Imaging modalities in general thoracic surgery

FORTHCOMING ISSUES

May 2004

Aggressive Surgery for Lung CancerValerie W. Rusch, MD, Guest Editor

August 2004

Postoperative Quality of LifeAnthony Yim, MD, Guest Editor

October 2004

MesotheliomaDavid J. Sugarbaker, MD, andMichael Chang, MD, Guest Editors

February 2005

Thoracic Anesthesia and Pain ManagementJerome M. Klafta, MD, Guest Editor

RECENT ISSUES

November 2003

Surgery for EmphysemaKeith S. Naunheim, MD, Guest Editor

August 2003

Lung TransplantationG. Alexander Patterson, MD, Guest Editor

May 2003

Tracheal SurgeryDouglas J. Mathisen, MD, Guest Editor

February 2003

Uncommon Tumors of the Tracheobronchial Tree:Diagnosis and ManagementMelvyn Goldberg, MD, andArthur S. Patchefsky, MD, Guest Editors

THE CLINICS ARE NOW AVAILABLE ONLINE!

Access your subscription at:http://www.TheClinics.com

CONTENTS

Foreword xiMark K. Ferguson and Catherine A. Bewick

Preface xiiiNasser K. Altorki and David F. Yankelevitz

Current State of Imaging for Lung Cancer Staging 1Michael S. Kent, Jeffrey L. Port, and Nasser K. Altorki

Lung cancer remains the leading cause of cancer death among men and women in theUnited States. The primary issue in the care of patients who have non–small-cell lungcancer is a determination of the stage of their disease. Several imaging techniques areavailable to help inform the determination of a patient’s stage, including CT, positronemission tomography, bone scintigraphy, and MRI. This article reviews these imagingtechniques and their indications for use based on current guidelines of clinical practice.

Imaging of Pleural and Chest Wall Tumors 15Michael J. Weyant and Raja M. Flores

Pleural and chest wall tumors encompass the relative minority of thoracic tumors.Advances in radiographic imaging modalities have allowed more accurate evaluation ofstaging and resectability of these tumors. CT and MRI appear to be relatively equal indetermining resectability and extent of invasion. Newer imaging modalities such as 18-flouro-deoxyglucose (FDG-PET) scanning appear to be most helpful in excluding thepresence of extrathoracic metastasis before surgical resection.

Imaging of the Mediastinum: Applications for Thoracic Surgery 25Dorith Shaham, Maria G. Skilakaki, and Orly Goitein

A wide variety of imaging modalities are available for evaluating the mediastinum, includ-ing plain radiography, CT, MRI, ultrasonography, and radionuclide imaging. CT is theimaging modality of choice for evaluating a suspected mediastinal mass or a widenedmediastinum; it provides the most useful information for diagnosis, planning of treatment,and evaluation of postoperative complications.

IMAGING MODALITIES IN GENERAL THORACIC SURGERY

VOLUME 14 • NUMBER 1 • FEBRUARY 2004 vii

State-of-the-Art Screening for Lung Cancer (Part 1): The Chest Radiograph 43Matthew Freedman

The chest radiographic methods used in prior studies of lung cancer screening and incurrent prospective clinical trials of lung cancer screening do not incorporate, as part oftheir prospective design, the newer methods available for the detection of lung nodules.Digital radiography, image processing, energy subtraction, and computer-aided detec-tion have been shown to enhance lung nodule detection. Temporal subtraction is a prom-ising method but with less supporting data currently available. These techniques, aloneor in combination, do not equal the nodule detection capability of lung CT, but they are likely to benefit patients having chest radiographs for other clinically indicated pur-poses and when the detection of a nodule is incidental to the clinical indication for theradiographic study.

State-of-the-Art Screening for Lung Cancer (Part 2): CT Scanning 53David Yankelevitz and Claudia I. Henschke

There have been dramatic improvements in technology in the past decade. In conjunctionthere have also been advances in our clinical knowledge that have led to changes in thescreening regimen. These changes are expected to continue in the future as CT scannerscontinue to improve and knowledge about screening accumulates, and computer-assistedtechniques are expected to play an ever more important role. This dynamic process willlead to continued improvements in the diagnostic distribution of lung cancers detectedunder CT screening.

Imaging for Esophageal Tumors 61Robert J. Korst and Nasser K. Altorki

The treatment of malignant tumors of the esophagus is stage-dependent, making accu-rate tumor staging of paramount importance. In this regard, imaging modalities play anintegral role in the staging of these lesions and are used to help determine the extent oflocoregional and distant disease. The accuracy of imaging for esophageal neoplasms isfar from perfect, however, with most suspicious lesions requiring biopsy for definitivestaging. The role of imaging techniques in the evaluation of esophageal tumors continuesto evolve and has recently begun to include assessment of the response to therapy.

Fluorescent Bronchoscopy 71Sebastien Gilbert, James D. Luketich, and Neil A. Christie

Detection of clinically occult lung neoplasms may represent an opportunity for early cur-ative intervention. Fluorescent bronchoscopy is a sensitive technique for detecting earlyendobronchial tumors that may be combined with CT scanning as part of a comprehen-sive lung cancer screening program. Identification and longitudinal follow-up of dys-plastic endobronchial changes with fluorescent bronchoscopy should facilitate studies ofchemoprevention and further knowledge regarding the natural history of these lesions.Analysis of bronchial epithelium with novel techniques such as genomic hybridizationand gene expression arrays might provide even better predictors of progression of dys-plastic endobronchial lesions.

Virtual Bronchoscopy for Evaluation of Airway Disease 79Steven E. Finkelstein, Ronald M. Summers, Dao M. Nguyen, and David S. Schrump

Virtual bronchoscopy (VB) is a novel modality for imaging airway anatomy that appearsto be highly useful for evaluation of airway anatomy due to endoluminal turmors or

viii CONTENTS

extrinsic compression. This modality is presently not reliable for evaluation of themucosal surface of the respiratory tract. Although form can be detected, mucosal color,irregularity, or friability cannot be assessed. As such, VB cannot be used for routine sur-veillance of patients at high risk of developing airway malignancies. The developmentof novel aerosolized contrast agents or spectroscopic techniques that can discriminatebenign versus malignant mucosal tissues might enhance the sensitivity and specificity ofVB for the detection of preinvasive cancers within the respiratory tract.

Chromoendoscopy and Magnification Endoscopy for Diagnosing 87Esophageal Cancer and DysplasiaMichael J. Connor and Prateek Sharma

Two primary subtypes of esophageal carcinoma are commonly seen in the esophagus:squamous cell carcinoma and adenocarcinoma. Currently, the diagnosis of metaplasticand dysplastic mucosa within the esophagus requires endoscopy with biopsy of abnormal-appearing tissue. Current practices of performing standard endoscopy with randombiopsies are inaccurate. Magnification and chromoendoscopy are among several toolsused in the esophagus to improve detection of squamous cell dysplasia/cancer, Barrett’sesophagus, and associated dysplasia. Current studies show that these techniques arepromising, although the results are still preliminary. These techniques will hopefullyimprove detection rates, decrease the number of biopsies required, and ultimately pro-vide a real-time diagnosis.

Radionuclide Imaging of Thoracic Malignancies 95Stanley J. Goldsmith, Lale A. Kostakoglu, Serge Somrov, and Christopher J. Palestro

Over the past decade a variety nuclear medicine imaging studies have become availablethat are of considerable value to patients who have pulmonary malignancies. By far thegreatest impact on the management of patients who have thoracic malignancy has beenthe availability of 18-flouro-deoxyglucose (18FDG-PET) imaging. In the patient who hasnewly diagnosed lung carcinoma, 18FDG-PET improves the accuracy of staging the disease by identifying or excluding mediastinal disease and distant metastatic foci.18FDG-PET is superior to anatomic methods for evaluating the response to therapy andfor distinguishing recurrent disease from posttreatment changes. Studies are in progressto evaluate the role of 18FDG-PET imaging in assessing prognosis.

Imaging of Acute Pulmonary Emboli 113Arfa Khan, Aaron Darius Cann, and Rakesh D. Shah

Pulmonary embolism (PE) is a significant cause of morbidity and mortality after surgi-cal procedures. Early diagnosis and prompt, effective management of this conditionpresent considerable clinical challenges to surgeons. Imaging studies form the mainstayof diagnosis of PE and include plain radiography, ventilation–perfusion scan, venogra-phy, echocardiography, catheter pulmonary angiogram, CT pulmonary angiogram, andMR pulmonary angiogram. Each imaging modality has a role in the diagnosis of PE.

Computer-Aided Diagnostics 125Anthony P. Reeves and Bryan M. Kressler

This article reviews the role of the computer in assisting physicians in interpreting CTimages of the lungs. Four primary computer functions are considered: visualization,detection, characterization and diagnosis, and whole-lung documentation and health

CONTENTS ix

evaluation. Computer-aided methods are emerging to aid the radiologist in the tasks ofdisease detection and diagnosis. Such methods might also be suitable to aid the surgeonin preoperative planning, the surgical operation, and postsurgical evaluation.

Future Generation CT Imaging 135Deborah Walter, Bruno De Man, Maria Iatrou, and Peter M. Edic

The article outlines some of the recent technological advances that will drive future CTevolution and describes the recently enabled applications and trends in thoracic imaging.Future technological developments in CT imaging will result in improvements in spatialresolution, coverage, temporal resolution, and dose reduction. The key to realizing thispotential is to combine improved imaging capability with advanced computer-assistedtools, which will expand the usefulness of CT imaging in many areas. This article dis-cusses examples of state-of-the-art and emerging clinical application using CT in theareas of lung cancer, chronic obstructive pulmonary disease, pulmonary embolism, andinterventional procedures.

Index 151

x CONTENTS

Thorac Surg Clin 14 (2004) xi

Foreword

As most readers know, the Clinics of North Amer- and the overall size of the issue has been increased

ica series has a long and illustrious history. The

Clinics were introduced in February 1912 with the

publication of the first issue of The Surgical Clinics of

North America, then known as the Surgical Clinics of

John B. Murphy, MD, at Mercy Hospital, Chicago.

The scope of the Clinics now includes 57 separate

series, with various subspecialty surgery titles. Many

of the Clinics have a Consulting Editor who is an in-

ternationally recognized expert in the subspecialty.

They are responsible for choosing topics and identi-

fying Guest Editors for each issue. The Guest Editors

identify appropriate content for each issue and select

authors to write the individual manuscripts. This for-

mat has now achieved a venerable place in the medical

and surgical literature, and the Clinics are regarded as

an authoritative source for clinical information written

and edited by leaders in the field.

The Chest Surgery Clinics of North America

series is undergoing some important changes. During

the past year, this series introduced a new Consulting

Editor. To the founding Consulting Editor for this se-

ries, Dr. L. Penfield Faber, we express gratitude on be-

half of thousands of readers who have benefited from

the extraordinary work he did for over 12 years. That

this series is among the most popular of its kind is

due entirely to his foresight and editorial skills. Main-

taining this legacy will be a tremendous challenge,

one that we relish.

Also, to better serve the readership, beginning

in 2004, the Chest Surgery Clinics of North America

has enhanced its cover art, enlarged its format, and

adopted a different title. The updated cover design

will now include artwork relevant to the issue’s topic,

1547-4127/04/$ – see front matter D 2004 Elsevier Inc. All right

doi:10.1016/S1547-4127(04)00044-1

to accommodate tables and illustrations more effec-

tively. Finally, the title has been changed to the

Thoracic Surgery Clinics to better reflect the contents

and the international character of the readership and

the authors.

Finally, we are pleased to announce that the

Clinics are now available online at www.TheClinics.

com. Subscribers to the printed version of the Clinics

will receive access to the online version at no ad-

ditional cost. We encourage all readers to visit

TheClinics.com to explore the features of the

Web site, including full text of all issues from 2002

to the present, comprehensive search capabilities, and

links to MEDLINE and other Elsevier journals.

It is our hope that these changes will enhance the

experience of the readership by providing the same

outstanding content they are accustomed to but in a

more accessible format. We welcome your comments

regarding these changes.

Mark K. Ferguson, MD

Department of Surgery

University of Chicago Medical Center

5841 S. Maryland Ave.

Chicago, IL 60637, USA

E-mail address: [email protected]

Catherine A. Bewick

Publisher, Elsevier

Elsevier Inc., The Curtis Center

Independence Square West, Suite 300

Philadelphia, PA 19106, USA


s reserved.

http:\\www.TheClinics.com

Thorac Surg Clin 14 (2004) xiii –xiv

Preface

Imaging modalities in general thoracic surgery

Nasser K. Altorki, MD David F. Yankelevitz, MD

Guest Editors

This issue of the Thoracic Surgery Clinics is onstrate where we can expect the intersection of these

dedicated to thoracic imaging with emphasis on the

recent advances and their relevance to the general

thoracic surgeon. Foremost among these advances are

the dramatic improvements in cross-sectional imag-

ing, particularly as they apply to CT scanning. Multi-

slice scanners have now largely replaced the single

slice scanners of the early 1990s, allowing for faster

image acquisition and higher spatial resolution. Im-

age acquisition using 10-mm slice thickness has now

given way to 0.675-mm slice thickness, a 15-fold

increase in spatial resolution. These changes have al-

lowed the detection of even smaller nodules and lung

cancers. Alongside improved spatial resolution, com-

puter-aided approaches are slowly entering the realm

of clinical practice. Faster computers with larger

image storage capacity have allowed manipulation

of high-resolution data and three-dimensional display.

Computer-aided diagnostic techniques enhance nod-

ule detection, analysis, and growth rates as well as

detection of pulmonary emboli.

In this issue, we admittedly pay special attention

to CT and its pivotal role in thoracic imaging. This

relates to issues such as lung cancer screening and

staging and the diagnosis of pulmonary emboli. We

also have included articles on the future of CT

scanners and image processing techniques that dem-


doi:10.1016/S1547-4127(04)00043-X

technologies to lead.

Although CT imaging plays an important role, it is

by no means an exclusive one, and many other

technologies have advanced and become quite useful.

MRI, which is most useful in tissue characterization,

is discussed with emphasis on its role in the staging of

tumors both in the lung and the mediastinum. Cur-

rently, MRI plays mainly a complementary role to CT;

however, rapid advances in this technology are occur-

ring as well, and its role in imaging of pulmonary

emboli may soon surpass CT.

In selecting contributions to this issue, we chose to

include articles that give state-of-the-art reviews on

the clinical role of various imaging procedures in

common thoracic surgical problems. We have also

included articles on specific imaging techniques that

have emerged and are now becoming readily availa-

ble; this includes various endoscopic techniques such

as virtual bronchoscopy and laser-induced fluores-

cence endoscopy (LIFE) bronchoscopy for the air-

ways, as well as chromo and magnification endoscopy

for the esophagus. These powerful new imaging

techniques have already found their way into clinical

practice, and these articles discuss their respective

advantages and limitations. We also included an

article on radiolabeled imaging for thoracic tumors.

s reserved.

N.K. Altorki, D.F. Yankelevitz / Thorac Surg Clin 14 (2004) xiii–xivxiv

Although positron emission tomography using fluo-

rodeoxyglucose (FDG) is the primary agent currently

used, this entire field is changing rapidly. New tar-

geted diagnostic and therapeutic agents are being

developed. This field will be best positioned to

leverage the advances in molecular biology and in-

corporate this into sophisticated imaging strategies.

The changes that have occurred in the past 10 years

have indeed been dramatic. Diseases are being diag-

nosed earlier, allowing for development of new thera-

peutic strategies. In addition, the thoracic surgeon can

now plan for more complex procedures. Combining

the advances in imaging technology with advances in

therapeutics will undoubtedly have a major impact on

the practice of thoracic surgery.

Nasser K. Altorki, MD

Department of Cardiothoracic Surgery

Weill Medical College

Cornell University

525 East 68th Street, Suite M404

New York, NY 10021, USA


David F. Yankelevitz, MD

Department of Radiology, Chest Division

Weill Medical College

Cornell University

525 East 68th Street



Thorac Surg Clin 14 (2004) 1–13

Current state of imaging for lung cancer staging

Michael S. Kent, MDa, Jeffrey L. Port, MDb, Nasser K. Altorki, MDb,*

aDepartment of Surgery, Weill Medical College, Cornell University, 525 East 68th Street, Suite K707, New York, NY 10021, USAbDepartment of Cardiothoracic Surgery, Weill Medical College, Cornell University, 525 East 68th Street, Suite M404,


Lung cancer remains the leading cause of cancer phy (BS), and MRI. Each of these studies carries a

death among men and women in the United States.

In 2002 169,400 patients were diagnosed with lung

cancer and 155,000 deaths resulted from the disease

[1]. In part, this poor survival reflects the fact that

the majority of patients who have lung cancer pre-

sent with locally advanced or metastatic disease.

Forty-nine percent of patients who were diagnosed

lung cancer in 2002 were found to have distant me-

tastases at the time of presentation, and 26% of pa-

tients had mediastinal lymph node involvement [1].

Therefore, less than 25% of patients are candidates

for surgery as the sole method of treatment.

From the perspective of the thoracic surgeon, the

primary issue in the care of patients who have non–

small-cell lung cancer is a determination of the stage

of their disease. Stage determines the treatment pa-

tients will receive and their prognosis. Inaccurate

staging might deny patients access to potentially

curative treatment and expose them to unnecessary

therapy. In effect, accurate staging is as critical to

the care of patients who have lung cancer as their

ultimate treatment.

The critical issue in staging is to identify patients

who have extrathoracic disease, who are not candi-

dates for surgery, and to identify patients who have

N2 disease, whose survival might be improved by

induction chemotherapy followed by surgery. Sev-

eral imaging techniques are available to help inform

the determination of a patient’s stage, including CT,

positron emission tomography (PET), bone scintigra-


doi:10.1016/S1547-4127(04)00031-3

* Corresponding author.


(N.K. Altorki).

financial cost and measurable false-positive and false-

negative rates. The injudicious use of imaging leads

to excessive costs and unnecessary invasive proce-

dures. Worse, a false-positive study might deny a pa-

tient potentially curative surgery. This article reviews

these imaging techniques and their indications for

use based on current guidelines of clinical practice.

Staging the primary tumor

When a pulmonary nodule is found to be malig-

nant, the initial step in defining the clinical stage

of the tumor is to determine the tumor (T) stage.

Outside the context of clinical trials, the distinction

between T1 and T2 disease does not usually impact

on the recommendation for treatment; however, the

distinction between invasion of the chest wall or other

resectable structures (T3) versus mediastinal struc-

tures such as the trachea or heart (T4) has significant

surgical implications.

CT

Tumors that invade the chest wall are considered

to be T3 disease. The finding of chest wall invasion

at the time of surgery does not preclude curative

resection; however, the preoperative diagnosis of

chest wall invasion does allow the surgeon and pa-

tient to anticipate en-bloc resection of the chest wall

with the primary tumor and the need for subsequent

reconstruction. Several findings on CT such as ex-

tensive contact with the parietal pleura, extrapleural

soft tissue, and obliteration of the extrapleural fat

plane suggest chest wall invasion but are relatively

s reserved.

M.S. Kent et al / Thorac Surg Clin 14 (2004) 1–132

nonspecific [2,3] (Fig. 1). The only findings on CT

that have been found to be highly predictive of chest

wall invasion are destruction of adjacent ribs and

clear extension of tumor beyond the chest wall [4],

and even these signs have a sensitivity of only 20%

[5]. The most accurate predictor of T3 disease is dy-

namic CT, which can document fixation of the tumor

to the chest wall through the respiratory cycle [6]. This

specialized study is not widely available, however.

The distinction between resectable tumors, which

invade the mediastinal pleura (T3), and unresectable

tumors, which invade structures such as the heart

or trachea (T4), is difficult to make on the basis of

CT imaging alone. Frequently, tumors abut the me-

diastinum and obliterate the normal fat plane on CT

but are deemed to be resectable at the time of tho-

racotomy (Fig. 2). For example, in a retrospective

study of 180 patients who had lung cancer staged by

conventional CT, only 62% of patients staged T4 by

Fig. 1. False-positive CT scan suggesting chest wall inv

CT were found to have T4 disease at the time of sur-

gery [7]. Findings on CT that increase the likelihood

of unresectability include involvement of the carina

or encasement of more than half the circumference

of the aorta, esophagus, or proximal left and right

pulmonary arteries [8]; however, even when these

signs are strictly applied, the predictive value of CT

in determining T4 disease is quite low [9,10]. Tumors

that have equivocal signs of invasion—even with

obliteration of the normal mediastinal fat planes—

should not be considered to be unresectable on the

basis of CT imaging alone [11].

MRI

MRI has found limited applicability in the imag-

ing of lung cancer, although it might be more useful

than CT scanning in specific circumstances. In 1991

the Radiologic Diagnostic Oncology Group (RDOG)

asion. (A) Lung window, (B) mediastinal window.

Fig. 2. False-positive CT scan of mediastinal invasion. The tumor (arrow) was completely resectable at the time of thoracotomy

and the mediastinal pleura was not invaded.

Fig. 3. T1-weighted MRI showing vertebral invasion

(arrow) by a Pancoast tumor.

M.S. Kent et al / Thorac Surg Clin 14 (2004) 1–13 3

directly compared the accuracy of MRI and CT in

170 patients who had operable non-small cell lung

cancer. The sensitivity and specificity of CT in

distinguishing T0–2 from T3–4 tumors were 63%

and 84%, respectively. No significant difference was

noted between CT and MRI, which had a sensitivity

and specificity of 56% and 80%, respectively [12].

Although no differences were noted in the determi-

nation of chest wall or airway invasion, MRI was

significantly more accurate in determining invasion

of the mediastinum.

Since the RDOG report, MRI technology has im-

proved, and its utility in evaluating patients who have

lung cancer has expanded. For example, the devel-

opment of MR angiography has allowed for much

improved resolution of hilar and mediastinal vessels.

In a pilot study of 50 patients imaged with MR an-

giography, the overall accuracy in predicting hilar or

mediastinal invasion was 88%, which was superior to

contrast-enhanced CT or conventional T1-weighted

MRI [13]. However, because of the low imaging

signal of air, MRI is inferior to conventional CT in

documenting endobronchial invasion [14].

One area in which MRI is clearly superior to CT

is in the evaluation of tumors of the superior sulcus.

The structures adjacent to the apex of the lung

(eg, the brachial plexus and subclavian vessels) are

not well visualized in the axial plane. MRI, unlike

CT, can image these structures in the coronal and

sagittal plane, and consequently is the imaging study

of choice for Pancoast tumors [15]. MRI can also

determine invasion of the vertebral body and exten-

sion of disease into the neural foramina, which is

critical information for preoperative planning [16]

(Fig. 3). Overall, MRI has been found to have a

94% correlation with surgical findings for Pancoast

tumors, compared with 63% accuracy for CT [17].

c Surg Clin 14 (2004) 1–13

Thoracoscopy

Although a detailed discussion is outside the

scope of this article, it should be mentioned that

minimally invasive techniques can be used to de-

termine resectability when imaging is equivocal.

Thoracoscopy allows for the cytologic evaluation

of pleural effusions and can determine invasion of

the chest wall and mediastinal structures by direct

visualization [18,19]. Thoracoscopy can also be used

to directly explore the pericardial cavity. In a small

study of 27 patients who had clinical T4 tumors, the

pericardial sac was explored using the same equip-

ment and port sites as for standard thoracoscopy.

This technique identified, with no complications,

six patients who were unresectable on the basis of

invasion of the heart or main pulmonary artery [20].

M.S. Kent et al / Thora4

Staging the mediastinum

The involvement of mediastinal lymph nodes

has a significant impact on the treatment and prog-

nosis of patients who have lung cancer. Mediastino-

scopy remains the gold standard to detect N2 nodal

metastases before thoracotomy. The procedure can be

performed with a complication rate well below 1%

and has a negative predictive value (NPV) of 93%

[21]. Although noninvasive modalities such as PET

have emerged to stage the mediastinum, none of

these techniques has a specificity high enough to ex-

clude patients from resection without confirmation

by tissue biopsy.

CT

The detection of nodal metastases on CT is based

on nodal size. By convention, a mediastinal node

larger than 1 cm in the short axis is considered to

be enlarged [22]; however, this convention suffers

from many limitations. First, the normal size of me-

diastinal lymph nodes varies by nodal station. Hilar

nodes can measure up to 7 mm, and benign sub-

carinal nodes can be as large as 15 mm [23]. In

addition, surrounding mediastinal structures and

volume averaging effects might make precise deter-

mination of nodal size difficult. Consequently, inter-

observer variability in the measurement of nodal size

is relatively high. Most importantly, normal-sized

nodes might harbor micrometastatic disease and en-

larged nodes might be reactive because of infection or

inflammatory processes rather than malignancy. The

accuracy of CT scanning, therefore, is relatively low.

In a meta-analysis of more than 20 studies with

3438 evaluable patients, the pooled sensitivity and

specificity of CT was 57% and 82%, respectively

[24]. There was marked heterogeneity between

studies, however, which was in part attributable to

variability between study populations. For instance,

the incidence of micrometastases to mediastinal

lymph nodes is higher in adenocarcinomas compared

with squamous cell cancers. As a consequence, the

false-negative rate of CT scans is significantly higher

in this group of patients [25]. Furthermore, the

specificity of CT varies with the location where the

study is performed. For example, the false-positive

rate will be higher in areas where sarcoidosis or other

granulomatous diseases are endemic [26].

MRI

MR signal characteristics and relaxation times

are unable to discriminate benign from malignant

nodes; therefore, the only criterion used to determine

nodal involvement in standard MR imaging is that

of size [27]. Consequently, the overall accuracy of

MRI in detecting nodal metastases is no better than

that of CT [9,12]. Other limitations in the imaging of

thoracic lymph nodes are unique to MRI. For exam-

ple, MRI is unable to visualize calcification within a

lymph node, a finding that would suggest a benign

etiology for nodal enlargement on CT. Because of the

poor spatial resolution of MRI, a group of normal-

sized nodes might be interpreted as a single node,

which would falsely raise the suspicion of metastatic

disease [28].

Refinements in MRI might make this modality

more useful for determining nodal stage in the future.

It has been shown in a small pilot study that the

pattern of enhancement of malignant nodes with

gadolinium is significantly different than for benign

nodes [29]. Although larger, confirmatory studies are

needed, this technique might prove to be a relatively

simple way to discriminate patients who have nodal

disease. Another emerging technology is that of MR

lymphography, in which superparamagnetic iron ox-

ide particles are used as the contrast agent. Iron oxide

particles are readily phagocytosed by macrophages in

normal nodal tissue and lower the signal intensity of

the node on T2-weighted sequences. Nodes that

harbor metastatic disease do not accumulate the

contrast agent as readily and therefore have greater

signal intensity on T2 images [30]. Early studies of

MR lymphography have demonstrated high sensi-

tivity and specificity in patients who have urologic

malignancies [31]; however, only small studies on

patients who have bronchogenic carcinoma have

been reported so far [32].

c Surg Clin 14 (2004) 1–13 5

Positron emission tomography

Without question, PET scanning using fluordeoxy

glucose (FDG) has shown the greatest promise in

staging the mediastinum noninvasively (Fig. 4). In

some centers PET scanning has become an almost

routine component of the preoperative evaluation of

patients who have lung cancer. This practice is

justified by several meta-analyses that have demon-

strated the superiority of PET over CT in staging the

mediastinum [20,33,34]. In a representative meta-

analysis [20] that included 1045 patients enrolled in

18 studies, the pooled sensitivity and specificity of

PET scanning were 84% and 89%, respectively. A

M.S. Kent et al / Thora

Fig. 4. Mediastinal spread of a right lower lobe lung cancer. (A) Su

tumor is also visible (arrowhead). (B) An axial FDG-PET scan dem

the subcarinal space.

direct comparison of PET and CT by receiver oper-

ating characteristic analysis demonstrated PET scan-

ning to be significantly more accurate. Perhaps the

most relevant measure of a staging study is the nega-

tive predictive value (NPV) of the test, which defines

the likelihood that a patient who has a negative test

result does not have the disease. The NPV of PET

scanning to stage the mediastinum in this study was

93%, compared with only 83% for CT scanning.

Several studies have documented the high impact

and cost-effectiveness of PET scanning on clinical

decision-making [35,36]. In addition to these retro-

spective series, the utility of PET scanning has been

evaluated in a prospective, randomized trial. The

bcarinal lymphadenopathy (arrow) on chest CT. The primary

onstrating increased glucose uptake in the primary tumor and


results of this trial, known as the PET in Lung Cancer

Staging Study (PLUS) were reported in 2002 [37].

In this trial 188 patients who had suspected or proven

non-small cell lung cancer were assigned to a con-

ventional workup (as determined by local practice)

or a conventional workup plus a PET scan. The end-

point of the study was a reduction in the number

of futile thoracotomies, which was defined as thora-

cotomy for benign disease, thoracotomy without

resection, unsuspected N2 or T4 disease, or relapse

within 12 months of surgery. In the conventional

workup group 41% of patients had a futile thora-

cotomy compared with 21% in the PET group, which

represents a relative reduction of 51%, which is

highly significant. One criticism of this study is that

the extent of the conventional workup was not

specified in the protocol. For example, it is not clear

whether or not the percentage of patients in whom the

suspicion of lung cancer was confirmed by a needle

biopsy was similar in both groups. Such a difference

might explain the observation that the number of

thoracotomies for benign disease was three times

higher in the conventional group than the PET scan

group. In centers in which needle biopsy is practiced

routinely, the impact of PET scans would be less than

that reported by the PLUS trialists.

There are other limitations of PET scanning. The

test carries considerable cost and limited availability.

In the United States the cost of a PET scan is ap-

proximately $2000. Furthermore, given a half-life of

110 minutes, the radioisotope must be produced by

an onsite cyclotron or be manufactured within 200 km

Fig. 5. A CT/PET fusion of a le

of the imaging center. Clinicians must also be cau-

tioned that not all PET scan centers use the same

technology. The published literature demonstrating

the superiority of PET to stage the mediastinum is

based on the use of dedicated PET scanners. Com-

peting systems using gamma cameras have been

introduced in an effort to lower the cost of the study.

It is estimated that there are nearly twice as many

camera-based scanners than dedicated PET scanners

currently in use [38]; however, imaging based on

gamma cameras is clearly less sensitive than that of a

dedicated PET system, and the overall accuracy might

not be much higher than standard CT alone [39].

Even with the use of dedicated systems, the

accuracy of PET scans should not be assumed in all

clinical situations. The spatial resolution of PET scans

is clearly inferior to that of CT, and PET is particu-

larly poor at documenting N1 disease [31]. In addi-

tion, the utility of PET in restaging patients after

induction chemotherapy has not been well estab-

lished. To date, two studies reporting on a total of

90 patients have been published with contradictory

findings [40,41]. In the authors’ experience PET did

not predict nodal status accurately in more than half

of patients restaged after induction chemotherapy,

with an equal proportion over- and understaged [42].

CT/positron emission tomography fusion

Interpretation of a PET scan in the presence of

CT images clearly improves the sensitivity and spec-

ft lower lobe lung cancer.


ificity of the study [43]. The development of hy-

brid PET/CT scanners is a natural outgrowth of this

observation (Fig. 5). The first prototype, which used

a single-detector CT scanner combined with a par-

tial-ring rotating PET scanner, was introduced re-

cently [44]. The benefits of this new technology

have not yet been clarified. Experience with a more

advanced scanner using multidetector CT combined

with a full-ring detector PET scanner was reported

in 2002. In this study of 53 patients who had a variety

of malignancies including lung cancer, PET/CT fu-

sion was felt to significantly improve diagnostic ac-

curacy over PET alone [45]. Another variation of

this technology is the combination of CT with a

camera-based PET scanner. A small study of 21 pa-

tients who had thoracic malignancies showed that

the accuracy of this system was equal to that of a

dedicated PET scanner [46]. If replicated in larger

studies, this finding might obviate the need for dedi-

cated PET scanners, which are more expensive and

limited in availability.

Endoscopic ultrasound

While mediastinoscopy is a proven tool for stag-

ing patients who have non-small cell lung cancer,

the technique has recognized limitations. Although

mediastinoscopy is an outpatient procedure, the pro-

cedure requires general anesthesia, is difficult to per-

form more than once, and has a small but defined

complication rate. Certain nodal stations such as

levels VIII and IX are also difficult to access by

standard mediastinoscopy. Endoscopic ultrasound

(EUS) has been proposed as an alternative to media-

stinoscopy in specific circumstances. The technique is

no different than EUS used for staging esophageal

cancer and involves the use of an ultrasound probe

placed at the tip of a modified endoscope. EUS

provides excellent visualization of the subcarinal

space and nodes in the inferior mediastinum. Suspi-

cious nodes are identified on the basis of size and

by disruption of the normal architecture, and they can

be sampled by fine-needle aspiration (FNA). In a

pooled analysis of five studies, the reported sensitivity

for this technique was 78% and the specificity was

71% [20]; however, a recent study in which all nodes

were sampled regardless of appearance showed that

the stage of 42% of patients was changed by EUS/

FNA [47]. A significant drawback of this technique is

its inability to visualize right-sided paratracheal

nodes. Given this limitation, it is likely that EUS will

at best complement, rather than replace, staging by

CT, PET, or mediastinoscopy.

The search for extrathoracic disease

The central questions in the search for extrathora-

cic disease are when such an investigation is worth-

while and to what extent it should be pursued.

Patients who have clinical signs or symptoms of

distant disease should undergo a full metastatic

workup; however, in the absence of clinical findings

the yield of such a workup is quite low. For example,

the incidence of silent metastases in patients who

have clinical stage I lung cancer is as low as 1% [48].

A uniform policy of imaging for extrathoracic dis-

ease in this group of patients would therefore incur

considerable expense, unnecessary invasive proce-

dures, and perhaps a significant delay in definitive

treatment [49].

The ability of a thorough clinical evaluation to

exclude metastatic disease has been well studied. Se-

venteen studies have been published in which clini-

cal evaluation was compared with the gold standard

of CT imaging of the brain. The pooled NPV among

1784 patients studied was 94% [20]. In the same

meta-analysis of studies evaluating the presence of

abdominal or bony metastases by the clinical exam-

ination (including routine serum chemistry), the NPV

was 95% and 90%, respectively [20].

If the search for silent metastases is restricted to

patients who have more advanced-stage disease, the

yield will be substantially higher. Approximately

25% of patients who have clinical N2 disease will

harbor metastatic disease [50], and patients who have

tumors greater than 3 cm are more likely to have brain

metastases when screened by MRI. Tumor histology

alone is not an independent risk factor for metastatic

disease [42]. Consequently, there is no indication that

patients who have adenocarcinoma require a more

thorough evaluation than patients who have squamous

cell cancer in the absence of clinical findings.

The single randomized study to address the issue

of screening for metastases in patients who have

non–small-cell lung cancer was reported by the

Canadian Lung Oncology Group in 2001 [51]. In

this study all patients were evaluated with a CT of

the chest and mediastinoscopy. Patients were then

randomized to immediate thoracotomy or additional

evaluation by bone scintigraphy and dedicated CT

scans of the abdomen and brain. The hypothesis

of the study was that additional evaluation would

lead to a lower rate of thoracotomies without cure,

defined as an incomplete resection or thoracotomy

with subsequent recurrence. Among the 634 patients

who were randomized, thoracotomy without cure

occurred in 73 patients in the limited investigation

group and in 58 patients in the full investigation


group. This trend was not statistically significant (P =

0.20) and no difference in survival was observed

between the two groups. An economic analysis cal-

culated less cost in the full investigation group

because of the avoidance of additional surgical

procedures; however, it is not clear whether or

not this would hold true in the United States’ health

care system.

Should a metastatic workup be deemed neces-

sary, some organ-specific considerations are dis-

cussed herein, followed by the authors’ current

imaging recommendations.

Brain

Central nervous system (CNS) metastases occur

in less than 3% of all asymptomatic lung cancer

patients [52]. Furthermore, in one study routine

CNS scanning led to a false-positive rate of 11%

[53]. While asymptomatic patients need not be

screened for brain metastases, the definition of what

constitutes symptoms differs widely among physi-

cians. Often, patients who have mild symptoms such

as headache of dizziness are classified as asymptom-

atic, although these patients are clearly documented

to have a higher rate of brain metastases [54].

CT and MRI are both suitable imaging studies for

evaluating for brain metastases. Gadolinium-en-

hanced MRI can detect smaller lesions and has a

higher sensitivity than a CT with contrast. Although

MRI can detect more lesions in a single patient, it has

not been shown to upstage a greater number of

Fig. 6. CT scan with contrast demonstrating

patients compared with CT [55]. Consequently, the

detection of smaller metastases by MRI is rarely of

clinical significance. Prolonged survival in patients

whose lesions were detected by MRI over CT is

likely caused by lead-time bias rather than a true

survival benefit [56].

Adrenal

Adrenal lesions are common in the general popu-

lation and most often represent adrenal adenomas

[57]. The assumption that an adrenal mass in a can-

cer patient represents a metastasis is not always valid.

Although an adrenal mass is more likely to be malig-

nant in patients who have advanced-stage disease

[58], adenomas predominate in patients who have

clinical stage IA cancer [59]. It is therefore critical

that these lesions be characterized precisely. A patient

can be denied potentially curative surgery if an

adenoma is mistakenly presumed to represent meta-

static disease. On the other hand, select patients might

be candidates for synchronous adrenalectomy and

pulmonary resection if a definitive diagnosis is made.

Typically, an adrenal mass is diagnosed on the

lower cuts of a contrast-enhanced chest CT per-

formed to evaluate the primary tumor (Fig. 6). Char-

acteristics of an adenoma include a low attenuation

lesion of less than 5 cm with a smooth, high attenua-

tion rim. A definitive diagnosis based on these crite-

ria is not always possible, however, and further

assessment becomes necessary [60]. One option is

to acquire delayed images to observe the pattern of

bilateral adrenal metastases (arrows).

Fig. 7. FDG-PET demonstrating multiple sites of meta-

static disease.


contrast washout. Adenomas typically display mod-

erate contrast enhancement with substantial washout

after 15 minutes. Adrenal metastases show the oppo-

site pattern: intense enhancement and little washout.

This technique has a reported sensitivity and speci-

ficity of 96% [61].

Another option is to repeat the CT without con-

trast. Adenomas are characterized by their high fat

content and consequently have a low attenuation

value on nonenhanced CT. The specificity of the

method will vary with the threshold used to define

malignancy. In a meta-analysis of 10 studies, the

specificity varied from 100% at a cutoff of 2 Houns-

field units (HU) to 87% at 20 HU. This study re-

commended that a threshold of 10 HU be used [62].

MRI has also been used to differentiate adenomas

from malignant disease on the basis of fat content.

Initial experience with MRI has suggested that

adenomas can be identified by their low signal on

T2-weighted images [63]. Further evaluation has

shown that this finding is relatively nonspecific, and

newer techniques using MR spectroscopy have sup-

planted routine MR imaging. Using chemical shift

imaging and dynamic gadolinium enhancement, MRI

was shown to have a specificity of 100% and spec-

ificity of 81% [64]. Unfortunately, this specialized

examination is not widely available.

Finally, PET scanning can also be used to char-

acterize adrenal masses. In three studies evaluating

88 patients who had a variety of malignancies, PET

scanning was shown to have a sensitivity of 100%

and a specificity between 80% and 100% [65–67].

Thus, an adrenal mass seen on CT that is negative on

PET is unlikely to be malignant. However, because

of a small but defined false-positive rate, patients

should undergo a confirmatory percutaneous needle

biopsy if the PET scan suggests an adrenal metastasis.

Bone

Routine BS in asymptomatic patients leads to

positive results in up to 40% of cases [68], however

bone scans are relatively nonspecific and have a

false-positive rate as high as 40% because of the

prevalence of preexisting traumatic or degenerative

skeletal disease [69]. MRI is also plagued by a high

number of false-positive scans, and it does not seem

that the overall accuracy of MRI surpasses that of

standard BS [70]. Although there are fewer studies of

PET scanning in this setting, they suggest that its

sensitivity and specificity are at least equal to, if not

superior to, bone scans [71,72]. In one study PETwas

shown to have an equivalent sensitivity but a superior

specificity (98% versus 61%) to bone scans, but

direct comparison between these techniques is diffi-

cult because of a flawed study design. In the majority

of reports a suspicious lesion was not definitively

diagnosed by a fine needle biopsy, so the true false-

positive rate could not be established.

Extrathoracic staging with positron emission

tomography

The hope that whole-body PET might replace the

standard metastatic workup for patients who have

lung cancer deserves special mention (Fig. 7). The

accuracy of PET in imaging metastases to the bone or

solid organs excluding the brain equals or surpasses

that of standard imaging. PET has been shown to de-

tect extrathoracic metastases in 11% to 14% of patients

who were thought to have localized disease by con-

ventional imaging [61,73]. Furthermore, negative PET

scans can exclude metastatic disease suggested by CT

scans with a reported 1% false-negative rate [61,63].

PET has some limitations in whole-body staging,

however. PET cannot replace standard imaging of

the brain. Because of the high metabolic rate of nor-

Fig. 8. False-positive FDG-PET of an early-stage lung

cancer. Supraclavicular lymph node biopsy revealed scle-

rosing lymphadenitis.


mal brain tissue, PET is extremely poor at detecting

cerebral metastases, with a sensitivity of only 60%

[61]. There is also a concern regarding the wide-

spread application of whole-body imaging in an

asymptomatic population. As more patients who

have early-stage lung cancer are staged by PET, the

issue of false-positive studies becomes more relevant

(Fig. 8). If every asymptomatic patient was screened

with PET, which has a specificity between 80% and

100% for bone and adrenal lesions, a significant

number of invasive and perhaps unnecessary diag-

nostic procedures would result.

Summary

Proper selection and interpretation of imaging

studies is essential to provide optimal treatment to

patients who have lung cancer. The following com-

bines the recommendations of the American College

of Chest Physicians [74] and the authors’ current

clinical practice guidelines:

� All patients who have known or suspected lung

cancer should undergo a CT of the chest and

upper abdomen.� An FDG-PET study should be performed, if

available.� Mediastinoscopy should be performed in all

patients except those who have peripheral small

(<2 cm) tumors and no evidence of N2 disease

on CT or PET imaging.� MRI should be performed for tumors of the

superior sulcus to define the relationship of the

tumor to adjacent neurovascular structures.� Patients who have neurologic signs or symp-

toms should undergo a brain imaging study (CT

or MRI).� Screening for extrathoracic disease is not nec-

essary in asymptomatic patients who have clini-

cal stage I or II disease.

References

[1] Jemal A, Thomas A, Murray T, et al. Cancer statistics,

2002. CA Cancer J Clin 2002;52:23–47.

[2] Ratto GB, Piacenza G, Frola C, et al. Chest wall

involvement by lung cancer: computed tomographic

detection and results of operation. Ann Thorac Surg

1991;51:182–8.

[3] Pennes DR, Glazer GM, Wimbish KJ, et al. Chest wall

invasion by lung cancer: limitations of CT evaluation.

Am J Roentgenol 1985;144:507–11.

[4] Quint LE, Francis IR, Wahl RL, Gross BH, et al. Pre-

operative staging of non-small cell carcinoma of the

lung: imaging methods. Am J Roentgenol 1995;164:

1349–59.

[5] Pearlberg JL, Sandler MA, Beute GH, et al. Limita-

tions of CT in evaluation of neoplasms involving the

chest wall. J Comput Assist Tomogr 1987;11:290–3.

[6] Murata K, Takahashi M, Mori M, et al. Chest wall and

mediastinal invasion by lung cancer: evaluation with

multisection expiratory dynamic CT. Radiology 1994;

191:251–5.

[7] Cetinkaya E, Turna A, Yildiz P, et al. Comparison of

clinical and surgical–pathological staging of patients

with non-small cell lung carcinoma. Eur J Cardio-

thorac Surg 2002;22:1000–5.

[8] Gat SB, Black WB, Armstrong P, et al. Chest CT

of unresectable lung cancer. Radiographics 1988;8:

735–48.

[9] Martini N, Heelan R, Westcott J, et al. Comparative

merits of conventional computed tomographic and

magnetic resonance imaging in assessing medias-

tinal involvement in surgically confirmed lung carci-

noma. J Thorac Cardiovasc Surg 1985;90:639–48.


[10] Glazer HS, Kaiser LR, Anderson DJ, et al. Indetermi-

nate mediastinal invasion in bronchogenic carcinoma:

CT evaluation. Radiology 1989;173:37–42.

[11] White PG, Adams H, Crane MD, et al. Preoperative

staging of carcinoma of the bronchus: can computed

tomographic staging readily identify stage III tumors?

Thorax 1994;49:951–7.

[12] Webb WR, Gatsonis C, Zerhouni EA, et al. CT and

MR imaging in staging non-small cell broncho-

genic carcinoma: report of the Radiologic Diagnostic

Oncology Group. Radiology 1991;178(3):705–13.

[13] Ohno Y, Adachi S, Motoyama A, et al. Multiphase

ECG-triggered 3D contrast-enhanced MR angiogra-

phy; utility for evaluation of hilar and mediastinal

invasion of bronchogenic carcinoma. J Magn Reson

Imaging 2001;13:215–24.

[14] Mayr B, Heywang S, Ingrisch H, et al. Comparison

of CT with MR imaging of endobronchial tumors.

J Comput Assist Tomogr 1987;11:43–8.

[15] McCloud T, Filion R, Edelman R, et al. MR imaging

of superior sulcus carcinoma. J Comput Assist Tomogr

1989;13:233–9.

[16] Freundlich I, Chasen M, Varma D. Magnetic reso-

nance imaging of pulmonary apical tumors. J Thorac

Imaging 1996;11:210–22.

[17] Heelan R, Demas B, Caravelli J, et al. Superior sul-

cus tumors: CT and MR imaging. Radiology 1989;170:

637–41.

[18] Eggeling S, Martin T, Bottger J, et al. Invasive staging

of non-small cell lung cancer—a prospective study.

Eur J Cardiothorac Surg 2002;22:679–84.

[19] Waller D, Clarke S, Tsang G. Is there a role for

video-assisted thoracoscopy in the staging of non-

small cell lung cancer? Eur J Cardiothorac Surg

1997;12:214–7.

[20] Loscertales J, Jimenez-Merchan R, Congregado-Lo-

scertales M, et al. Usefulness of videothoracoscopic

intrapericardial examination of pulmonary vessels

to identify resectable clinical T4 lung cancer. Ann

Thorac Surg 2002;73:1563–6.

[21] Patterson G, Ginsberg R, Poon P, et al. A prospective

evaluation of magnetic resonance imaging, computed

tomography, and mediastinoscopy in the preopera-

tive assessment of mediastinal node status in broncho-

genic carcinoma. J Thorac Cardiovasc Surg 1987;94:

679–84.

[22] Grenier P, Dubray B, Carette M, et al. Pre-operative

staging of lung cancer: CT and MR evaluation. Diagn

Interv Radiol 1989;1:23–30.

[23] Genereux G, Howie J. Normal mediastinal lymph

node size and number: CT and anatomic study. Am J

Roentgenol 1984;142:1095–100.

[24] Toloza E, Harpole L, McCroy D. Noninvasive stag-

ing of non-small cell lung cancer: a review of the cur-

rent evidence. Chest 2003;1(S):137–45.

[25] Vallieres E, Waters P. Incidence of mediastinal node

involvement in clinical T1 bronchogenic carcinomas.

Can J Surg 1987;30:341–2.

[26] Brudin L, Valind S, Rhodes C, et al. Fluorine-18 de-

oxyglucose uptake in sarcoidosis measured with posi-

tron emission tomography. Eur J Nucl Med 1994;21:

297–305.

[27] Glazer G, Orringer M, Chenevert T, et al. Mediastinal

lymph nodes: relaxation time/pathologic correlation

with implications in staging of lung cancer with MR

imaging. Radiology 1988;168:429–31.

[28] Boiselle P. MR imaging of thoracic lymph nodes.

MRI Clin N Am 2000;8:33–41.

[29] Laissey J, Gay-Depassier P, Soyer P, et al. Enlarged

mediastinal lymph nodes in bronchogenic carcinoma:

assessment with dynamic contrast-enhanced MR

imaging. Radiology 1994;191:263–7.

[30] Vassallo P, Matei C, Heston W, et al. AMI-227-

enhanced MR lymphography: usefulness for differ-

entiating reactive from tumor-bearing lymph nodes.

Radiology 1994;193:501–6.

[31] Bellin M, Roy C, Kinkel K, et al. Lymph node meta-

stases: safety and effectiveness of MR imaging

with ultrasmall superparamagnetic iron oxide par-

ticles—initial clinical experience. Radiology 1998;

207:799–808.

[32] Nguyen B, Stanford W, Thompson B, et al. Multicenter

clinical trial of ultrasmall superparamagnetic iron

oxide in the evaluation of mediastinal lymph nodes

in patients with primary lung carcinoma. J Magn

Reson Imaging 1999;10:468–73.

[33] Dwamena B, Sonnad S, Angobaldo J, et al. Metastases

from non-small cell lung cancer: mediastinal staging

in the 1990s—meta-analytic comparison of PET and

CT. Radiology 1999;213:530–6.

[34] Hellwig D, Ukena D, Paulsen F, et al. Meta-analysis

of the efficacy of positron emission tomography with

F-18 fluorodeoxyglucose in lung tumors. Pneumologie

2001;55:367–77.

[35] Hicks R, Kalff V, MacManus M, et al. 18F-FDG PET

provides high-impact and powerful prognostic stratifi-

cation in staging newly-diagnosed non-small cell lung

cancer. J Nucl Med 2001;42:1596–604.

[36] Vesselle H, Pugsley J, Vallieres E, Wood D. The im-

pact of F18 positron-emission tomography on the sur-

gical staging of non-small cell lung cancer. J Thorac

Cardiovasc Surg 2002;124:511–9.

[37] Tinteren H, Hoekstra O, Smit E, et al. Effectiveness

of positron emission tomography in the preopera-

tive assessment of patients with suspected non-small

cell lung cancer: the PLUS multicentre trial. Lancet

2002;359:1388–92.

[38] Coleman R, Tesar R. Clinical PET: are we ready?

J Nucl Med 1997;38:5N–9N.

[39] Stevens H, Bakker P, Schlosser N, et al. Use of a dual-

head coincidence camera and 18F-FDG for detection

and nodal staging of non-small cell lung cancer: accu-

racy as determined by 2 independent observers. J Nucl

Med 2003;44:336–40.

[40] Akhurst T, Downey R, Ginsberg M, et al. An initial

experience with FDG-PET in the imaging of residual


disease after induction therapy for lung cancer. Ann

Thorac Surg 2002;73:259–64.

[41] Cerfolio R, Ojha B, Mukherjee S, et al. Positron emis-

sion tomography scanning with 2-fluoro-2-deoxy-

D-glucose as a predictor of response of neoadjuvant

treatment for non-small cell carcinoma. J Thorac Car-

diovasc Surg 2003;125:938–44.

[42] Port JL, Kent M, Kerestzes R, et al. Positron emission

tomography scanning poorly predicts response to pre-

operative chemotherapy in non-small cell lung cancer.

Ann Thorac Surg 2004;77:254–9.

[43] Weng E, Tran L, Rege S, et al. Accuracy and clinical

staging of mediastinal lymph node staging with FDG-

PET imaging in potentially respectable lung cancer.

Am J Clin Oncol 2000;23:47–52.

[44] Kluetz P, Villemagne V, Metzer G, et al. The case for

PET/CT: experience at the University of Pittsburgh

[abstract]. Clin Positron Imaging 2000;3:174.

[45] Hany T, Steinert H, Goerres G, et al. PET diagnostic

accuracy: improvement with in-line PET-CT system:

initial results. Radiology 2002;225:575–81.

[46] D’Amico T, Wong T, Harpole D, et al. Impact of com-

puted tomography–positron emission tomography fu-

sion in staging patients with thoracic malignancies.


[47] Wallace M, Silvestri G, Sahai A, et al. Endoscopic

ultrasound-guided fine-needle aspiration for staging

patients with carcinoma of the lung. Ann Thorac Surg

2001;72:1861–7.

[48] Tanaka K, Kubota K, Kodama T, et al. Extrathoracic

staging is not necessary for non-small cell lung cancer

with clinical stage T1–2 N0. Ann Thorac Surg 1999;

68:1039–42.

[49] Billings J, Wells F. Delays in the diagnosis and surgical

management of lung cancer. Thorax 1996;51:903–6.

[50] Grant D, Edwards D, Goldstraw P. Computed tomog-

raphy of the brain, chest, and abdomen in the pre-

operative assessment of non-small cell lung cancer.

Thorax 1988;43:883–6.

[51] The Canadian Lung Oncology Group. Investigating

extrathoracic disease in patients with apparently

operable lung cancer. Ann Thorac Surg 2001;75:

425–34.

[52] American Thoracic Society. Pretreatment in the evalua-

tion of non-small cell lung cancer. Am J Respir Crit

Care Med 1997;156:320–32.

[53] Patchell R, Tibbs P, Walsh J, et al. A randomized trial

of surgery in the treatment of single metastases to

the brain. N Engl J Med 1990;322:494–500.

[54] Guyatt G, Cook D, Griffith L, et al. Surgeon’s assess-

ment of symptoms suggesting extrathoracic metastases

in patients with lung cancer. Ann Thorac Surg 1999;

68:309–15.

[55] Davis P, Hudgins P, Peterman S, et al. Diagnosis of

cerebral metastases: double-dose delayed CT vs. con-

trast-enhanced MR imaging. AJR Am J Roentgenol

1991;156:1039–46.

[56] Yokoi K, Kamiya N, Matsuguma H, et al. Detection

of brain metastasis in potentially operable non-small

cell lung cancer: a comparison of CT and MRI. Chest

1999;115:714–9.

[57] Glazer H, Weyman P, Sagel S, et al. Non functioning

adrenal masses: incidental discovery on computed

tomography. AJR Am J Roentgen 1982;139:81–5.

[58] Eggesbo H, Glazer G, Gross B, et al. Clinical impact

of adrenal expansive lesions in bronchial carcinoma.

Acta Radiol 1996;37:343–7.

[59] Pearlberg J, Sandler M, Beute G, et al. T1N0M0 bron-

chogenic carcinoma: assessment by CT. Radiology

1985;157:187–90.

[60] Gillams A, Roberts C, Shaw P, et al. The value of

CT scanning and percutaneous fine needle aspiration

of adrenal masses in biopsy-proven lung cancer. Clin

Radiol 1992;46:18–22.

[61] Korobkin M, Brodeur F, Francis I, et al. CT time-

attenuation washout curves of adrenal adenomas

and nonadenomas. AJR Am J Roentgenol 1998;170:

747–52.

[62] Boland G, Lee M, Gazelle G, et al. Characterization of

adrenal masses using unenhanced CT: an analysis of

the CT literature. Am J Roentgenol 1998;171:201–4.

[63] Glazer G, Woolsey E, Borrello J. Adrenal tissue

characterization using MR imaging. Radiology 1986;

158:73–9.

[64] Korobkin M, Lombardi T, Aisen M, et al. Characteri-

zation of adrenal masses with chemical shift and gado-

linium-enhanced MR imaging. Radiology 1995;197:

411–8.

[65] Erasmus J, Patz E, McAdams H, et al. Evaluation

of adrenal masses in patients with bronchogenic

carcinoma using 18F-fluorodeoxyglucose positron

emission tomography. AJR Am J Roentgenol 1997;

168:1357–60.

[66] Boland G, Goldberg M, Lee M, et al. Indeterminate

adrenal mass in patients with cancer: evaluation at

PET with 2-F-18-fluor-2-deoxy-D-glucose. Radiology

1995;194:131–4.

[67] Yun M, Kim W, Alnafisi N, et al. 18F-FDG PET

in characterizing adrenal lesions detected on CT or

MRI. J Nucl Med 2001;42:1795–9.

[68] Tornyos K, Garcia O, Karr B, et al. A correlation study

of bone scanning with clinical and laboratory find-

ings in the staging of non-small cell lung cancer. Clin

Nucl Med 1991;16:107–9.

[69] Quinn D, Ostrow L, Porter D, et al. Staging of non-

small cell bronchogenic carcinoma: relationship of

the clinical evaluation to organ scans. Chest 1986;89:

270–5.

[70] Earnest I, Ryu J, Miller G, et al. Suspected non-small

cell lung cancer: incidence of occult brain and skeletal

metastases and effectiveness of imaging for detec-

tion-pilot study. Radiology 1999;211:137–45.

[71] Marom E, McAdams H, Erasmus J, et al. Staging non-

small cell lung cancer with whole body PET. Radiol-

ogy 1998;168:187–94.

[72] Durski J, Srinivas S, Segall G. Comparison of FDG-


PET and bone scans for detecting skeletal metastases

in patients with non-small cell lung cancer. Clin Posi-

tron Imag 2000;3:97–105.

[73] Valk P, Pounds T, Hopkins D, et al. Staging non-small

cell lung cancer by whole-body positron emission

tomographic imaging. Ann Thorac Surg 1995;60:

1573–82.

[74] Silvestri G, Tanoue L, Margolis M, et al. The nonin-

vasive staging of non-small cell lung cancer: the guide-

lines. Chest 2003;123:147S–56S.


Imaging of pleural and chest wall tumors

Michael J. Weyant, MDa, Raja M. Flores, MDa,b,*

aCardiothoracic Surgery, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021, USAbWeill Medical College, Cornell University, 525 East 68th Street, New York, NY 10021, USA

The visceral and parietal pleural lining of the tho- cal treatment, and the only long-term survivors are

rax is a serosal membrane arising from mesoderm.

Together these membranes invest the lungs, interlobar

fissures, ribs, diaphragm, and mediastinum. Pleural

tumors comprise multiple histologic forms of benign

and malignant types. Approximately 90% of pleural

tumors arise from metastatic deposits, whereas only

10% are truly primary pleural neoplasms. Histologic

types of primary pleural tumors include malignant

mesothelioma, fibrous pleural tumors, lymphoma,

pleural liposarcoma, and other less common types

(Box 1) [1–3].

The response of the pleura by an infiltrating dis-

ease process is manifested radiologically by effusion,

thickening, or calcification. Several imaging modali-

ties can be used to evaluate pleural masses, and the

most common noninvasive methods include chest ra-

diograph, CT, and MRI. Because of the inability of

plain radiographs to provide detailed information

regarding tissue specificity, CT and MRI are usually

used as adjunctive studies.

Malignant mesothelioma

Malignant pleural mesothelioma (MPM) is an un-

common, highly lethal tumor with an incidence of

2000 to 3000 cases and 1500 deaths per year in

the United States [4,5]. This tumor is thought to be

of mesodermal origin and has a strong relation to pre-

vious asbestos exposure. There is no effective medi-


doi:10.1016/S1547-4127(04)00033-7

* Corresponding author. Cardiothoracic Surgery, Me-

morial Sloan-Kettering Cancer Center, 1275 York Avenue,

New York, NY 10021.

E-mail address: [email protected] (R.M. Flores).

those who have undergone surgical resection by ex-

trapleural pneumonectomy or pleurectomy/decorti-

cation. Evaluation for resectability is a challenging

process involving multiple imaging modalities includ-

ing chest radiograph, CT, MRI, and, more recently,

18-flouro-deoxyglucose (FDG-PET) scanning. It is

important to attempt to rule out the presence of ad-

vanced disease because of the high morbidity and

mortality associated with surgical resection. The ra-

diographic criteria for resectability are listed in Box 2.

The most frequent radiologic abnormality found

initially is a pleural irregularity and unilateral pleural

effusion on plain chest radiograph (Fig. 1; Table 1)

[3]. Other findings occasionally found on chest ra-

diograph include osseous destruction, periosteal re-

action, or calcification [1,5]. Isolated pleural masses

without effusion are uncommon and occur in less

than 25% of patients in the initial radiograph [1].

CT provides greater detail in imaging and clinical

staging of MPM compared with chest radiography.

The most common CT finding of MPM is pleural

thickening, which usually involves the parietal and

visceral pleurae (Fig. 2) [6].

A large effusion without mediastinal shift is also

a common finding. The lack of mediastinal shift is

caused by the restrictive nature of the tumor peel.

MRI, which has also been used to evaluate MPM

and determine its resectability, has the characteristic

finding of increased signal intensity on T2-weighted

images compared with adjacent tissue, the sensitivity

of which approaches 100% [1]. MRI also allows the

visualization of diffuse and nodular pleural thicken-

ing and fissural involvement, which often occurs in

MPM. CT and MRI have been compared side-by-side

in two studies regarding staging and resectability.

Heelan et al [7] reported the results of 65 patients

s reserved.

Box 1. Classification of pleural and chestwall tumors

Pleural tumors

Malignant pleural mesotheliomaSolitary fibrous tumorPleural liposarcomaPleural metastasis

Chest wall tumors

Soft tissue sarcomasCartilaginous tumorsPrimary bone tumorsMetastatic lesionsTumors invading from contiguous

organs

Fig. 1. Plain radiograph demonstrating left side pleural effu-

sion in a patient who had MPM.

M.J. Weyant, R.M. Flores / Thorac Surg Clin 14 (2004) 15–2316

who underwent CT, MRI, and attempt at surgical re-

section in an effort to determine the accuracy of CT

and MRI with specific correlation to staging. The

accuracy of CT and MRI was relatively poor in most

areas. Their accuracy, respectively, to assess visceral

pleural involvement was 67% versus 58%; diffuse

chest wall involvement 65% versus 52%; invasion of

diaphragm 55% versus 82%; and invasion of lung

parenchyma 46% versus 69%. The ability to detect

Box 2. Imaging criteria for resectability inmalignant pleural mesothelioma

Resectable tumors

Preserved extrapleural fat planesNormal CT attenuation values and

MR signal intensity of adjacentstructures

Absence of extrapleural soft tissuemasses

Smooth inferior diaphragmatic surface

Unresectable tumors

Tumor encasement of the diaphragmInvasion of extrapleural soft tissueInfiltration or displacement of ribs by

tumorInvasion of essential mediastinal

structures

nodal involvement is even worse: 49% for CT versus

51% for MRI. The authors attribute these low accu-

racy rates to the diffuse nature of the tumor. MRI

was found to be slightly better in revealing solitary

foci of chest wall invasion, endothoracic fascia in-

volvement, and diaphragmatic muscle invasion.

These findings did not affect surgical decision-mak-

ing, and the authors advocate the routine use of

CT, not MRI, because of the increased cost of MRI.

Patz et al [4] performed a study comparing CT and

MRI in 41 patients to determine the resectability of

MPM. The unresectability rate of patients undergoing

thoracotomy was 30%. There was no significant dif-

ference in the predictive values of these modalities,

Table 1

Most common CT findings of malignant pleural mesothe-

lioma

Finding % Cases

Pleural thickening 92

Thickening of interlobar fissures 86

Pleural effusion 74

Loss of volume of involved hemithorax 42

Pleural calcification 20

Invasion of chest wall 18

Data from Refs. [2,6,18].

Fig. 2. (A) CT imaging demonstrating medial pleural mass and small pleural effusion in a patient who had MPM. (B) CT image

demonstrating diffuse pleural thickening in a patient who had MPM.

M.J. Weyant, R.M. Flores / Thorac Surg Clin 14 (2004) 15–23 17

leading the authors to recommend CT as the preferred

imaging choice for determining resectability.

Recently, the use of imaging with FDG-PET has

been applied to several tumor types, including tho-

racic malignancies (Fig. 3) [5]. MPM is reported to

have increased uptake on FDG-PET compared with

Fig. 3. FDG-PET uptake in patients who had solitary focus (A) and

benign pleural lesions in the majority of cases [8,9].

It has also been reported to have increased the detec-

tion of nodal metastasis compared with CT [9]. The

authors reported their experience in 63 patients who

had MPM who also underwent FDG-PET during a

4-year period [10]. Increased uptake was seen in all

diffuse (B) pleural involvement in patients who had MPM.


but one tumor. PET findings yielded sensitivities of

only 19% and 11% in determining tumor (T) and

node (N) status, respectively. A high standard uptake

value (SUV) did, however, correlate with the pres-

ence of N2 disease. The authors’ results demonstrated

that although FDG-PET is a poor predictor of stage, it

identified occult stage IV disease that was otherwise

undetected by CT scan alone in 10% of patients.

In addition to the benefit of identifying patients

who have stage IV disease, PET scans might have

prognostic significance. The authors evaluated their

now-larger cohort of 85 patients who underwent PET

scanning with the diagnosis of malignant pleural

mesothelioma and found that there was a linear rela-

tionship between increasing SUV and poor median

survival time. In addition, the relative risk of death in

patients who had an SUV of greater than four when

compared with an SUVof less than four was 3.3 (P =

0.03), which is a clinically significant finding that is

equivalent to impact of histology on survival. The

relative risk of death for nonepithelial histology com-

pared with epithelial histology was 3.2 (P = 0.03).

These findings suggest that PET can be used to strat-

ify patients for treatment [11].

Pleural plaques

Pleural plaques, which are usually a result of as-

bestos exposure, can present as diffuse thickening of

the visceral and parietal pleural layers. These lesions

can vary from diffuse, nodular lesions on the pleural

surface to lesions as wide as 6 cm. The coalescence

of pleural surfaces and the propensity for the lower

hemithorax can cause these lesions to be clinically

mistaken for diffuse MPM. These plaques are thought

to be formed by lymphatic transport of asbestos fi-

bers from the visceral to the parietal pleura, with the

fibers undergoing phagocytosis by macrophages that

secrete substances stimulating submesothelial fibro-

Fig. 4. (A, B) Solitary fibrous tumor attached to sta

blasts [12]. The physician should remember that it is

not uncommon for mesothelioma and pleural plaques

to be present simultaneously.

The distinction between benign pleural plaques

and mesothelioma can generally be recognized easily

by CT scan. Calcifications are usually present in pa-

tients who have a history of asbestos exposure, and

extensive calcification usually indicates benign pleu-

ral pathology [13]. In certain cases PET scans have

been useful in distinguishing between benign and

malignant pleural pathology [14]; however, when a

significant question arises, the gold standard in dis-

tinguishing a benign from a malignant pleural pro-

cess is surgical biopsy, preferably by way of the video

assisted thoracic surgery (VATS) technique.

Solitary fibrous tumor

Solitary fibrous tumors are thought to arise from

submesothelial mesenchymal cells. They compromise

only 10% of primary pleural tumors [15]. The inci-

dence of these tumors is highest in patients older

than 50 years of age [1]. Approximately 50% of be-

nign fibrous tumors of the pleura are asymptomatic

and are found incidentally on routine chest radio-

graphs. Symptomatic patients might present with

chest pain, cough, dyspnea, and fever. Pierre-Marie-

Bamberg syndrome (pulmonary osteoarthropathy and

clubbing) has been described in approximately 15%

of cases from the tumor production of hyaluronic acid

[16]. Doege-Potter Syndrome (refractory hypogly-

cemia) has been described in approximately 5% of

cases; these lesions might secrete an insulin-like sub-

stance [17].

The typical radiographic appearance is a rounded

or oval, frequently lobulated mass abutting the pleu-

ral surface. Calcification might be present in ap-

proximately 5% of cases. The location might be in

the fissure (30%), adjoining the mediastinal pleura

lk. Note change in position with respiration.

M.J. Weyant, R.M. Flores / Thorac

(18%), thoracic pleura (46%), or diaphragm (6%).

These lesions are reported to be benign in 63% of

patients and malignant in 37% of patients [15]. CT

imaging provides no pathognomonic findings to

evaluate the malignant potential of such lesions ex-

cept in cases in which a lesion is identified with a

stalk (it is more likely to be benign). Variation in

location during respiration on CT scan might also

indicate the presence of a lesion attached by a stalk

(Fig. 4A, B).

Pleural metastasis

Pleural metastases account for a large majority of

pleural-based tumors. The most common primary

tumors responsible for producing these lesions are

bronchogenic carcinoma (36%), breast cancer (25%),

lymphoma (10%), and ovarian and gastric carcinoma

(V5%; Fig. 5). The most common radiologic finding

of pleural metastasis is pleural effusion. Effusions

are thought to be produced by tumor blockage of

lymph ducts, thereby causing an exudative process.

Leung et al [18] reported on the CT findings of pleu-

ral lesions, and the most common radiologic findings

in patients who had pleural metastasis were effusion

in 88%, lung base involvement in 88%, and nodu-

larity in 50%. Other less common findings included

pleural thickening and plaque formation in metastatic

disease. Pleural lymphoma can produce pleural in-

volvement by extension from mediastinal lymph

nodes or, less commonly, by primary involvement. In-

vasive thymoma can also produce pleural thickening

by direct extension from the mediastinum, resulting

in pleural thickening or plaques.

Fig. 5. CT image of choriocarcinoma metastatic to the pleura.

Pleural liposarcoma

Pleural liposarcoma is a rare primary pleural tu-

mor. Only 100 cases have been reported in the lit-

erature. These lesions present as well-defined pleural

masses [19]. The pathognomonic CT and MRI find-

ings are a heterogeneous mixture of fat and soft tissue

densities. The surgeon should be able to differenti-

ate pleural liposarcoma from a lipoma, which has a

homogeneous tissue density consistent with fat.

Surg Clin 14 (2004) 15–23 19

Chest wall tumors

Tumors of the chest wall can be benign or ma-

lignant and can arise from any of its components,

including muscle, bone, adipose tissue, nerves, blood,

or lymphatic tissue. The majority of malignant chest

wall tumors are metastatic lesions from other organs

or they are the result of direct invasion of a tumor

from the lung parenchyma. Primary chest wall neo-

plasms represent only 5% of all thoracic neoplasms,

with approximately half of all primary chest wall

tumors being benign lesions [20].

Surgical excision is the modality of choice to treat

most chest wall tumors. There are few medical op-

tions for therapy. Advances in plastic and recon-

structive surgery have provided physicians with the

ability to reconstruct even the largest of chest wall

defects; however, accurate radiologic evaluation of

these tumors is essential in determining resectability

and planning reconstruction. Radiographic tools for

imaging these lesions consist mainly of plain radio-

graphs, CT, and MRI.

Metastatic lesions

Metastatic tumors are the most common chest

wall malignancy. The most common sources are

lung, breast, kidney, and prostate carcinomas. The

most common radiologic manifestation of these tu-

mors is a lytic lesion of one of the ribs [21]. Other

tumors such as metastatic thyroid carcinoma might

produce expansile, or ‘‘blown-out,’’ lesions in the

ribs [22].

Lung cancer with chest wall invasion

Direct extension of lung tumors occurs in up to

8% of cases and accounts for a significant proportion

of chest wall malignancies (Fig. 6) [21]. CT is su-

perior to plain radiographs in evaluating the extent of

chest wall invasion in these lesions because of the

large amount of bone destruction required for the

Fig. 6. T3 lung cancer demonstrating invasion of chest

wall on CT imaging.

Fig. 7. Pancoast MRI demonstrating invasion of lower

trunk of brachial plexus.


lesion to become visible on plain radiographs [21].

MRI has also been studied to evaluate the extent of

chest wall invasion. Padovani et al [23] evaluated

34 patients who had bronchogenic carcinoma and re-

ported a sensitivity of 90% by MRI for evaluat-

ing chest wall invasion. Webb et al [24] performed

a study comparing the accuracy of CT and MRI in

evaluating tumor classification and found that there

was no difference in the ability of MRI versus CT

in delineating chest wall invasion; however, they did

find that MRI might be superior in identifying

mediastinal invasion. Identifying lung cancers that

invade into the chest wall is essential to the pre-

operative planning and the clinical staging of these

patients, particularly since these tumors became clas-

sified as T3.

Pancoast tumors

Bronchogenic carcinomas that develop in the apex

of the lung and invade the superior pulmonary sulcus

were described in 1932 by Pancoast, who noted their

association to the clinical findings of Horner’s syn-

drome, unilateral arm pain, and wasting [25]. This

type of tumor represents less than 5% of all lung

cancers, and the survival rate is reported to range

from 15% to 56% (Fig. 7) [26,27]. Vital structures

such as the subclavian artery and vein, brachial

plexus, and vertebral bodies can be invaded early

in the course of growth of superior sulcus tumors.

Surgical resection is an important predictor of cure.

Accurate radiologic assessment of these tumors is

essential to ensure appropriate selection of patients

as candidates for resection to avoid the morbidity

associated with surgery.

These cancers are most often found on plain ra-

diograph, but more detailed studies are required to

delineate the extent of tumor invasion. CT and MRI

are used extensively, and crucial therapeutic deci-

sions are made based on these initial imaging studies.

The critical anatomic areas to be addressed include

(1) the apical layer of fat between the pleura and

the subclavian artery and vein, (2) tumor invasion

into the supraclavicular area, (3) invasion into the

subclavian vein and artery, (4) brachial plexus inva-

sion, and (5) involvement of adjacent ribs or verte-

bral bodies [28]. The decision to offer preoperative

chemotherapy or radiation therapy is based on deter-

mining invasion into these specific areas.

Heelan et al [29] reported experience with 31 pa-

tients who had superior sulcus tumors at Memorial

Sloan Kettering Cancer Center, comparing CT and

MRI in all patients. MRI was superior to CT, with an

accuracy of 94% compared with 63% in determining

invasion extending beyond the apex of the lung. This

increased accuracy is believed to be caused by MRI’s

ability to image in different planes, particularly the

coronal and sagittal planes. Laissy et al [30] reported

experience and efficacy with MR angiography to de-

termine the presence of vessel involvement. Other

authors reported that there might be an advantage to

CT in delineating vertebral body and rib destruction

[28]. The authors find that most patients initially

obtain a CT because of its convenience and availa-

Fig. 9. CT imaging of osteosarcoma arising from the lateral

chest wall.


bility. MRI has been demonstrated to be superior in

many aspects when compared with CT and should be

employed in this group of patients.

Primary osseous and cartilaginous lesions

Fibrous dysplasia is the most common benign tu-

mor arising from bone. These tumors account for ap-

proximately 30% of benign tumors of the chest wall

[21,22]. These fibrous tumors are slow-growing and

are usually seen in the lateral or posterior aspect of

one of the ribs. The tumor progresses by filling in the

medullary cavity with fibrous tissue, which can be

demonstrated on CT or MRI [21]. The usual radio-

logic finding is an expanding lytic lesion in one of

the ribs with a ground-glass appearance [22]. Osteo-

chondroma and chondromas together comprise an-

other 30% to 40% of benign chest wall tumors and

usually arise at the sternocostal junction.

Chondrosarcoma is the most common malignant

lesion arising from the bone, most frequently from

the anterior portion of the ribs and less frequently

from the sternum, scapula, or clavicle (Fig. 8) [20].

Chondrosarcomas frequently appear as a large, lobu-

lated mass arising from a rib with scattered calcifi-

cations consistent with a bony matrix [22]. These

lesions might be similar radiographically with their

benign counterparts, enchondromas, osteochondro-

mas, and osteoblastomas, therefore necessitating tis-

sue biopsy for diagnosis. The size of the lesion can be

used as a predictor of malignancy; lesions larger than

Fig. 8. CT imaging of chondrosarcoma arising from the

chest wall.

4 cm are considered to be malignant [21,22]. Osteo-

sarcomas are true malignant bony primary tumors and

usually arise from a rib. These tumors carry a worse

prognosis and have similar radiologic findings to

chondrosarcomas (Fig. 9).

Soft tissue tumors

Soft tissue tumors of the chest wall arise from

muscular, connective, or neural tissue. The most com-

mon soft tissue primary malignant tumors of the chest

wall include fibrosarcoma, malignant fibrous histio-

cytoma, and neurofibrosarcoma [20]. These lesions

appear radiographically similar and present as masses

of soft tissue density that might be associated with a

low-density necrotic area and areas of focal calcifica-

tion. Malignant schwannomas, rarer tumors, often

appear as rounded or elliptical masses adjacent to a

rib [21].

The most common benign soft tissue lesion of the

chest wall is the lipoma. These lesions can have in-

trathoracic and extrathoracic components, a dumb-

bell-shaped appearance, and tissue density consistent

with fat, which makes these lesions easy to identify

using CT or MRI. Neurogenic tumors are often be-

nign and usually appear radiographically to originate

from intercostal nerve roots. MRI is extremely useful

in identifying lesions that encroach on the neural

foramen. Hemangiomas are soft tissue masses that

are occasionally found in the chest wall and are

identified radiographically by the presence of phlebo-

liths and irregular tissue density [22]. Other rare le-

sions include plasmacytomas and desmoid tumors.

The imaging modalities used most frequently to

evaluate chest wall tumors include plain radiographs,


CT, and MRI. CT and MRI clearly provide supe-

rior resolution to plain radiographs; however, most

tumors are detected initially by a plain radiograph.

Because of its broad diagnostic spectrum, low sus-

ceptibility to artifacts, wide availability, and lower

cost, CT is usually the preferred initial imaging

choice in most institutions [31]. Although MRI is

expensive, time-consuming, and can be susceptible

to motion artifacts, it has several qualities that make it

desirable for evaluation of chest wall tumors. Multi-

planar imaging and superior soft tissue resolution

allow MRI to visualize the relations of tumors to

vessels and planes of tissue, which is helpful when

evaluating the extent of invasion of a tumor into

the chest wall [21,31]. The ability of MRI to help

determine the position of a neurogenic tumor in re-

lation to the neural foramen is also an advantage.

Lastly, the lack of iodinated contrast material allows a

viable option for imaging in patients who had allergic

reactions to standard intravenous contrast solutions.

Summary

MPM is a difficult disease to characterize radio-

graphically because of its diffuse nature and propen-

sity to infiltrate between tissue planes. Although

significant information is obtained by CT, MRI, and

PET, correlation with intraoperative findings is in-

consistent. Overall, CT and MRI are similar in pre-

dicting surgical resectability of pleural and chest wall

malignancies. MRI has a slight advantage in select

situations such as Pancoast tumors; however, CT

is less expensive and is sufficient in the majority of

cases. Because radiologic imaging cannot differenti-

ate benign from malignant lesions with 100% accu-

racy, surgical biopsy remains the gold standard for

diagnosis. Newer imaging modalities such as PET

scan and combined PET/CT might provide greater

information and warrant further study in the pre-

operative evaluation of pleural and chest wall tumors.

References

[1] Bonomo L, Feragalli B, Sacco R, Merlino B, Storto

ML. Malignant pleural disease. Eur J Radiol 2000;34:

98–118.

[2] Dynes MC, White EM, Fry WA, Ghahremani GG.

Imaging manifestations of pleural tumors. Radio-

graphics 1992;12:1191–201.

[3] Schmutz GR, Fisch-Ponsot C, Regent D, Sylvestre J.

Computed tomography (CT) and magnetic resonance

imaging (MRI) of pleural masses. Crit Rev Diagnost

Imaging 1993;34:309–83.

[4] Patz Jr EF, Shaffer K, Piwnica-Worms DR, Jochelson

M, Sarin M, Sugarbaker DJ, et al. Malignant pleu-

ral mesothelioma: value of CT and MR imaging in

predicting resectability. Am J Roentgenol 1992;159:

961–6.

[5] Marom EM, Erasmus JJ, Pass HI, Patz Jr EF. The

role of imaging in malignant pleural mesothelioma.

Semin Oncol 2002;29:26–35.

[6] Ng CS, Munden RF, Libshitz HI. Malignant pleural

mesothelioma: the spectrum of manifestations on CT

in 70 cases. Clin Radiol 1999;54(7):415–21.

[7] Heelan RT, Rusch VW, Begg CB, Panicek DM, Cara-

velli JF, Eisen C. Staging of malignant pleural meso-

thelioma: comparison of CT and MR imaging. Am J

Roentgenol 1999;172:1039–47.

[8] Schneider DB, Clary-Macy C, Challa S, Sasse KC,

Merrick SH, Hawkins R, et al. Positron emission to-

mography with f18-fluorodeoxyglucose in the stag-

ing and preoperative evaluation of malignant pleural

mesothelioma. J Thorac Cardiovasc Surg 2000;120:

128–33.

[9] Benard F, Sterman D, Smith RJ, Kaiser LR, Albelda

SM, Alavi A. Metabolic imaging of malignant pleural

mesothelioma with fluorodeoxyglucose positron emis-

sion tomography. Chest 1998;114:713–22.

[10] Flores RM, Akhurst T, Gonen M, Larson SM, Rusch

VW. Positron emission tomography defines metastatic

disease but not locoregional disease in patients with

malignant pleural mesothelioma. J Thorac Cardiovasc

Surg 2003;126:11–6.

[11] Flores R, Akhurst T, Gonen M, Larson S, Rusch V.

FDG-PET predicts survival in patients with malignant

pleural mesothelioma. ASCO Proceedings, Chicago,

IL, 2003.

[12] Hillerdal G. The pathogenesis of pleural plaques and

pulmonary asbestosis: possibilities and impossibilities.

Eur J Respir Dis 1980;61:129–38.

[13] Tiitola M, Kivisaari L, Zitting A, Huuskonen MS,

Keleva S, Tossavainen A, et al. Computed tomography

of asbestos-related pleural abnormalities. Int Arch

Occupat Environ Health 2002;75:224–8.

[14] Benard F, Sterman D, Smith RJ, Kaiser LR, Albelda

SM, Alavi A. Metabolic imaging of malignant pleural

mesothelioma with fluorodeoxyglucose positron emis-

sion tomography. Chest 1998;114:713–22.

[15] England DM, Hochholzer L, McCarthy MJ. Localized

benign and malignant fibrous tumors of the pleura.

A clinicopathologic review of 223 cases. Am J Surg

Pathol 1989;13:640–58.

[16] Cardillo G, Facciolo F, Cavazzana A, Capece G, Gas-

parri R, Martelli M. Localized (solitary) fibrous tumors

of the pleura: an analysis of 55 patients. Ann Thorac

Surg 2000;70:1808–12.

[17] Chamberlain M, Taggart D. Solitary fibrous tumor as-

sociated with hypoglycemia: an example of the Doege-

Potter syndrome. J Thorac Cardiovasc Surg 2000;119:

185–7.


[18] Leung AN, Muller NL, Miller RR. CT in differential

diagnosis of diffuse pleural disease. Am J Roentgenol

1990;154:487–92.

[19] Munk PL, Muller NL. Pleural liposarcoma: CT diag-

nosis. J Comput Assist Tomogr 1988;12:709–10.

[20] Incarbone M, Pastorino U. Surgical treatment of chest

wall tumors. World J Surg 2001;25:218–30.

[21] Schaefer PS, Burton BS. Radiographic evaluation of

chest-wall lesions. Surg Clin N Am 1989;69:911–45.

[22] Jeung MY, Gangi A, Gasser B, Vasilescu C, Massard

G, Wihlm JM, et al. Imaging of chest wall disorders.

Radiographics 1999;19:617–37.

[23] Padovani B, Mouroux J, Seksik L, Chanalet S, Sedat J,

Rotomondo C, et al. Chest wall invasion by broncho-

genic carcinoma: evaluation with MR imaging. Radi-

ology 1993;187:33–8.

[24] Webb WR, Gatsonis C, Zerhouni EA, Heelan RT,

Glazer GM, Francis IR, et al. CT and MR imaging

in staging non-small cell bronchogenic carcinoma: re-

port of the Radiologic Diagnostic Oncology Group.

Radiology 1999;178:705–13.

[25] Komaki R, Putnam Jr JB, Walsh G, Lee JS, Cox JD.

The management of superior sulcus tumors. Surg

Oncol 2000;18:152–64.

[26] Beale R, Slater R, Hennington M, Keagy B. Pancoast

tumor: use of MRI for tumor staging. South Med J

1992;85:1260–3.

[27] Komaki R, Roth JA, Walsh GL, Putnam JB, Vaporciyan

A, Lee JS, et al. Outcome predictors for 143 patients

with superior sulcus tumors treated by multidisciplinary

approach at the University of Texas M.D. Anderson

Cancer Center. Int J Radiat Oncol Biol Phys 2000;48:

347–54.

[28] Kuhlman JE, Bouchardy L, Fishman EK, Zerhouni

EA. CT and MR imaging evaluation of chest wall

disorders. Radiographics 1994;14:571–95.

[29] Heelan RT, Demas BE, Caravelli JF, Martini N, Bains

MS, McCormack PM, et al. Superior sulcus tumors:

CT and MR imaging. Radiology 1989;170:637–41.

[30] Laissy JP, Soyer P, Sekkal SR, Tebboune D, Servois V,

Sibert A, et al. Assessment of vascular involvement

with magnetic resonance angiography (MRA) in

Pancoast syndrome. Magn Reson Imaging 1995;13:

523–30.

[31] Landwehr P, Schulte O, Lackner K. MR imaging of

the chest: mediastinum and chest wall. Eur Radiol

1999;9:1737–44.


Imaging of the mediastinum: applications for

thoracic surgery

Dorith Shaham, MDa,*, Maria G. Skilakaki, MDb, Orly Goitein, MDa

aDepartment of Radiology, Hadassah University Hospital, Ein-Kerem, Jerusalem 91120, IsraelbDepartment of Radiology, Evangelismos General Hospital, 45–47 Ipsiladou Street, 10675 Athens, Greece

The mediastinum is a complex anatomic division pericardial reflection and posteriorly by the posterior

of the thorax, extending from the thoracic inlet

superiorly to the diaphragm inferiorly. The mediasti-

num is bordered anteriorly by the sternum, poste-

riorly by the vertebral column, and laterally by the

parietal pleura.

The mediastinum is further subdivided into supe-

rior, anterior, middle, and posterior divisions. The

exact anatomic borders of these divisions are unclear,

and different authors have different definitions [1].

Additionally, these borders do not have clear-cut

implications to the development of disease and do

not form barriers to the spread of disease; however,

each compartment of the mediastinum has its own

most common lesions, and knowing the location of

the mass, the patient’s age, and the presence or ab-

sence of symptoms considerably narrows the range of

possible diagnoses [2,3].

The complex anatomy of the mediastinum is best

understood by cross-sectional images provided by CT

or MRI.

According to Gray’s anatomy [4], the mediasti-

num is divided into superior and inferior compart-

ments by an imaginary line from the lower border of

the manubrium to the lower border of the fourth

thoracic vertebra. The anterior mediastinum lies an-

terior to the pericardium and ascending aorta. The

posterior mediastinum is bounded in front by the

trachea, the pulmonary vessels, and the pericardium

and behind by the vertebral column. The middle

mediastinum is bordered anteriorly by the anterior


doi:10.1016/S1547-4127(04)00039-8


E-mail address: [email protected] (D. Shaham).

pericardial reflection.

The bulk of the mediastinum is composed of the

heart and blood vessels. The carina, major airways,

and the esophagus are also identified easily in the

normal mediastinum and are surrounded by a variable

amount of fatty areolar tissue.

The contents of the anterosuperior mediastinum

include the thymus gland, the aortic arch and its

branches, the great veins, and the lymphatics. The

middle mediastinum contains the heart, pericardium,

phrenic nerves, carina and main bronchi, hila, and

lymph nodes. The contents of the posterior medias-

tinum include the esophagus, vagus nerves, sympa-

thetic nervous chain, thoracic duct, descending aorta,

azygos and hemiazygos veins, and paravertebral

lymph nodes.

Various imaging modalities

Almost half of all mediastinal masses do not

produce symptoms and are discovered on imaging

examinations obtained for other reasons [2,5,6]. In

recent years several developments in radiographic

techniques and immunohistochemistry have led to

more accurate preoperative delineation and histologic

diagnosis of mediastinal lesions. Today the presur-

gical evaluation of a mediastinal mass often involves

an array of imaging modalities and percutaneous or

transbronchial biopsy techniques [3,7].

Plain chest radiography

The standard posterior–anterior and lateral chest

roentgenogram continue to form the cornerstone of

s reserved.

D. Shaham et al / Thorac Surg Clin 14 (2004) 25–4226

diagnostic imaging [3]. High kilovoltage techniques

[>120 peak kilovoltage (KVp)] have significant ad-

vantages over low kilovoltage techniques (f70–

90 KVp) for demonstrating mediastinal interfaces

and providing better penetration of the mediastinum

[6]. In most cases deformation of the mediastinal

contours must be present for the radiologist to iden-

tify a mass, manifested as focal or widespread dis-

placement of normal structures or of the mediastinal

pleura [8]. Other features to be evaluated include

lesion shape, margins, location, the presence of single

or multifocal masses, the presence and type of cal-

cification (eg, rim-like calcification suggests a cystic

or vascular lesion), and associated findings such

as pleural involvement [9,10]. Mediastinal masses

are typically rounded and well circumscribed with

smooth margins. Occasionally they might be insepa-

rable from adjacent mediastinal structures and have

an obtuse angle or interface. A poor margin at the

pulmonary interface usually indicates invasiveness of

the lesion, but the most reliable sign of malignancy is

spread of disease [6,11].

Old films, if available, are often helpful. Obser-

vation of growth rate, duration, and change in nature

of the mass can contribute greatly to diagnostic

accuracy and guide further investigation [7,9].

CT

After an initial assessment using plain chest ra-

diography, the next step in radiologic evaluation is

CT. CT is extremely valuable in the radiographic

evaluation of the mediastinum and might be the only

imaging modality needed in the investigation of a

mediastinal mass [3,9,10,12–14]. CT is commonly

used to define and further characterize a mediastinal

abnormality diagnosed on plain chest radiographs.

Additionally, CT is also often used to evaluate the

mediastinum in patients who have normal chest

radiographs but a clinical reason to suspect medias-

tinal disease [12,14]. CT can depict vascular abnor-

malities and small masses that do not deform the

mediastinal contour on chest radiographs following

intravenous administration of contrast material [9].

The attenuation of a mediastinal lesion, as mea-

sured in Hounsfield units (HU), allows detection of

cysts, fat, soft tissue masses, calcification, and air and

is extremely important in the differential diagnosis of

mediastinal masses [15–17]. Masses can be catego-

rized according to their attenuation [12].

Fat attenuation

Fat attenuation (�70 to �100 HU) masses include

lesions composed primarily of or partially containing

fat or lipid-rich tissues. Abnormalities of fat distribu-

tion can be diffuse, as in mediastinal lipomatosis, or

focal, as in lipoma, thymolipoma, and lipoblastoma.

Most fatty masses are seen in the peridiaphragmatic

areas, and they most often represent herniation of

abdominal fat. As a general rule, the fatty nature of a

mediastinal mass is a strong indication toward benig-

nancy [12,17–19].

Low attenuation

Low attenuation (about �20 to +20 HU) masses

have a density greater than fat but less than muscle.

These masses are usually cystic and include con-

genital benign cysts (bronchogenic, esophageal du-

plication, neurenteric, pericardial, and thymic cysts),

meningocele, mature cystic teratoma, and lymphan-

gioma. Additionally, many tumors can undergo cys-

tic degeneration, especially after radiation therapy

or chemotherapy, and demonstrate mixed solid and

cystic components at CT, including thymoma, lym-

phoma, germ cell tumors, mediastinal carcinoma,

metastases to lymph nodes, and nerve root tumors.

Sometimes, when degeneration is extensive, such

tumors might mimic the appearance of congenital

cysts; however, clinical history and other manifesta-

tions allow correct diagnosis in most cases. Finally, a

mediastinal abscess or pancreatic pseudocyst might

also appear as a fluid-containing mediastinal cystic

mass [12,14,20–22].

High attenuation

High attenuation masses have a density greater

than that of muscle (>60 HU). The high density can

be attributed to calcium (calcified lymph nodes,

partially calcified primary neoplasms including

germ-cell tumors, thymoma, and neurogenic tumors,

calcified goiter, calcified vascular lesions) or to the

presence of fresh blood in a mediastinal hematoma

[12,16].

Enhancement

Enhancing masses show a significant increase in

attenuation following the injection of contrast. These

lesions are highly vascular and include substernal

thyroid, parathyroid glands, carcinoid tumor, para-

ganglioma, Castleman’s disease, lymphangioma, and

hemangioma [12,23–26].

In recent years the advent of spiral (helical) CT

has fundamentally revised the approach to scanning

the mediastinum [12]. Spiral CT data sets coupled

with a real-time volume-rendering technique allow

creation of accurate three-dimensional images, which,

although they are not required for diagnosis, can

D. Shaham et al / Thorac Surg Clin 14 (2004) 25–42 27

aid radiologists and referring clinicians by dem-

onstrating anatomic relationships and the extent of

disease. Volume-rendered images can be helpful in

assessing chest wall extension and collateral vessels

caused by obstruction of the superior vena cava [27].

Spiral CT also allows two-dimensional imaging in

various planes, including coronal, sagittal, and vari-

ous angled planes.

MRI

MRI is used less frequently compared with CT

in the evaluation of mediastinal masses, mainly

because of its lesser availability and higher cost

[3,10,28]; however, MRI has a capacity for multi-

planar imaging and the ability to image vessels, and it

can provide better tissue characterization than CT.

Additionally, MRI is excellent in the evaluation of

regions of complex anatomy such as the thoracic inlet,

the perihilar, paracardiac, and peridiaphragmatic

regions, and for the assessment of posterior mediasti-

nal or paravertebral masses [6,12,29]. MRI has com-

pletely replaced myelography for the evaluation of

potential spinal involvement of posterior neurogenic

tumors [3].

MRI is the primary imaging modality for investi-

gating mediastinal abnormalities that are suspected to

be vascular. Additionally, the difference in signal

between flowing blood and stationary tissues can be

used to demonstrate invasion or narrowing of the

large arteries and veins of the mediastinum. In se-

lected cases magnetic resonance angiography can be

used to demonstrate vascular disorders and distortion,

displacement, or stenosis of vessels by mediastinal

masses [6,12,30,31]. Conventional angiography and

venography, previously performed routinely in the

preoperative assessment of invasive primary medias-

tinal tumors, are now only occasionally used [7].

Additional indications for MRI include the diagnosis

of cystic lesions not of cystic attenuation on CT scans

(ie, identification of fluid with high protein content)

[9,29] and the differential diagnosis between residual

tumor and fibrous tissue in a patient who has lym-

phoma or carcinoma that has been treated [6,9,12,29].

Ultrasonography

Ultrasonography (US) is not commonly used in

the evaluation of mediastinal lesions, but it has been

reported as a useful alternative to more costly tech-

niques in the assessment of mediastinal masses in

selected cases, especially in children [3,9,10,32].

Transesophageal US has been introduced recently

to demonstrate mediastinal lesions adjacent to the

esophagus, particularly subcarinal lymph nodes and

cysts [33,34].

This method can determine whether or not a

lesion is cystic and demonstrate its relationship to

adjacent structures. Transesophageal US appears to

be the best method to verify if an esophageal impres-

sion is intramural or extrinsic to the esophageal wall,

thus giving additional information about the origin of

a mediastinal cyst [34].

Radionuclide imaging

Radionuclide imaging can be helpful in the differ-

ential diagnosis of certain mediastinal lesions. Iodine

scanning using iodine-123 or iodine-131 can demon-

strate functioning thyroid tissue while scanning with

technetium-99m (Tc-99) sestamibi can detect para-

thyroid tissue [21,35].

Preoperative differentiation between thymoma

and thyroid hyperplasia or between recurrent tumor

and scar tissue can be facilitated by somatostatin

receptor scintigraphy with indium-111-octreotide.

Additionally, thallium-201 scintigraphy has been

reported to enable distinction between normal thy-

mus, lymphoid follicular hyperplasia, and thymoma

in patients who have myasthenia gravis [35–38].

Metaiodobenzylguanidine (a precursor of epi-

nephrine) scans detect pheochromocytomas and neu-

roblastomas, and Tc-99 pertechnate scans can help

identify gastric mucosa in suspected neuroenteric

cysts [3,6].

Radionuclide scintigraphy has met with variable

success in the assessment of malignant lymphomas

over the past 30 years. The appearance of the anterior

mediastinum after treatment is quite variable, and

neither CT nor MRI has proven to be reliable in

excluding the presence of active disease in certain

cases. Gallium-67 citrate and thallium-201 scintigra-

phy have been reported recently as being highly

sensitive and specific in the detection of residual or

recurrent disease [39,40].

Finally, the role of fluorodeoxyglucose (FDG)

positron emission tomography (PET) in the assess-

ment of the extent of malignant mediastinal tumors

and its utility for initial staging and for predicting

prognosis are under investigation [6,38,41–43], and

initial results seem to be promising [38]. Recently,

combined PET-CT scanners have been introduced

that might further facilitate the diagnosis and fol-

low-up of mediastinal masses.

Fig. 1. Normal thymus. Contrast-enhanced CT with medi-

astinal window settings of a 3-year-old child shows a smooth,

well-defined anterior mediastinal structure (arrows).


Differential diagnosis of mediastinal tumors by

compartments

Classification of mediastinal masses into anterior,

middle, and posterior compartments is a convenient

categorization method, although there are no anatom-

ical boundaries that limit the extension of masses

from these compartments. In general, the most com-

mon mediastinal tumor location is the anterior com-

partment (50–60% in most series) [44,45]. Anterior

mediastinal masses include thymoma, lymphoma,

teratoma, and germ cell tumors. The most frequent

lesions seen in the middle mediastinum are reactive

lymph nodes, bronchogenic cysts, and pleuroperi-

cardial cysts. Tumors arising in the posterior me-

diastinum tend to be neurogenic in origin (Box 1)

[3,20,36].

Thymic masses

The normal thymus is located anterior to the proxi-

mal ascending aorta and superior vena cava (SVC).

The gland is bilobed, with the left lobe usually larger

than the right (Fig. 1) [46]. It is the largest between

Box 1. Classification of the most frequentmediastinal masses according to theirtypical location

Anterior mediastinal masses

Thyroid massesThymic massesGerm cell tumorsLymph nodesPericardial cyst

Middle mediastinal masses

Lymph nodesCarcinoma of bronchusBronchogenic cystAneurysm of the aorta

Posterior mediastinal masses

Neurogenic tumorsExtramedullary hemopoiesisEsophageal massesDilated, ruptured aortaHiatal hernia

the ages of 12 to 19 years and has an attenuation of

30 HU at this stage. Later, fatty involution takes place

and the gland is gradually replaced by fat.

Thymoma

Thymoma is the most common primary tumor of

the anterior mediastinum (f20%) [46]. There is a

slight female predominance, and the typical present-

ing age is in the mid-40s. Approximately 30% of

patients who have thymoma have myasthenia gravis,

and 10% to 15% of all myasthenia gravis patients

have thymomas. These thymomas are less aggressive

and have a better prognosis. Hematologic disorders

such red cell aplasia and hypogammglobulinemia are

associated with thymoma. In patients who have

myasthenia gravis, CT is indicated even in the ab-

sence of pathology on the plain roentgenogram be-

cause 25% of thymomas are not apparent on plain

radiographs [47,48].

On CT, thymomas usually appear as oval, round,

or lobulated masses mostly in the location of the

normal thymus, related to the root of the aorta or

pulmonary artery. In most cases the contour of the

mass is smooth and well defined, and it usually grows

asymmetrically to one side of the anterior medias-

tinum. The mass might be completely or partially

outlined by fat or it might replace the anterior

mediastinal fat completely. The absence of fat planes

between the mass and the mediastinal structures does

not necessarily denote the presence of invasion [47].

Homogenous attenuation is common with values of

45 to 75 HU, and mild enhancement is seen following

contrast injection [47,49]. Low attenuation areas can

represent cyst formation, necrosis, or hemorrhage

[46,47,50]. Calcification, even when subtle, can be

detected easily by CT [47]. A reliable distinction


between benign and malignant thymoma based on

CT characteristics is often impossible. Nevertheless,

some CT features are considered to be suspicious of

tumor invasion, including heterogeneous mass atten-

uation, complete obliteration of fat planes, pericardial

thickening, encasement of mediastinal vessels, irregu-

lar interface with the adjacent lung, and focal or

diffuse pleural thickening (Fig. 2) [47,51]. Extension

of invasive thymomas into the posterior mediastinum,

retrocrural space, and retroperitoneum has been de-

scribed [51,52].

Treatment consists of surgical excision. Maintain-

ing clear surgical margins is of paramount importance

because even noninvasive thymomas can recur if not

excised completely.

Lymphoma

Mediastinal lymphadenopathy can be a manifesta-

tion of Hodgkin’s disease (HD), non-Hodgkin’s lym-

phoma (NHL), infection, metastases, or sarcoidosis

(Fig. 3) [2]. Lymphoma accounts for 20% of anterior

mediastinal abnormalities in adults and 50% in chil-

dren. Patients might experience chest pain, dyspnea,

dysphagia, shoulder pain, congestive heart failure,

hypotension, and SVC syndrome. HD involves the

anterior mediastinum or paratracheal region in 90% to

100% of patients. HD typically spreads in contiguous

lymph node groups then spreads to the anterior

mediastinal compartment [46,53]. Additional thoracic

manifestations include pleural or pericardial effusion,

sternal erosion, and chest wall erosion. Pulmonary

involvement occurs in up to 11% of patients [54]. Low

attenuation areas associated with necrosis are seen in

Fig. 2. Invasive thymoma. (a) Contrast-enhanced CT with mediasti

mass with solid (short, white arrow) and fluid (black arrow) atten

hemithorax. (b) Section at the level of the heart demonstrates invas

the rib cage (long, thin arrow) and muscle infiltration (thick arro

arrow). The patient had previously undergone thoracotomy for res

20% to 50% of newly diagnosed cases of HD. The

presence of necrotic nodes has no prognostic value

[55]. NHL in the chest characteristically involves the

middle mediastinum. Extrathoracic disease is also

present in 90% of patients. Adenopathy in the cardio-

phrenic angle is typical for NHL and an unusual site

for HD [56].

Lymphoma is treated nonsurgically by chemother-

apy and radiotherapy. Calcification can be seen in HD

after treatment.

Thyroid masses

Substernal thyroid abnormality is defined as the

presence of thyroid tissue below the thoracic inlet.

Goiter

Substernal goiter represents 10% of mediastinal

masses. Most thyroid tumors (75–80%) arise from a

lower pole or the isthmus and extend into the anterior

mediastinum. The remaining 20% to 25% arise from

the posterior aspect of each lobe and involve the

posterior mediastinum.

Characteristic imaging features include a well-

defined mass with a spherical or lobulated border

continuous with the thyroid gland in the neck [50].

Thyroid tissue has high density before contrast injec-

tion (>100 HU) and undergoes intense immediate and

prolonged enhancement after contrast injection. At-

tenuation of intrathoracic goiter is usually higher than

muscle but less than that of the thyroid gland itself.

Low-density areas representing cysts or hemorrhage

are identified easily on postcontrast scans because

they do not enhance, contrary to normal thyroid tissue.

nal window settings demonstrates an extensive heterogenous

uation occupying the left and the right (long, white arrow)

ion into the left anterior chest wall, including destruction of

w). Also note invasion of the left pericardium (short, thin

ection of an invasive thymoma; this is a recurrent tumor.

Fig. 3. Mediastinal lymphadenopathy. Contrast-enhanced CT with mediastinal window settings reveals lymphadenopathy in the

(a) precarinal and retrocarinal regions (arrows), (b) subcarinal region (arrow), and (c) azygo–esophageal recess (arrow). These

lymph nodes have areas of low attenuation caused by necrosis.


Displacement or narrowing of the trachea is typi-

cal. Retrotracheal position of the goiter can occur,

with splitting of the trachea and the esophagus.

Calcifications are common. Benign calcifications

are well defined with a nodular, curvilinear, or circular

configuration. Malignant calcifications are usually a

group of fine dots corresponding to the psammoma

bodies found in papillary and follicular carcinoma of

the thyroid [57]. Primary thyroid cancer presents only

rarely in the anterior mediastinum, but it can invade it

as a direct extension.

Differentiation of benign from malignant thyroid

masses on CT is not possible unless obvious invasion

beyond the thyroid gland with invasion into the

mediastinal fat or chest wall vessels and lymphade-

nopathy are evident [50,58]. Fine needle aspiration

biopsy is not always possible and is rarely reliable for

excluding malignancy.

CT is currently the imaging modality of choice for

determining the presence and extent of such masses

and whether or not impingement on adjacent struc-

tures is present. MRI has a limited role (if any) in

imaging these masses because of its low sensitivity in

detecting calcifications and its high cost [59]. Radio-

nuclide imaging is an accurate method of determining

the thyroid nature of an intrathoracic mass. Iodine-131

is the agent of choice, but iodine-123 and Tc-99m are

also employed [60]. Proper imaging provides the

surgeon with all the relevant information to choose a

surgical versus a conservative approach [61].

Germ cell tumors

Germ cell tumors are thought to originate from

pluripotent primitive germ cells. They usually occur

in the gonads themselves. Extragonadal germ cell tu-

mors are considered to arise from multipotent cells

that are misplaced along midline structures during

their migration from the urogenital ridge to the

primitive gonad [50].

Mediastinal germ cell tumors represent only 1% to

3% of all germ cell neoplasms, and the anterior

mediastinum is the most common extragonadal site.

These tumors represent 15% of anterior mediastinum

tumors in adults and 24% in children [62]. They

occur in young adults between the ages of 20 to

Fig. 4. Teratoma. Contrast-enhanced CT with mediastinal

window settings reveals a well-defined, encapsulated (dotted

arrow points to capsule) anterior mediastinal mass with

heterogeneous density consisting of fluid (thick arrow) and

fat (thin arrow).


40 years. Women tend to develop benign tumors,

whereas men are prone to developing the malignant

germ cell tumors (Fig. 4) [46,63].

Neurogenic tumors

Most neurogenic tumors arise in the paraspinal

region, originating in an intercostal or sympathetic

nerve. Tumors of neural tissue origin represent 20%

of all primary mediastinal tumors in adults and 35%

in children [50].

Peripheral nerve tumors

The majority of peripheral nerve tumors arise

from an intercostal nerve. Histologic classification

includes neurilemoma (schwannoma), neurofibroma

(plexiform and nonplexiform types), and neurogenic

sarcoma (malignant schwannoma; Fig. 5).

Most peripheral nerve tumors are benign, and com-

plete surgical excision is associated with excellent

prognosis. When malignant, these tumors are aggres-

sive and commonly present with metastases, mainly to

the lungs [64–66].

Sympathetic ganglia tumors

Tumors of sympathetic ganglia the include gan-

glioneuroma, ganglioneuroblastoma, and neuroblas-

toma, a histologic continuum from differentiated

benign tissue to frank malignancy. Ganglioneu-

roma is a benign tumor occurring in children and

young adults [65]. Ganglioneuroblastoma includes

varying degrees of malignancy and occurs in children

younger than 10 years of age. Neuroblastoma is a

highly malignant tumor occurring in children younger

than 5 years of age, and in this age group a posterior

mediastinal mass is considered to be a neuroblastoma

until proven otherwise [50]. Vanillymendelic acid, ho-

movanillymendelic acid, and cystathionine are found

to be elevated in 90% of patients who have neu-

roblastomas [65]. Radiographically, they present as

elongated, elliptical masses extending over three to

five vertebral bodies. The elongated tapering config-

uration of sympathetic ganglia tumors help distin-

guish them from other neurogenic tumors. On CT

scans they appear well margined with homogeneous

or heterogeneous attenuation. Calcifications are dem-

onstrated in 25% of patients. Erosion of the nearby

vertebral bodies or ribs is seen more frequently in

malignant tumors [67,68]. Neuroblastomas can also

show invasion of posterior mediastinal structures and

a tendency to cross the midline.

Ganglioneuromas are benign and slow-growing,

and surgical excision offers a cure. Neurobalstomas

are highly aggressive, presenting as metastatic dis-

ease to regional lymph nodes, skeleton, and liver in

some patients, in whom 5-year survival does not

exceed 30%. The prognosis for ganglioneuroblas-

toma varies and relates to the patient’s age (younger

patients show a better outcome), stage, and histologic

tumor type [50].

Mediastinal cysts

Cystic masses of the mediastinum are well-de-

fined, round, epithelium-lined masses that contain

fluid. Mediastinal cysts represent 15% to 20% of

mediastinal masses [21].

Bronchogenic cyst

Bronchogenic cyst is a congenital abnormality

caused by ventral budding of the tracheobronchial

tree during embryogenesis (Figs. 6). Pseudostrati-

fied columnar respiratory epithelium lines these cysts,

and cartilage, smooth muscle, and mucous glands are

evident in the walls. The content of these cysts is

serous fluid or a mixture of mucus and protein. They

occur mainly near the carina but can also be found in

the middle or posterior mediastinum. Bronchogenic

cysts can also be found within the lung parenchyma,

pleura, or diaphragm [21]. Other congenital abnor-

malities such as lobar emphysema, pulmonary seques-

tration, or a pedicle attaching the cyst to adjacent

structures can also be seen. Most patients are asymp-

tomatic, but compression of adjacent structures can

cause symptoms such as chest pain, cough, dysp-

nea, fever, and purulent sputum [20,21]. CT scans

demonstrate a round mass with an imperceptible wall.

Attenuation values of the cyst content vary from

clear fluid to soft tissue attenuation values resulting

Fig. 5. Malignant peripheral nerve sheath tumor. (a) Posterior–anterior chest radiograph demonstrates a spherical mass arising

from the mediastinum on the left (arrows). (b) Lateral chest radiograph confirms that the mass is posterior (arrows). (c) Non-

contrast-enhanced CT with mediastinal window settings shows a large, round mass (thick arrows) in the left paraspinal region

with scoliosis and enlargement of the neural foramen on this side (thin arrow). (d) Coronal T1-weighted MRI. A large mass

isointense to muscle is noted in the left paraspinal region abutting the vertebral bodies. (e) Coronal T1-weighted MRI follow-

ing gadolinium administration. The aforementioned mass demonstrates heterogeneous enhancement. Scalloping of the left

lateral aspects of the midthoracic vertebrae is also seen. Scoliosis curving toward the left side is evident. (Courtesy of Paul

Cronin, MD, University of Michigan, Ann Arbor, MI.)


Fig. 5 (continued ).

Fig. 7. Neuroenteric cyst. Contrast-enhanced CT with medi-

astinal window settings shows a posterior mediastinal mass

with viscous content measuring 65 HU abutting the de-

scending aorta and esophagus (arrows).


from a high content of protein debris or hemorrhage.

Cysts containing calcifications have also been de-

scribed [69].

Surgical excision is indicated in symptomatic

patients. Young patients are also advised to remove

these cysts because of the low surgical risk and the

possibility of complications such as infection, hem-

orrhage, and neoplasia.

Gastroenteric (neuroenteric) cyst

Esophageal duplication cysts are uncommon. The

majority occur within the wall of the esophagus or

Fig. 6. Bronchogenic cyst. Contrast-enhanced CT with

mediastinal window settings shows a well-defined poste-

rior mediastinal mass with clear fluid content and an im-

perceptible wall (arrows). The mass is located adjacent to

the trachea.

adjacent to it (see Fig. 7), and they are usually lined

entirely or partially by gastric or small intestine

mucosa. The term neuroenteric cyst designates the

association with spinal column abnormalities [70].

The presence of ectopic gastric mucosa (50% of

patients) can cause hemorrhage, perforation, or in-

fection. On CT it resembles a bronchogenic cyst, the

only clue being the esophageal proximity or a thicker

wall. Patent communication to the gastrointestinal

tract is rare when cysts are connected to the esopha-

gus. Upper gastrointestinal barium studies demon-

strate extrinsic or intramural esophageal compression

[70]. Radionuclide studies with Tc-99 sodium per-

technetate can identify the ectopic gastric mucosa

existing in 50% of patients [71]. Neuroenteric cysts

demonstrate a fibrous connection to the spine or an

intraspinal component [50]. Association with verte-

bral body anomalies is common. The majority of

cysts present in the posterior mediastinum above the

level of the carina [21]. CT and MRI characteristics

are similar to other foregut cysts. MRI optimally

demonstrates the extent and degree of the spinal

involvement [72].

Pericardial cyst

Pericardial (mesothelial) cysts are a result of

aberrations in the formation of coelomic cavities.

The cysts usually contain clear fluid and the walls

are composed of a single layer of mesothelial cells

and connective tissue [21]. The majority of patients

are asymptomatic and discovered incidentally. Peri-

cardial cysts vary in size and shape. Seventy percent


of the cysts arise in the right and the remainder in the

left cardiophrenic angle or the superior portion of the

mediastinum [73]. On CT scans they appear as

round–oval cystic masses abutting the pericardium.

The benign nature of these lesions can be ascertained

by echocardiography, CT, and MRI.

Meningocele

Intrathoracic meningocele results from an abnor-

mal herniation of leptomeninges through either an

intervertebral foramen or a vertebral defect. The ma-

jority of meningoceles are diagnosed in adults, and

association with neurofibromatosis is frequent [74].

CT demonstrates a well-circumscribed, paraverte-

bral, low attenuation mass with distension of the

intervertebral foramina and rib anomalies, vertebral

anomalies, or scoliosis. When scoliosis is present, the

lesion occurs on the convex side [74]. MRI depicts

the continuity between the cerebrospinal fluid in the

thecal sac and the meningocele [28]. CT myelogra-

phy following intraspinal injection of contrast can

confirm the diagnosis by demonstrating filling of the

meningocele [72].

Invasion of mediastinal structures

Malignant primary mediastinal tumors remain a

relatively uncommon finding, although their inci-

dence seems to be increasing over the past decades;

however, when a malignant mediastinal tumor is

present, possible invasion of mediastinal structures

has to be determined preoperatively because a deci-

sion to resect the mass along with involved neighbor-

ing structures must be weighed against the morbidity

of such a procedure. In addition, the potential long-

term survival benefit must be considered [7].

In general, absolute contraindications to resection

of mediastinal masses are invasion of the myocar-

dium or the great vessels and invasion of a long

tracheal segment [7]. Overdiagnosis of invasion

should be avoided; direct contact between the tumor

and mediastinal structures and the absence of cleav-

age planes are not strictly reliable criteria for pre-

dicting invasion. Conversely, clear definition of fat

planes surrounding a tumor indicates the absence of

macroscopic invasion of adjacent structures [6,12].

Thymomas, germ-cell tumors, lymphomas, and

neurogenic tumors account for the vast majority of

primary mediastinal tumors in adults. Approximately

30% to 35% of thymomas, 20% of germ-cell tumors,

and 15% of nerve sheath tumors are invasive

[6,12,14]. Radical excision is the standard of care

for invasive thymic tumors and tumors of nerve

sheath origin, whereas chemotherapy is the primary

treatment modality for invasive germ-cell tumors. A

combination of chemotherapy and radiation therapy

is required in most cases of lymphoma [7,13,75–77].

In a recent study focused specifically on patients

who had malignant mediastinal tumors invading

adjacent organs or structures [7], the most commonly

invaded structure of the mediastinum was the peri-

cardium, followed by the pleura, the lung (mainly

invasion of the anterior segments of the upper lobes

or lingula), the phrenic nerve, and the SVC. In cases

of massive invasion of the pulmonary hilum or

extensive subpleural and pulmonary thymoma metas-

tases, a pneumonectomy cannot be avoided. When

clinical SVC syndrome is present and the vein is

invaded extensively by the tumor, total SVC replace-

ment is indicated. Widespread collateral venous cir-

culation or extensive thrombosis of subclavian veins

increases the likelihood of postoperative thrombosis

[7,78].

Invasive thymomas infiltrate adjacent structures

including the SVC, great vessels, airways, lungs, and

chest wall. An irregular interface with the adjacent

lung is suggestive of invasion [47]. There can also be

spread to the pericardium and pleura along pleural

reflections and along the aorta through the diaphragm

into the abdomen and retroperitoneum, usually on one

side of the body only (see Fig. 2) [9,12]. It is therefore

important, when investigating a potentially invasive

thymoma, to include the deep pleural reflections and

the upper abdomen on any imaging examination [6,9].

Rarely, thymoma might appear as predominantly

pleural disease, usually unilateral, with nonspecific

radiographic patterns such as pleural thickening,

pleural masses, or diffuse, nodular, circumferential

pleural thickening that encases the ipsilateral lung.

The latter manifestation mimics malignant mesothe-

lioma or metastatic adenocarcinoma [6,47].

Neural tumors in the posterior mediastinum usu-

ally arise close to the spine and can extend through the

neural exit foramina into the spinal canal. This intra-

dural extension is not necessarily a sign of malig-

nancy, but it requires a combined neurosurgical and

thoracic surgical approach [6,79]. Bone invasion,

when present, is a strong indication of malignancy [9].

When a nerve sheath tumor is localized it is not

possible to distinguish between benign and malig-

nant tumors [6]. Tumors that grow on intercostal

nerves can cause rib erosion. When a sclerotic bor-

der is present, the possibility of malignancy is low.

Conversely, spreading of multiple ribs with erosion

or frank destruction is suggestive of a malignant

lesion [6,9].


Vascular supply of mediastinal tumors

The vascular supply of mediastinal tumors depends

on their anatomic location, extent, and histopathologic

features. In general, congenital mediastinal cysts and

the majority of neurogenic tumors are hypovascular

lesions, most commonly supplied by the intercostal

vessels [13,34,45,79,80]. Anterior intercostal arteries

arise from the internal mammary (thoracic) artery and

posterior intercostal arteries arise from the thoracic

Table 1

Definition of 1996 American Joint Committee on Cancer/Union

mediastinal lymph nodes

Number Name L

1 Highest mediastinal nodes A

br

to

2 Upper paratracheal nodes A

m

of

3A

3P

Prevascular and

Retrotracheal nodes

A

an

4 Lower paratracheal nodes O

be

m

ri

lo

en

O

a

ao

br

br

co

5 Subaortic nodes (aortopulmonary window) L

le

of

pl

6 Paraaortic nodes (ascending aorta or phrenic) A

ar

a

7 Subcarinal nodes C

lo

8 Paraesophageal nodes A

le

su

9 Pulmonary ligament nodes W

po

10 Hilar nodes P

pl

in

Lymph node stations 1–9 are N2 nodes and lie with the mediasina

N1 nodes that are distal to the mediastinal pleural reflection and w

aorta [81]. In each intercostal space there are one

posterior and two anterior intercostal veins. The

anterior veins drain into the internal mammary veins,

the superior four posterior veins drain into the bra-

chiocephalic (innominate) veins, and the lower eight

posterior intercostal veins drain into the azygos vein

on the right and the accessory hemiazygos and hemi-

azygos veins on the left [81,82].

Thymomas are supplied by the internal mammary

arteries [44], which are located within the adipose and

Internationale Contre le Cancer classification of regional

ocation

bove a horizontal line at the upper rim of the left

achiocephhalic (innominate) vein where it ascends

the left, crossing in front of the trachea at its midline

bove a horizontal line drawn tangential to the upper

argin of the aortic arch and below the inferior boundry

number 1 nodes

nterior to the aortic arch branches (3A)

d posterior to the trachea (3P)

n the right: to the right of the midline of the trachea

tween a horizontal line drawn tangential to the upper

argin of the aortic arch and a line extending across the

ght main bronchus at the upper margin of the right upper

be bronchus and contained within the mediastinal pleural

velope; azygos nodes are included in this station

n the left: to the left of the midline of the trachea between

horizontal line drawn tangential to the upper margin of the

rtic arch and a line extending across the left main

onchus at the upper margin of the left upper lobe

onchus medial to the ligamentum arteriosum and

ntained within the mediastinal pleural envelope

ateral to the ligamentum arteriosum or the aorta or

ft pulmonary artery and proximal to the the first branch

the left pulmonary artery and within the mediastinal

eural envelope

nterior and lateral to the ascending aorta and the aortic

ch or brachiocephalic or the brachiocephalic artery, beneath

line tangential to the upper margin of the aortic arch

audad to the tracheal carina but not associated with the

wer lobe bronchi or arteries within the lung

djacent to the wall of the esophagus and to the right or

ft of the midline below the tracheal carina, excluding

bcarinal nodes

ithin the pulmonary ligament, including those in the

sterior wall and lower part of the inferior pulmonary vein

roximal to lobar nodes and distal to the mediastinal

eural reflection and the nodes adjacent to the bronchus

termedius on the right

l pleural envelope. Lymph node station 10 is included in the

ithin the visceral pleura.


connective tissues bordered anteriorly by the costal

cartilage and intercostal muscles and posteriorly

by the endothoracic fascia and transverse thoracic

muscles [83]. Occasionally the arterial supply of

thymic tumors can be derived from the inferior

thyroid arteries [4]. Venous drainage of the thymic

gland is through a variable number of thin vessels

(veins of Keynes) that drain the thymus from its

posterior surface into the anterior aspect of the left

innominate vein. Frequently, one or more veins might

join together to form a common trunk before opening

into the left innominate vein [4,13,84]. Additionally,

one or two small veins from the upper pole of thymus

end in the inferior thyroid veins [4].

Intrathoracic goiters receive blood supply from

the superior and inferior thyroid arteries. The superior

thyroid arteries arise from the external carotids and

the inferior thyroid arteries arise from the thyrocer-

vical trunks [4,13,84]. Venous drainage is through the

superior, middle, and inferior thyroid veins. The

superior and middle thyroid veins end in the internal

jugular veins, and the inferior thyroid veins end in the

brachiocephalic veins. Sometimes the inferior thyroid

veins might join together to form a common trunk

ending in the left brachiocephalic vein [4,84].

Since the advent of cisplatin-based chemotherapy,

the role of surgery as the primary treatment of germ

cell tumors has been more limited [7,75]. Following

initial chemotherapy, persistent radiographic abnor-

malities accompanied by elevated marker levels in

the serum that continue to rise denote persistent car-

cinoma; these patients should be treated with an

alternative chemotherapy regimen [75]. Resection of

Fig. 8. Mediastinal lymphadenopathy. (a) Noncontrast CT with

paratracheal lymph nodes (dotted arrow, station 4R) and bilateral

lymph nodes are seen on the left (thin arrow, station 6). (b) Non

lymphadenopathy further down in the lower paratracheal region (

axilla (thick arrows).

residual masses after chemotherapy has been advo-

cated in patients whose marker levels have normal-

ized [7,75,85 – 87]. The blood supply of these

residual tumors varies according to their size, precise

anatomic location, histopathology, and the degree of

postchemotherapy necrosis.

Sampling procedures for mediastinal lymph nodes

Lymph nodes are widely distributed throughout

the mediastinum. Two systems that have been in used

for classifying regional lymph node stations for lung

cancer staging were unified in 1996 [88,89]. These

were the American Joint Committee on Cancer

(AJCC) classification, adapted from the work of

Naruke [90], and the classification of the American

Thoracic Society and the North American Lung

Cancer Study Group [91]. The unified classification

was adopted by the AJCC and the Prognostic TNM

Committee of the Union Internationale Contre le

Cancer. The following discussion of lymph node

sampling is according to this unified classification

(Table 1).

Surgical procedures used for mediastinal lymph

node sampling include cervical mediastinoscopy, an-

terior mediastinotomy, and video-assisted thoraco-

scopic surgery (VATS).

Regional lymph nodes accessible by cervical

mediastinoscopy include stations 1, 2, 4, and 7 (an-

terior and superior nodes). When performing anterior

or parasternal mediastinoscopy, lymph node stations

5 and 6 can be sampled. VATS offers a panoramic

mediastinal window settings demonstrates enlarged lower

axillary lymphadenopathy (thick arrows). Small para-aortic

contrast CT with mediastinal window settings demonstrates

thin arrow, station 4R). Lymphadenopathy is noted in both


view of the ipsilateral hemithorax including the

hilum, mediastinum, visceral pleura, and chest wall.

Lymph node stations accessible by VATS in the right

hemithorax include 4R, 9R, and 10R; in the left

hemithorax 5, 6, 9L, and 10L are accessible. Right-

sided thoracoscopy allows sampling of lymph node

stations 3A, 3P, 7 (posterior and inferior nodes), and 8

(Fig. 8) [92].

In studies documenting the size of normal medias-

tinal lymph nodes by CT, 95% of these lymph nodes

were less than 10 mm in diameter [93,94]. The short

axis nodal diameter is used for measuring mediastinal

lymph nodes because it was found to be the best CT

predictor of nodal volume [95]. FDG-PET scanning

adds to the accuracy of detecting lymph node involve-

ment in lung cancer staging and has a particularly high

negative predictive value [96–98].

Postoperative complications

Postoperative mediastinal complications include

mediastinal hemorrhage, mediastinitis, and chylo-

thorax.

Significant hemorrhage can follow thoracic opera-

tions, particularly procedures involving the heart and

great vessels, which require cardiopulmonary bypass.

The clinical presentation is variable and might in-

clude retrosternal pain radiating to the back and neck.

With increased accumulation of blood in the medias-

tinum, signs and symptoms related to compression of

Fig. 9. Postoperative mediastinal hematoma. (a) Contrast-enhanc

rosternal dense fluid collection (short, thick arrow) measuring 50 H

collection anterior to it (long, thick arrow). These fluid collections

small, bilateral pleural effusions. (b) Contrast-enhanced CT with m

in the chest. A dense fluid collection is seen anterior to the ascendin

arrow). Two anterior air bubbles are also seen (thin arrows), cons

mediastinal structures, particularly veins, can occur

and manifest as dyspnea and cyanosis. With further

accumulation of blood, mediastinal tamponade can

develop, presenting with circulatory compromise

[6,99].

Plain chest radiographs might demonstrate widen-

ing of the mediastinal shadow, which can be focal or

general. The blood might also track extrapleurally

over the lung apices and give rise to apical capping

[100]. Severe hemorrhage can rupture into the pleural

cavity. Rapid widening of the mediastinum on serial

films is an important clue to the diagnosis of medias-

tinal hemorrhage. CT can show the characteristic

appearance of blood, the high density related to a

fresh clot, and the relationship of the hematoma to

adjacent mediastinal structures (Fig. 9).

Infection of the mediastinum is a relatively rare,

serious, and potentially fatal condition that currently

occurs most frequently following median sternotomy

for open-heart surgery. Postoperative mediastinitis

usually occurs between 3 days and 3 weeks following

surgery, but delayed manifestations can occur up to

months later. The clinical manifestations of mediasti-

nitis are fever, tachycardia, and chest pain, and when it

occurs postoperatively there might be wound ery-

thema, pain, effusion, and an unstable sternum [99].

The radiologic features of acute mediastinitis

include mediastinal widening and pneumomediasti-

num. On the lateral chest radiograph, an abnormal

soft tissue density, air – fluid levels (representing

abscess formation), and sternal dehiscence might be

seen. Accompanying pleural effusion on one side or

ed CT with mediastinal window settings demonstrates ret-

U. Note the sternotomy site (thin arrow) and a similar fluid

are consistent with postoperative hematomas. Also note the

ediastinal window settings. This slice is slightly lower down

g aorta and the pulmonary trunk, abutting the sternum (thick

istent with postoperative air.


bilaterally are common. These findings are recognized

more easily on CT scan. CT can also show associated

findings such as venous thrombosis or pericardial

effusion and contiguous infections such as emphyma,

subphrenic abscess, or cervical soft tissue infection

[6,99].

Distinguishing retrosternal hematomas from reac-

tive granulation tissue or cellulitis might be difficult,

as is differentiating osteomyelitis from postsurgical

changes in the sternum. Substernal fluid collections

and minimal amounts of air are normal for up to

20 days following sternotomy (Fig. 9). Air can be

seen on chest radiographs in the presternal or retro-

sternal soft tissues for up to 50 days following

sternotomy, so only newly appearing air collections

or collections that increase in size can be diagnosed

as gas-forming infections [101].

Mediastinitis should be diagnosed as early as

possible; delays in diagnosing this condition and

initiating treatment result in increased morbidity and

mortality. Treatment options include incision, de-

bridement and drainage of the involved area, the use

of closed irrigation systems, and using a tissue flap

(pectoralis or rectus abdominis muscle or omen-

tum) [99].

Chylothorax can develop 1 to 2 weeks following a

surgical procedure in the region of the aorta, esopha-

gus, or posterior mediastinum. The anatomy of the

thoracic duct is constant only in its variability [102].

The duct originates from the cysterna chili, a globular

structure 3 to 4 cm long and 2 to 3 cm in diameter

that lies adjacent to the vertebral column between

L3 and T10, just to the right of the aorta. Usually a

single thoracic duct enters the chest through the aortic

hiatus at the level of T12 to T10, just to the right of

the aorta. Above the diaphragm the duct lies on the

anterior surface of the vertebral column behind the

esophagus and between the aorta and the azygos vein.

At the level of T5 the duct courses to the left and

ascends behind the aortic arch into the left side of the

posterior mediastinum, where it passes adjacent to

the left side of the esophagus. In the root of the neck,

the thoracic duct passes behind the left carotid sheath

and jugular vein and enters the venous system at the

left jugulo–subclavian junction. There are several

anastomoses between the duct and the azygos, inter-

costals, and lumbar veins. This normal anatomic

description exists in slightly more than half of indi-

viduals. In the remainder there are two or more main

ducts in some part of its course [103].

Injury to the duct has occurred in almost every

known thoracic operation. The duct is most vulne-

rable in the upper part of the left side of the chest,

particularly during procedures that involve mobiliza-

tion of the aortic arch, left subclavian artery, or

esophagus. Because of the course of the duct, injury

below the level of T5 to T6 usually causes a right-

sided chylothorax, whereas injury above this level

results in a left-sided chylothorax [103].

There is usually an interval of 2 to 10 days

between rupture of the thoracic duct and the onset

of a chylous pleural effusion. This delay is caused by

the accumulation of lymph in the posterior mediasti-

num until the mediastinal pleura ruptures [103].

Chylous effusion is typically (but not necessarily)

milky, particularly during starvation, as might occur

following surgery. The diagnosis of chylous effusion

is made by measuring the triglyceride levels of the

effusion; levels above 110 mg/dL are regarded as

positive. Chylothorax should be differentiated from

pseudochylothorax, which is also milky but caused

by high levels of cholesterol or lecitin–globulin com-

plexes in the effusion. This condition characteris-

tically occurs in chronic pleural disease with pleural

thickening and chronic encysted effusion [104].

A chylous effusion on a plain chest radiograph

cannot be distinguished from pleural effusion result-

ing from other causes. It can be large or small,

unilateral or bilateral. On CT, the density of chyle

is indistinguishable from that of other effusions de-

spite the high fat content because it is also protein-

rich; therefore, the density of the effusion is not as

low as would be expected based on the rich fat con-

tent [104].

Conservative therapy for chylothorax includes

thoracostomy tube drainage, correction of fluid losses,

prevention of electrolyte imbalance, and parenteral

nutritional support. Surgical therapy is indicated when

the chylous effusion does not respond to conservative

management and there is no contraindication to sur-

gery. Surgery includes a combination of direct closure

of the thoracic duct–pleural fistula, suturing of the

leaking mediastinal pleura, and supradiaphragmatic

ligation of the duct [103].

Summary

The diagnostic approach to patients who have

mediastinal masses should include thorough preoper-

ative imaging. Once limited to plain radiographic

techniques, the radiologist now has a wide variety

of imaging modalities to aid in the evaluation of the

mediastinum. CT is the imaging modality of choice

for evaluating a suspected mediastinal mass or a

widened mediastinum, and it provides the most useful

information for the diagnosis, treatment, and evalua-

tion of postoperative complications.


References

[1] Armstrong P. Normal chest. In: Armstrong P, Wilson

AG, Dee P, Hansell DM, editors. Imaging of diseases

of the chest. 2nd edition. St. Louis (MO): Mosby;

1995. p. 15–47.

[2] Fraser RS, Pare JAP, Fraser RG, Pare PD. Diseases

of the mediastinum. In: Fraser RS, Pare JAP, Fraser

RG, Pare PD, editors. Synopsis of diseases of the

chest. Philadelphia: WB Saunders; 1994. p. 896–942.

[3] Kohman LJ. Approach to the diagnosis and staging

of mediastinal masses. Chest 1993;103:328S–30S.

[4] Williams PL, Warwick R, Dyson M, et al. Gray’s

anatomy. 37th edition. Edinburgh: Churchill Living-

stone; 1989.

[5] Wychulis AR, Payne WS, Clagett OT, Woolner

LB. Surgical treatment of mediastinal tumors. A

40-year experience. J Thorac Cardiovasc Surg 1971;

62:379–92.

[6] Armstrong P. Mediastinal and hilar disorders. In:

Armstrong P, Wilson AG, Dee P, Hansell DM, edi-

tors. Imaging of diseases of the chest. London:

Mosby; 2000. p. 789–892.

[7] Bacha EA, Chapelier AR, Macchiarini P, Fadel E,

Dartevelle PG. Surgery for invasive primary medias-

tinal tumors. Ann Thorac Surg 1998;66:234–9.

[8] Shaffer K, Pugatch RD. Diseases of the mediastinum.

In: Freundlich IM, Bragg DG, editors. A radiologic

approach to diseases of the chest. Baltimore: Williams

& Wilkins; 1992. p. 171–85.

[9] Templeton PA. Mediastinal lesions. In: Greene R,

Muhm JR, editors. Syllabus: a categorical course in

diagnostic radiology. Chest radiology. Oak Brook, IL:

Radiological Society of North America Inc; 1992.

p. 273–86.

[10] Merten DF. Diagnostic imaging of mediastinal

masses in children. AJR 1992;158:825–32.

[11] Pugatch R, Spirn PW. Mediastinal neoplasms. In:

Taveras JM, Ferrucci JT, editors. Radiology: diagno-

sis—imaging—intervention, Vol 1. Philadelphia: Lip-

pincott; 1995.

[12] Naidich DP, Webb WR, Muller NL, Krinsky GA, Zer-

houni EA, Siegelman SS. Mediastinum. In: Naidich

DP, Webb WR, Muller NL, Krinsky GA, Zerhouni

EA, Siegelman SS, editors. Computed tomography

and magnetic resonance of the thorax. Philadelphia:

Lippincott Williams & Wilkins; 1999. p. 37–159.

[13] Roviaro G, Rebuffat C, Varoli F, Vergani C, Macio-

cco M, Scalambra SM. Videothroacoscopic excision

of mediastinal masses: indications and technique.


[14] Tecce PM, Fishman EK, Kuhlman JE. CT evaluation

of the anterior mediastinum: spectrum of disease.

Radiographics 1994;14:973–90.

[15] Dedrick CG. Non-neoplastic disorders of the medias-

tinum. In: Taveras JM, Ferrucci JT, editors. Radi-

ology: diagnosis—imaging—intervention, Vol 1.

Philadelphia: Lippincott; 1995.

[16] Glazer HS, Molina PL, Siegel MJ, Sagel SS. Pictorial

essay: high-attenuation mediastinal masses on unen-

hanced CT. AJR 1991;156:45–50.

[17] Glazer HS, Wick MR, Anderson DJ, Semenkovich

JW, Molina PL, Siegel MJ, et al. CT of fatty thoracic

masses. AJR 1992;159:1181–7.

[18] Gaerte SC, Meyer CA, Winer-Muram HT, Tarver RD,

Conces Jr DJ. Fat-containing lesions of the chest.

Radiographics 2002;22:S61–78.

[19] Mullins ME, Stein J, Saini SS, Mueller PR. Preva-

lence of incidental Bochdalek’s hernia in a large adult

population. AJR 2001;177:363–6.

[20] McAdams HP, Kirejczyk WM, Rosado-de-Christen-

son ML, Matsumoto S. Bronchogenic cyst: imaging

features with clinical and histopathologic correlation.

Radiology 2000;217:441–6.

[21] Jeung MY, Gasser B, Gangi A, Bogorin A, Charneau

D, Wihlm JM, et al. Imaging of cystic masses of the

mediastinum. Radiographics 2002;22:S79–93.

[22] Kawashima A, Fishman EK, Kuhlman JE, Nixon MS.

CT of the posterior mediastinal masses. Radio-

graphics 1991;11:1045–67.

[23] Spizarny DL, Rebner M, Gross BH. CT of enhancing

mediastinal masses. J Comput Assist Tomogr 1987;

11:990–3.

[24] Doppman JL, Skarulis MC, Chen CC, et al. Parathy-

roid adenomas in the aortopulmonary window. Radi-

ology 1996;201:456–62.

[25] McAdams HP, Rosado-de-Christenson ML, Moran

CA. Mediastinal hemangioma: radiographic and CT

features in 4 patients. Radiology 1994;193:399–402.

[26] Kirsch CFE, Webb EM, Webb WR. Multicentric Cas-

tleman’s disease and POEMS syndrome: CT findings.

J Thorac Imaging 1997;12:75–7.

[27] Johnson PT, Fishman EK, Duckwall JR, Calhoun PS,

Heath DG. Interactive three-dimensional volume ren-

dering of spiral CT data: current applications in the

thorax. Radiographics 1998;18:165–87.

[28] Webb WR, Sostman HD. MR imaging of thoracic

disease: clinical uses. Radiology 1992;182:621–30.

[29] Zerhouni EA. MR imaging in chest disease: present

status and future applications. In: Green R, Muhm JR,

editors. Syllabus: a categorical course in diagnostic

radiology. Chest radiology. Oak Brook, IL: Radiologi-

cal Society of North America Inc; 1992. p. 25–41.

[30] Ho VB, Prince HR. Thoracic MR aortography: imag-

ing techniques and strategies. Radiographics 1998;18:

287–309.

[31] Leung DA, Debatin JF. Three-dimensional contrast-

enhanced magnetic resonance angiography of the tho-

racic vasculature. Eur Radiol 1997;7:981–9.

[32] Wernecke K, Vassallo P, Potte R, Lukener HG, Peters

PE. Mediastinal tumors: sensitivity of detection with

sonography compared with CT and radiography.

Radiology 1990;175:137–43.

[33] Gress FG, Savides TJ, Sandler A, Kesler K, Conces

D, Cummings O, et al. Endoscopic ultrasonography,

fine-needle aspiration biopsy guided by endoscopic

ultrasonography, and computed tomography in the

preoperative staging of non–small-cell lung cancer:


a comparative study. Ann Intern Med 1997;127:

604–12.

[34] Cioffi U, Bonavina L, De Simone M, Santambrogio

L, Pavoni G, Testori A, et al. Presentation and surgical

management of bronchogenic and esophageal dupli-

cation cysts in adults. Chest 1998;113:1492–6.

[35] Nguyen BD. Parathyroid imaging with Tc-99m sesta-

mibi planar and SPECT scintigraphy. Radiographics

1999;19:601–14.

[36] Santana L, Givica A, Camacho C. Thymoma. Radio-

graphics 2002;22:S95–102.

[37] Lastoria S, Vergara E, Palmieri G, Acampa W, Var-

rella P, Caraco C, et al. In vivo detection of malignant

thymic masses by indium-111-DTPA-D-Phe1-octreo-

tide scintigraphy. J Nucl Med 1998;39:634–9.

[38] Higuchi T, Taki J, Kinuya S, Yamada M, Kawasuji M,

Matsui O, et al. Thymic lesions in patients with my-

asthenia gravis: characterization with thallium-201

scintigraphy. Radiology 2001;221:201–6.

[39] Waxman AD, Eller D, Ashook G, Ramanna L, Brach-

man M, Heifetz L, et al. Comparison of gallium-

67-citrate and thallium-201 scintigraphy in peripheral

and intrathoracic lymphoma. J Nucl Med 1996;37:

46–50.

[40] Fletcher BD, Xiong X, Kauffman WM, Kaste SC,

Hudson MM. Hodgkin disease: use of T1-201 to

monitor mediastinal involvement after treatment.

Radiology 1998;209:471–5.

[41] Kubota K, Yamada S, Kondo T, Yamada K, Fukuda

H, Fujiwara T, et al. PET imaging of primary medi-

astinal tumors. Br J Cancer 1996;73:882–6.

[42] Sadato N, Tsuchida T, Nakaumra S, Waki A, Uematsu

H, Takahashi N, et al. Non-invasive estimation of the

net influx constant using the standardized uptake

value for quantification of FDG uptake of tumors.

Eur J Nucl Med 1998;25:559–64.

[43] Sasaki M, Kuwabara Y, Ichiya Y, Akashi Y, Yoshida T,

Nakagawa M, et al. Differential diagnosis of thymic

tumors using a combination of 11c-methionine PET

and FDG PET. J Nucl Med 1999;40:1595–601.

[44] Landreneau RJ, Dowling RD, Castillo WM, Ferson

PF. Thoracoscopic resection of an anterior mediasti-

nal tumor. Ann Thorac Surg 1992;54:142–4.

[45] Sugarbaker DJ. Thoracoscopy in the management of

anterior mediastinal masses. Ann Thorac Surg 1993;

56:653–6.

[46] Tecc PM, Fishman EK, Kuhlman JE. CT evaluation

of the anterior mediastinum: spectrum of disease.

Radiographics 1994;14(5):973–90.

[47] Rosado-de-Christenson ML, Galobardes J, Moran

CA. From the archives of the AFIP. Thymoma: ra-

diologic pathologic correlation. Radiograpgics 1992;

12:151–68.

[48] Morgenthaler TI, Brown LR, Clby TV. Thymoma.

Mayo Clin Proc 1993;68:110–1123.

[49] Santana L, Givica A, Camacho C. Best cases from the

AFIP thymoma. Radiograhpics 2002;22:S95–102.

[50] Fraser RS, Muller N, Colman N, Pare PD. Mediastinal

disease. In: Fraser RS, Muller N, Colman N, Pare PD,

editors. Fraser and Pare’s diagnosis of diseases of the

chest. Philadelphia: Saunders; 1999. p. 2875–974.

[51] Zerhouni EA, Scott WW, Baker RR. Invasive thymo-

mas: diagnosis and evaluation by computed tomogra-

phy. J Coput Assist Tomogr 1982;6:92.

[52] Scatarige JC, Fishman EK, Zerhouni EA. Transdia-

phragmatic extension of invasive thymoma. AJR

1985;144:31.

[53] Diehl L, Hopper K, Giguere J, Garnger E, Lesar M.

The pattern of intrathoracic Hodgkin’s disease as-

sessed by computed tomography. J Clin Oncol 1991;

9:438–43.

[54] Guermazi A, Brice P, de Kerviler E, Ferme C,

Hennequin C, Meignin V, et al. Extranodal Hodgkin

disease: spectrum of disease. Radiographics 2001;21:

161–79.

[55] Salonen O, Kivisaari L, Somer K. Differential diag-

nosis of anterior upper mediastinal expansions by

contrast-enhanced computed tomography. Comut

Radiol 1984;8:217–22.

[56] Filly R, Blank N, Castellino R. Radiographic distri-

bution of intrathoracic disease in previously untreated

patients with Hodgkin’s disease and non Hodgkin’s

lymphoma. Radiology 1976;120:277–81.

[57] Kowolafe F. Radiological patterns and significance

of thyroid calcifications. Clin Radiol 1981;32:571–5.

[58] Takashima S, Morimoto S, Ikezoe J. CT evaluation

of anaplastic thyroid carcinoma. AJR 1990;154:1079.

[59] Jennings A. Evaluation of substernal goiters using

computed tomography and MR imaging. Endocrinol

Metab Clin N Am 2001;30(2):401–14.

[60] Park HM, Traver RD, Siddiqui AR. Efficacy of thy-

roid scintigraphy in the diagnosis of intrathoracic goi-

ter. AJR 1987;148:527.

[61] Sanders LE, Rossi RL, Shahian DM, Willamson WA.

Medaistinal goiters. The need for an aggressive ap-

proach. Arch Surg 1992;127(5):609–13.

[62] Rosado-de-Christenson ML, Tempelton PA, Moran

CA. From the archives of the AFIP—mediastinal

germ cell tumors: radiologic pathologic correlation.

Radiogarphics 1992;12:1013–30.

[63] Nichols CR. Mediastinal germ cell tumors: clini-

cal features and biologic correlates. Chest 1991;99:

472–9.

[64] Wood DE. Mediastinal germ cell tumors. Semin

Thorac Cardiovasc Surg 2000;12(4):278–89.

[65] Gale AW, Jelihovsky T, Grant AF. Neurogenic tu-

mors of the mediastinum. Ann Thorac Surg 1974;

17:434.

[66] Kawashima A, Fishman EK, Kuhlman J, Nixon M.

CT of posterior mediastinal masses. Radiographics

1991;11:1045–67.

[67] Bourgouin PM, Shepard JO, Moore EH. Plexiform

neurofibromatosis of the mediastinum: CT appear-

ance. AJR 1992;159:279.

[68] Reed JC, Kagan-Hallett K, Feigin DS. Neural tumors

of the thorax: subject review from the AFIP. Radiol-

ogy 1978;126:9.

[69] Mendelson DS, Rose SJ, Efermidis SC, Kirschner


PA, Cohen BA. Bronchogenic cysts with high CT

numbers. AJR 1983;140:463–5.

[70] Azzie G, Beasley S. Diagnosis and treatment of fore-

gut duplications. Semin Pediatr Surg 2003;12(1):

46–54.

[71] Salo JA, Ala-Kulju K. Congenital esophageal cyst in

adults. Ann Thorac Surg 1987;44:135–8.

[72] Erasmus JJ, Mcadams HP, Donnelly LF, Spritzer CE.

MR imaging of mediastinal masses. Magn Reson

Imaging Clin N Am 2000;8(1):59–89.

[73] Feigin DS, Fenoglio JJ, Mcalsiter HA, Medawell JE.

Pericardial cysts: a radiographic pathologic correla-

tion and review. Radiology 1977;25:15–20.

[74] Aughenbaugh GL. Thoracic manifestations of neu-

rocutaneous diseases. Radiol Clin N Am 1984;22:

741–56.

[75] Kantoff P. Surgical and medical management of germ

cell tumors of the chest. Chest 1993;103:331S–3S.

[76] Costello P. Chest lymphoma. In: Greene R, Muhm

JR, editors. Syllabus: a categorical course in diag-

nostic radiology: Chest radiology. Oak Brook, IL:

Radiological Society of North America Inc; 1992.

p. 245–58.

[77] Gawrychowski J, Rokicki M, Gabriel A. Thymoma

the usefulness of some prognostic factors for diagno-

sis and surgical treatment. Eur J Surg Oncol 2000;26:

203–8.

[78] Dartavelle PG, Chapelier AR, Pastorino U, Corbi P,

Lenot B, Cerrina J, et al. Long-term follow-up after

prosthetic replacement of the superior vena cava com-

bined with resection of mediastinal–pulmonary ma-

lignant tumors. J Thorac Cardiovasc Surg 1991;102:

259–65.

[79] Naunheim KS. Video throacoscopy for masses of the

posterior mediastinum. Ann Thorac Surg 1993;56:

657–8.

[80] Demmy TL, Krasna MJ, Detterbeck FC, Kline GG,

Kohman LJ, DeCamp Jr MM, et al. Multicenter

VATS experience with mediastinal tumors. Ann

Thorac Surg 1998;66:187–92.

[81] Sobotta J. In: Ferner H, Staubesant J, editors. Atlas

der anatomie des menschen [atlas of human anato-

my], Vol. 2. Munich: Uran-Schwarzenberg; 1972.

p. 27–39 [in German].

[82] Lawler LP, Corl FM, Fishman EK. Multi-detector row

and volume-rendered CT of the normal and accessory

flow pathways of the thoracic systemic and pulmo-

nary veins. Radiographics 2002;22:S45–60.

[83] Kuzo RS, Ben-Ami TE, Yousefzadeh DK, Ramirez

JG. Internal mammary compartment: window to the

mediastinum. Radiology 1995;195:187–92.

[84] DeCamp Jr MM, Swanson SJ, Sugarbaker DJ. The

mediastinum. In: Baue AE, Geha AS, Hammond GL,

et al, editors. Glenn’s thoracic and cardiovascular sur-

gery. 6th edition. Stanford: Appleton & Lange; 1996.

p. 643–63.

[85] Toner GC, Panicek DM, Heelan RT, Geller NL, Lin

SY, Bajorin D, et al. Adjunctive surgery after chemo-

therapy for nonseminomatous germ cell tumors:

recommendations for patients selection. J Clin Oncol

1990;8:1683–94.

[86] Nichols CR, Saxman S, Williams SD, Loehrer PJ,

Miller ME, Wright C, et al. Primary mediastinal non-

seminomatous germ cell tumors: a modern single

institution experience. Cancer 1990;65:1641–6.

[87] Wright CD, Kesler KA, Nichols CR, Mahomed Y,

Einhorn LH, Miller ME, et al. Primary mediastinal

nonseminomatous germ cell tumors. Results of a

multimodality approach. J Thorac Cardiovasc Surg

1990;99:210–7.

[88] Mountain CF, Dresler CM. Regional lung cancer clas-

sification for lung cancer staging. Regional lymph

node classification in lung cancer staging. Chest

1997;111:1718–23.

[89] Mountain CF. Revisions in the international system

for staging lung cancer. Chest 1997;111:1710–7.

[90] American Joint Committee on Cancer. Lung. In:

Beahrs OH, Henson DE, Hutter RVP, et al, editors.

Manual for staging cancer. 4th edition. Philadelphia:

Lippincott; 1992. p. 115–21.

[91] American Thoracic Society. Medical section of the

American Lung Association. Clinical staging of

primary lung cancer. Am Rev Respir Dis 1983;127:

659–64.

[92] Cymbalista M, Waysberg A, Zacharias C, Ajavon Y,

Riquet M, Rebibo G, et al. Demonstration of the

1996 AJCC-UICC regional lymph node classifica-

tion for lung cancer staging. Radiographics 1999;

19:899–900.

[93] Genereux GP, Howie JL. Normal mediastinal lymph

node size and number: CT and anatomic study. AJR

Am J Roentgenol 1984;142(6):1095–100.

[94] Quint LE, Glazer GM, Orringer MB, Francis IR,

Bookstein FL. Mediastinal lymph node detection

and sizing at CT and autopsy. AJR Am J Roentgenol

1986;147(3):469–72.

[95] Glazer GM, Gross BH, Quint LE, Francis IR, Book-

stein FL, Orringer MB. Normal mediastinal lymph

nodes: number and size according to American

Thoracic Society mapping. AJR Am J Roentgenol

1985;144(2):261–5.

[96] Graeter TP, Hellwig D, Hoffmann K, Ukena D,

Kirsch CM, Schafers HJ. Mediastinal lymph node

staging in suspected lung cancer: comparison of

positron emission tomography with F-18-fluoro-

deoxyglucose and mediastinoscopy. Ann Thorac

Surg 2003;75(1):231–5 [discussion 235–6].

[97] von Haag DW, Follette DM, Roberts PF, Shelton D,

Segel LD, Taylor TM. Advantages of positron emis-

sion tomography over computed tomography in me-

diastinal staging of non-small cell lung cancer. J Surg

Res 2002;103(2):160–4.

[98] Kernstine KH, Mclaughlin KA, Menda Y, Rossi NP,

Kahn DJ, Bushnell DL, et al. Can FDG-PET reduce

the need for mediastinoscopy in potentially respect-

able nonsmall cell lung cancer? Ann Thorac Surg

2002;73(2):394–401 [discussion 401–2].

[99] Davis RD, Oldham HN, Sabiston DC. The mediasti-


num. In: Sabiston DC, Spencer FC, editors. Surgery

of the chest. 6th edition. Philadelphia: WB Saunders;

1995. p. 576–611.

[100] Simeone JF, Minagi H, Putman CE. Traumatic dis-

ruption of the thoracic aorta: significance of the

left apical extrapleural cap. Radiology 1975;117:

265–8.

[101] Carter AR, Sostman HD, Curtis AM, Swett HA. Tho-

racic alterations after cardiac surgery. AJR Am J

Roentgenol 1983;140:475–81.

[102] Davis HK. A statistical study of the thoracic duct in

man. Am J Anat 1915;17:211.

[103] Cohen RG, DeMeester TR, Lafontaine E. The pleura.

In: Sabiston DC, Spencer FC, editors. Surgery of the

chest. 6th edition. Philadelphia: WB Saunders; 1995.

p. 523–75.

[104] Wilson AG. Pleura and pleural disorders. In: Arm-

strong P, Wilson AG, Dee P, Hansell DM, editors.

Imaging of diseases of the chest. 2nd edition.

St. Louis (MO): Mosby; 1995. p. 641–716.


State-of-the-art screening for lung cancer (part 1):

the chest radiograph

Matthew Freedman, MD, MBA

Lombardi Cancer Center & Imaging Science and Information Systems (ISIS) Research Center,

Georgetown University Medical Center, Lombardi Building S150, Box 20057- 1465, 3800 Reservoir Road,

NW, Washington, DC 20057-1465, USA

A series of studies performed primarily in the ing of CR and DR CXRs. Other cancers can be

1970s have been interpreted to show that the chest

radiograph (CXR) is not an effective method for

reducing mortality from lung cancer. While a stage

shift was seen for the detected cancers, compared with

current or historic controls, mortality was not shown

to decrease. Since the 1970s, when these studies were

performed, there have been substantial improvements

in the technologies for CXRs and for the detection of

small nodules on these radiographs. New develop-

ments include computed radiography (CR), direct

digital radiography (DDR), image processing, energy

subtraction (ES), temporal subtraction of serial radio-

graphs (TS), and computer-aided detection (CAD). In

this article the term digital radiography (DR) will be

used to include CR and DDR. Examples of DR, ES,

and CAD are shown in Figs. 1 and 2.

Identified causes of missed lung cancer

Analyses of CXRs for missed cancers [1–4] show

that there are identifiable causes for missing lung

cancer on CXRs. Small lesions can lie in lightly or

darkly exposed portions of the CXR, a problem that

can be at least partially overcome by image process-


doi:10.1016/S1547-4127(04)00036-2

The author is Clinical Director and a stockholder in Deus

Technologies Limited Liability Corporation, Rockville, MD.


hidden partially or completely behind bony structures,

a problem that is decreased by energy subtraction

imaging, in which bone structures are made less

visible. Other cancers might be simply overlooked,

a problem that is partially corrected by CAD, a

computer program that alerts the radiologist to find-

ings that might represent lung nodules. These major

changes in available techniques for chest radiography

have not been incorporated into current lung cancer

screening trials. DR, ES, and CAD have been shown

to enhance the detection of lung nodules on CXRs

when compared with conventional chest radiographic

techniques. Most studies to date have reported on

synthetic nodules, nodules representing metastases to

the lungs, or mixtures of primary and metastatic

cancer. CR, DDR, ES, and CAD have received U.S.

Food and Drug Administration (FDA) approval. CR

has been used in lung cancer screening trials in Japan

and at some sites in the United States. The combina-

tions of DR with ES, DR with CAD, and DR with ES

and CAD have not been used in prospective studies

for lung cancer detection.

Several methods of statistical analysis have been

applied in the prior reported studies. Receiver oper-

ating characteristic (ROC) is a standard method for

analyzing the effect of new methods of imaging on

radiologists’ interpretation. The most commonly ap-

plied statistic for this is the area under the ROC curve

(Az). For practical purposes, the Az range is 0.5 to

1.0. The higher the value, the better the system. Az

can be interpreted as the sensitivity averaged at all

possible levels of specificity. When clinicians write of

improvements in sensitivity improvements in an ROC

s reserved.

Fig. 1. Case 1. (A) Small nodule in the left midlung. (B) Circle drawn by CAD program. (C) Nodule on ES image. (D) Nodule on

ES image outlined by CAD program. The CAD product used (RS2000D) is FDA approved for use with digitally acquired CXRs

but not for use with ES CXRs. (Courtesy of Deus Technologies LLC, Rockville, MD.)

M. Freedman / Thorac Surg Clin 14 (2004) 43–5244

study, this average improvement in sensitivity for all

possible specificity levels is being referenced. Some

studies have used free ROC (FROC) and alternate

FROC (AFROC), methods that allow multiple find-

ings (in this case one or more nodules) on each image.

Identification of the location of the finding is impor-

tant. There are various methods of reporting the

FROC and AFROC statistic. In each case the maxi-

mum value is 1.0, and higher values are better.

Digital radiography

CR and DR are different methods for acquiring

digital radiographs of the chest. DR with image

processing is FDA-approved, and with certain types

of image processing it has been shown to increase

radiologists’ ability to detect pulmonary nodules. It is

designed to correct for exposure differences among

subjects and within a single subject. Intersubject and

intrasubject image optical density is better controlled

with digital methods [5–8]. It is well recognized that

there is an optimal range of optical densities on chest

images for the detection of minimal findings such as

small lung nodules. The International Labor Organi-

zation has provided standards for conventional CXRs,

and similar settings have been recommended by the

American College of Radiology [9,10]. The reason

for these limits is that film is a nonlinear recorder of

exposure, and if the image is too light or too dark the

contrast of a small object might be so decreased that it

cannot be seen.

DR uses an x-ray sensing system that allows a

wider range of exposures to be recorded. This wider

range of exposures can then be adjusted by computer

vision techniques (usually referred to as image pro-

cessing) to produce images of optimal exposure (if

displayed on film) or luminance (if displayed on a

monitor). There are several types of DR that fall into

two categories: (1) CR (also referred to in the literature

as storage phosphor radiography [SR]) and (2) DDR.

There are analog and digital image processing

methods. Fundamental changes that can be produced

include changes in optical density, contrast, unsharp

masking to balance or correct optical density, spatial

frequency filtering, and mathematical methods to

enhance specific frequencies in images, resulting

in improved contrast for objects of specific sizes or

shapes, edge enhancement, and image noise reduc-

tion. These processing methods can be applied across

the entire image (global processing) or to specific por-

tions of an image (adaptive processing). The process-

Fig. 2. Case 2. (A) Small nodule in the left upper lung on DR chest image. (B) Circles drawn by CAD program. The CAD

program misses the nodule. (C) Nodule on ES image. (D) Nodule on ES image outlined by CAD program. On the ES image the

performance of the CAD program is improved. The CAD product used (RS2000D) is FDA approved for use with digitally

acquired CXRs but not for use with ES CXRs. (Courtesy of Deus Technologies LLC, Rockville, MD.)

M. Freedman / Thorac Surg Clin 14 (2004) 43–52 45

ing of specific portions of an image can be based on

anatomic regions identified by the computer or on re-

gions of specific optical density.

Initial work with analog processing

Sorenson used analog unsharp masking technique

and demonstrated that improving the contrast in the

retrocardiac region improved detection of nodules

metastatic to the lung in that region [11]. He used

FROC analysis with Bunch transform to ROC coor-

dinates. FROC mean true-positives in the retrocardiac

region were 0.500 for conventional and 0.700 for the

unsharp mask images, a 40% increase in true-posi-

tives. For all nodules the conventional mean was

0.625 and for unsharp mask images it was 0.677.

Initial work with image processing of digitized screen

film images

Sherrier had eight radiologists interpret digitized

CXRs containing 150 nodules [12]. They viewed

the images unprocessed, processed with histogram

equalization, and with adaptive filtration applied to

underexposed regions of the images. The highest per-

formance was seen with the adaptive filtration. Az

improved from 0.68 on the unprocessed images to

0.78 with adaptive filtration, a 15% improvement.

Using synthetic nodules superimposed on digi-

tized CXRs, Hoffmann applied optical density cor-

rection for under- and overexposed images [13]. In an

ROC study, Az for underexposed retrocardiac and

retrodiaphragmatic regions improved from 0.708 to

0.849 (P < 0.01), a 20% increase in sensitivity.

For overexposed lung periphery Az improved from

0.958 to 1.0 (P < 0.05). The authors concluded that

the improvement resulted from adjusting the contrast

at the location of the nodules to the area of steepest

contrast gradient in the image.

Initial work on synthetic nodules with digital image

acquisition and image processing

Initial work with digitally acquired images and

image processing used synthetic nodules and anthro-

pomorphic phantoms or digitally synthesized nodules

that were superimposed electronically on digital or

digitized CXRs. Prokop reported that in phantoms,

simulated nodules in CR chest images, nodule detec-

tion was better with large masks than with small

masks [14]. Schaefer-Prokop used phantom and simu-

lated nodules superimposed on lung tissue and me-


diastinum and found no differences between screen

film (SF), CR, and a selenium system (Thoravision,

Philips Medical Systems, Shelton, Connecticut) in the

phantoms but did find an improvement in the detec-

tion of micronodules and thin simulated lines with the

selenium system [15]. Leppert compared SF, asym-

metric SF, and Fuji AC-1, (Fuji Medical Systems,

Stamford, Connecticut) images using synthetic nod-

ules placed on human volunteers before obtaining the

CXRs [16]. This work showed that the asymmetric SF

system was best overall for pulmonary nodules and

that the Fuji system and the asymmetric SF combi-

nation were better than conventional SF images for

synthetic nodules superimposed on the mediastinum.

In 1994 (1 year before this report) the Fuji AC-1 was

in the process of being replaced by newer Fuji

systems (FCR 9000 and AC-3) that had better signal

to noise characteristics and new image processing

methods that produced images similar to those pro-

duced with the asymmetric SF system. It would be

expected that these technical advances would result in

improved nodule detection. Li used 5 and 10 mm

synthetic nodules of two compositions and shapes to

simulate dense and less dense nodules in an anthro-

pomorphic phantom [17]. This work showed that

image processing changes had no effect on the de-

tection of 10 mm nodules but improved detection of

5 mm nodules. They recommend processing with un-

sharp masking with midrange frequency suppression

and low frequency enhancing filters.

This important work with synthetic nodules super-

imposed on anthropomorphic phantoms or on digi-

tized or digital CXRs showed that image processing

applied to digital images enhanced the detection of

smaller nodules, particularly when they occurred in

regions of the image that had low contrast (ie, in

lightly or heavily exposed regions).

Studies in subjects who had actual lung nodules

confirmed by CT

Van Heesewijk obtained SF and selenium images

of the chest in patients who had several types of CT-

confirmed pulmonary and pleural diseases (eg, pul-

monary opacities, interstitial disease, mediastinal

disease, and pleural disease) [18]. The patients had

12 solitary nodules less than 2 cm. No differences

were found between SF and Thoravision, but the

authors might not have stressed the difference with

the use of small nodules. No range of nodule size

was given.

Muller reported a complex experiment in which

he compared two SF systems (200 and 400 speed),

digital CR images obtained at the same two expo-

sures, and six filtering masks for the digital images

[19]. Two hundred eighty-four CT-documented nod-

ules were evaluated with these methods and rated by

six observers. A nodule detectability score was used,

with 0 points for nondetection of a nodule and

20 points for a well-visualized nodule; mean nodule

detectability scores were then calculated. Overall, CR

processed with large kernel sizes was superior to CR

processed with small kernel sizes. CR was superior to

SF for nodules superimposed on the heart, dia-

phragm, and mediastinum and for smaller nodules.

Properly processed CR was always at least as good as

SF and sometimes better. For nodules in lung fields,

the mean nodule detectability score for SF was 12.52;

for CR it was 14.26. For nodules obscured by the

heart, diaphragm, or mediastinum, mean values were

8.63 for SF and 12.81 for CR. Nodules were also

studied by size. For nodules less than 10 mm, the

mean score was 11.9 for SF and 13.5 for CR. For

nodules that were 10 to 20 mm and greater than

20 mm, SF at 200 speed and CR performed the same.

For SF at 400 speed, CR at 400 speed was better

than SF (SF 9.6, CR 13.2 for nodules 10–20 mm).

An improvement in detectability score indicates that

the nodules were more conspicuous, but it does not

indicate that more nodules were detected.

Woodard compared chest images obtained with the

selenium Thoravision and system-optimized SF im-

ages obtained at 150 kilovoltage peak (KVp). Sele-

nium and SF images were obtained in 34 subjects

who had 78 lung nodules that were identified by

CT previously [20]. The nodule size range was 0.5 to

3.5 cm with an average size of 1.5 cm. Overall there

was no significant difference in nodule detection

(64% SF, 66% digital). For the subgroup of 19 nod-

ules that were less than 1 cm, radiologists detected

53% of the nodules on the selenium images and only

46% on the asymmetric SF images, a 15% increase in

nodules detected, which was not significant at the

small sample size (P = 0.69). For the 13 nodules that

were obscured on the postero-anterior (PA) view

because they were superimposed on the heart or

diaphragm, radiologists detected 46% on the sele-

nium system images and 36% on the asymmetric SF

system, a 28% increase in nodules detected, which was

not significant at the small sample size of 13 cases.

Krupinski reported a study in which six radiol-

ogists compared unprocessed images with five dif-

ferent image processing methods on 168 CR cases of

disease that were initially missed on interpreta-

tion [21]. There were 38 subtle nodules among the

168 cases. No difference was shown in the Az for

these cases or for individual subsets. Enhanced confi-

dence ratings were shown with the use of image


processing (P < 0.0001), but it is not known if en-

hanced confidence ratings enhanced disease detection.

Diagnostic decisions were not changed in this case.

Yang studied 18 patients who had CT-detected

nodules less than 20 mm. Regius CR (Konica, Tokyo,

Japan) was used for data acquisition with the original

image and two degrees of unsharp mask [22]. Overall,

no significant difference was shown. Radiologists

interpreting the standard image had an Az of 0.65;

with each of the filters the Az was 0.68. Sensitivity

increased with tumor size (P < 0.5). Tumors that had

alveolar lining growth were less visible than those

that were solid. While the authors showed no statis-

tically significant difference, eight of the ten measure-

ments showed nonsignificant improvement in Az for

the filtered images and one was equivalent.

Awai reported a comparison of CR to selenium-

based radiography in 31 patients who had CT-docu-

mented solitary noncalcified lung nodules that were

5 to 30 mm [23]. Nineteen nodules were smaller than

10 mm. Five radiologists interpreted the images,

comparing the CR and selenium images. Az was

0.64 with the CR system and 0.72 with the selenium

system (P < 0.05).

Overall, these reports support the assertion that

radiologists interpreting images of patients who have

lung nodules show improved performance in nodule

detection when digital images with appropriate image

processing are used when compared with nodule

detection with SF systems. The benefit of digital

imaging is greatest with the smallest nodules and

for those that are obscured by the heart or diaphragm.

Energy subtraction

ES radiography (also called dual energy radiogra-

phy in the literature) is a well-established research

imaging technique that has been implemented in

clinical systems over the past 7 years. Energy sub-

traction is an FDA-approved method for chest imag-

ing and has been shown in several clinical studies to

improve the detection of noncalcified nodules super-

imposed on bony structures.

There are now at least two FDA-approved sys-

tems. For clinical use it is still considered to be novel,

and no large United States trial has used it for lung

cancer screening. With ES, two images of the chest are

obtained simultaneously (in CR systems) or in rapid

succession (in DR systems). Because of the underly-

ing physical properties of atoms, the ratio of the x-ray

absorption of water and calcium varies at different

x-ray energies. Because this ratio varies at different

energies, image processing methods can be used to

take a low and a high energy radiographic image and

present them to the viewer as a bone emphasis and a

water emphasis image. The work of Stitik [1,2] and

Austin [3] show that many cancers missed on CXR

screening in clinical practice are missed because they

are small or because they are hidden behind ribs or

the clavicles. Limited clinical studies have confirmed

that ES increases the conspicuity of lung nodules.

Ho evaluated the effect of ES imaging on nodule

detection in images of anthropomorphic phantoms

that had simulated nodules using analog film as the

detector [24]. He compared two methods for ES. He

demonstrated that readers of the conventional SF

image had an Az of 0.876, those who read dual-

exposure ES had an Az 0.945 (an increase of 8% in

the mean sensitivity for all potential specificities),

and those who read single-exposure ES had an Az

of 0.929. Both ES methods were statistically signifi-

cantly superior to the conventional SF image.

Ishigaki performed an experiment in which 140

subjects had CR ES images taken. In two studies a

total of nine radiologists reviewed the images [25].

Sixty images showed primary lung cancer and 34 had

metastatic nodules. The others were normal or had

nonmalignant findings confirmed by biopsy or fol-

low-up. In the first experiment the sensitivity for

nodule detection was 0.47 for CR and 0.72 for ES

(P < 0.05). Specificity also improved from 0.74 to

0.86 (P < 0.05). For the second experiment nodule

sensitivity improved from 0.45 to 0.67 (P < 0.05) and

specificity improved from 0.80 to 0.92 (P < 0.05).

While improved sensitivity was shown for all nod-

ules, the effect was greater for those that projected

under a rib. For these nodules the sensitivity im-

proved from 0.17 to 0.73 in the first study and from

0.13 to 0.68 in the second study (P < 0.05). In the

second study the radiologists were asked to determine

if the nodule showed benign calcification. Sensitivity

for calcium detection improved from 0.29 to 0.96

on the ES images.

Kelcz reported on a study of 116 CT nodules in

50 patients [26]. Conventional SF and CR single-

exposure ES images were compared. Five observers

showed improvements in Az for nodule detection and

for characterization of the nodules as calcified. For

nodule detection the average Az was 0.597 for SF and

0.695 for ES (P < 0.005). For detecting nodule

calcification, the Az was 0.815 for SF and 0.958

for ES (P < 0.05).

Kido reported a comparison of standard CR with

an older and a newer method for detecting CT-con-

firmed lung nodules [27]. Forty nodules 5 to 20 mm in

size were seen in 22 subjects. Fourteen radiologists

interpreted the images. AFROC methods of data


analysis were used. For all nodules, the average Az for

the radiologists was 0.61 with CR, 0.72 with the new

ES method (P < 0.01), and 0.66 with the older ES

method (P < 0.01). For nodules that were super-

imposed on ribs, the average Az for the radiologists

was 0.55 on standard CR, 0.71 with the new ES

method (P < 0.05), and 0.63 with the older ES method

(P < 0.05). No improvement or decrement was shown

among the three methods for the nodules that were

not superimposed on the ribs (Az of 0.69, 0.69, and

0.72, respectively).

This evidence demonstrates that ES imaging pro-

vides improved lung nodule detection. Sensitivity is

improved for the detection of all nodules, but particu-

larly for those that are obscured by ribs. The detection

of calcification in nodules is also improved. Patients

who had calcified nodules were less likely to be

classified as having potentially malignant lesions.

Computer-aided detection

CAD of solitary pulmonary nodules is an FDA-

approved method for application to digitized SF and

digitally acquired CXRs. The FDA-approved ver-

sion was validated on primary non-small cell lung

cancer (NSCLC) 9 to 27 mm in size.

Computer analysis of radiographs for nodules has

been shown to be an effective method for increasing

radiologists’ detection of small lung nodules. With

CAD, the computer searches the image for findings

that could indicate the presence of a lung nodule. The

radiologist first views the image without CAD infor-

mation, then the CAD information is provided. If the

CAD system indicates sites that contain nodules that

the radiologist has overlooked and if the radiologist

accepts the information, the radiologist’s detection

rate for nodules improves. Many of the articles to

date cover technical details of the development of

these systems. In the fields of computer-aided diag-

nosis of lung cancer, two types of organizations

(academic and commercial) have been conducting

investigations for more than a decade. A group of

outstanding researchers led by Dr. Kunio Doi has

done substantial work [28,29] in pulmonary nodule

detection and CAD in other diseases at the University

of Chicago [30–32]. Doi’s group has presented two

effective lung nodule detection methods: (1) evalua-

tion of circularity with incremental thresholding, and

(2) evaluation of circularity using a morphological

open operation. Their results indicate that these meth-

ods achieve a true-positive detection rate of ap-

proximately 70% with an average of three to four

false-positive detections per chest image [33,34].

The clinical studies are reviewed below. One of the

benefits of CAD is that the marking of the lesion

appears to enhance its conspicuity. Krupinski reported

that the placement of a solid circular boundary around

a lung nodule enhanced its detection significantly

[35]. When no circle was provided the Az was

0.523; with a dashed circle it was 0.690 and with a

solid circle it was 0.800.

Kobayashi assessed the improvement in radiolo-

gists’ performance using a case sample of 60 normal

patients and 60 patients who had a single pulmonary

nodule [36]. The mean pulmonary nodule size was

14 mm. Thirty-seven percent of the nodules were

confirmed primary lung cancer and 42% had solitary

metastases. Sixteen radiologists participated and

ROC analysis was used. The Az without CAD was

0.894, and with CAD it was 0.940 (P < 0.001), a 5%

improvement. MacMahon reported the results of a

study in which 20 CXRs containing a single pulmo-

nary nodule and 20 normal CXRs were interpreted

with and without CAD [37]. On hundred forty-six

observers participated. Chest radiologists’ Az im-

proved from 0.825 to 0.889, an 8% improvement;

other radiologists’ Az improved from 0.810 to 0.876,

an 8% improvement. Radiology residents’ Az im-

proved from 0.774 to 0.855, and nonradiologists’ Az

improved from 0.697 to 0.808. All of these improve-

ments were highly statistically significant.

Recently, Deus Technologies LLC received FDA

approval for a CAD system for enhanced lung nodule

detection on CXRs (FDA PMA000041). The system

was validated on cases of NSCLC from the Johns

Hopkins Early Lung Cancer Trial, one of the major

lung cancer screening trials from the 1970s. Using the

RS-2000 (Deus Technologies, Rockville, Maryland)

radiologists showed that at their operating points (the

point of sensitivity and specificity at which they

decided on the need for diagnostic CT averaged for

the 15 study radiologists) there was an average in-

crease in cancer detection of 11% for primary NSCLC

9 to 27 mm, 21% for NSCLC 9 to 14.5 mm, and

38% for cancers that had been missed prospectively

[38–41]. As defined for this study, a cancer was a

missed cancer if the two radiologists at Johns Hopkins

who subspecialized in chest radiography both missed

the cancer on the film obtained approximately 1 year

before actual detection. In ROC numbers the radiolo-

gists, on average, went from Az 0.829 to 0.865 for all

cancers, from Az 0.798 to 0.848 for cancers 9 to

14 mm, and from Az 0.702 to 0.744 for the missed

cancers. The improvement for all cancers and for the

cancers 9 to 14 mm in size was statistically significant

(P < 0.05). For missed cancers the results were not

statistically significant for the sample size. The im-


provement demonstrated in this retrospective study

suggests that a prospective screening trial using

CAD would result in improved (ie, earlier) detection

of NSCLC. A new version of the CAD system, the

RS2000D, which uses digitally acquired CXRs, has

also received FDA approval.

Fig. 3. TS of CXRs. (A) Prior CXR. (B) CXR obtained 1 year later.

apex can be identified as a white focal nodule. This method is exp

S-C Ben Lo, Hui Zhao, and Matthew Freedman, Imaging Scienc

Computer-aided detection on energy subtraction

images

In 2002 Kido provided two reports of CAD on ES

images. In the first report [42] 12 patients had CT-

confirmed nodules, eight of whom had bronchogenic

(C) Subtraction image on which the nodule in the right lung

erimental and has not received FDA approval. (Courtesy of

e Center, Georgetown University, Washington DC.)


cancer (presumed to have one nodule each, although

this is not stated in the article). Four patients had

metastases to the lung [42]. In only one of the

12 patients (one patient who had several nodules)

did the four radiologists, on average, detect more true-

positive nodules than the CAD. In seven patients they

were equivalent, and in four patients the CAD system

detected nodules that one or more radiologists missed.

Overall, the average was 1.60 [standard deviation

(SD) 1.03] for the radiologists and 1.83 (SD 1.34)

for the CAD system.

In 2002 Kido also reported a study of 25 images of

chest phantoms with nylon nodules of three degrees of

thickness producing different degrees of contrast on

the image [43]. Twenty-five nodules were present.

Each lung was viewed separately so there were 25

lungs with nodules and 25 lungs without nodules.

Digital chest x-radiograph (D-CXR) and digital chest

x-radiograph with energy subtraction (D-CXR-ES)

images were studied. For the D-CXR without ES, the

CAD system detected 14 of 25 nodules; the 12 ra-

diologists, on average, detected 13.3 of the 25 nod-

ules (SD 2.6; not significant). If one looked at the

maximum potential improvement that radiologists

could have had if they tested with CAD, the 12 ra-

diologists would have detected 17.7 (SD 1.4) of the

25 nodules, a potential improvement of 4.4 nodules.

On the ES images, CAD detected 23 nodules com-

pared with 14 without ES (P < 0.005). The average

detections of the 12 radiologists on the D-CXR-ES

images were 21.2 (SD 2.1) compared with 13.3

(SD 2.6) without ES (P < 0.005). If the radiologists

had used CAD, the maximum potential nodule detec-

tions would have increased from 21.2, on average, to

24.4 (SD 0.5). These articles show the potential

contribution of ES when added to D-CXR and CAD

when added to D-CXR with ES.

A series of studies have been reported in which

evidence is presented that (1) DR CXR with appro-

priate image processing is superior to standard CXR

for nodule detection, (2) DR CXR with ES is superior

to DR CXR without ES, (3) CXR and DR CXR with

CAD is superior to CXR and DR CXR, and (4)

preliminary data suggest that CAD might provide

further potential improvements when added to DR

CXR with ES.

Temporal subtraction

TS is an important proposed method for enhancing

the detection of lung nodules on serial CXRs. In this

method two studies of the chest are taken at different

times. The older image is then subtracted from the

newer image, resulting in increased conspicuity of any

change that might have occurred over time. The

method is complex because of differences in position-

ing of the chest on the two radiographs and because

of differences in the degree of lung inflation. An

example of TS is shown in Fig. 3.

Kakeda reported on a study in which a temporal

series of D-CXR images containing 20 solitary lung

nodules (10 benign and 10 lung cancers, all less than

30 mm) were assessed by four radiologists and four

radiology residents [44]. The radiologists showed

improved Az from 0.873 to 0.969 with the use of

temporally subtracted images. The residents’ Az im-

proved from 0.825 to 0.958. The combined improve-

ment was significant (P = 0.027). Similar results for

lung infiltrates were reported by Tsubamoto [45].

In a study of sequential radiographs in 30 patients

who had CT-confirmed solitary pulmonary nodules,

Johkoh [46] showed that temporal subtraction pairs

resulted in improved performance among residents

but not attending radiologists. The residents’ perfor-

mance improved from Az 0.855 to 0.907 (P < 0.05).

For the attending radiologists, the Az was 0.964

without and 0.907 with the TS images.

Preliminary data suggest that TS imaging might,

in the future, provide additional benefit in the detec-

tion of lung nodules, but additional investigation and

improvements in this method are needed.

Summary

The chest radiographic methods used in prior

studies of lung cancer screening and in current pro-

spective clinical trials of lung cancer screening do not

incorporate, as part of their prospective design, the

newer methods available for the detection of lung

nodules. DR, image processing, ES, and CAD have

been shown to enhance lung nodule detection. TS is a

promising method but with less supporting data cur-

rently available. These techniques, alone or in com-

bination, do not equal the nodule detection capability

of lung CT, but they are likely to benefit patients

having CXRs for other clinically indicated purposes

and when the detection of a nodule is incidental to the

clinical indication for the radiographic study.

References

[1] Stitik F, Tockman M. Radiographic screening in the

early detection of lung cancer. Radiol Clin N Am

1978;216:347–66.

[2] Stitik F, Tockman M, Khouri N. Chest radiology. In:


Miller AB, editor. Screening for cancer. New York:

Academic Press; 1985. p. 163–91.

[3] Austin JHM, Romney BM, Goldsmith LS. Missed

bronchogenic carcinoma: radiographic findings in

27 patients with a potentially resectable lesion evident

in retrospect. Radiology 1992;182:115–22.

[4] Quekel L, Kessels A, Goei R, van Engelshoven J. Miss

rate of lung cancer on the chest radiograph in clinical

practice. Chest 1999;115:720–4.

[5] Freedman MT, Lo SCB, Nelson M, Horii SC, Mun SK.

Tests of the AGFA Phosphor Plate Imaging System in

chest imaging. In: Heshiki A, Mun SK, editors. Pro-

ceedings of IMAC 91, Kyoto, Japan, April 11–13,

1991. Los Alamitos, CA: IEEE Computer Society

Press; p. 425–9.

[6] Freedman M, Leftridge C, Nelson M, Lo S-CB, Mun

SK. Tests of the AGFA ADC Phosphor Plate Imaging

System in Pediatrics. In: Brody WR, Johnson GS, edi-

tors. SCAR 92, Symposium for Computer Assisted

Radiology, Baltimore MD (June 14–17, 1992). Carls-

bad, CA: Symposia Foundation; p. 716–7.

[7] Freedman MT, Lo SCB, Nelson M, Horii SC, Mun SK.

Tests of the AGFA ADC Phosphor Plate Imaging

System. In: Brody WR, Johnson GS, editors. Proceed-

ings of SCAR 92: Symposium for Computer Assisted

Radiology. Baltimore, MD, June 14–17, 1992. Carls-

bad, CA: Symposia Foundation; p. 718–9.

[8] Freedman MT, Lo SCB, Nelson M, Pe E, Mun SK.

Comparative tests of two storage phosphor plate imag-

ing systems: the AGFA ADC and the Fuji AC-1+.

SCAR 92. In: Brody WR, Johnson GS, editors. Pro-

ceedings of Symposium for Computer Assisted Radi-

ology. Baltimore, MD, June 14–17, 1992. Carlsbad,

CA: Symposia Foundation; p. 203–9.

[9] International Labor Organization. Guidelines for the

use of the international classification of radiographs

of pneumoconiosis. Geneva (Switzerland): Interna-

tional Labor Organization; 1992.

[10] American College of Radiology. Standards 2002.

Reston (VA): American College of Radiology; 2002.

[11] Sorenson JA, Mitchell CR, Armstrong JDN, Mann H,

Bragg DG, Mann FA, et al. Effects of improved con-

trast on lung-nodule detection. A clinical ROC study.

Invest Radiol 1987;22(10):772–80.

[12] Sherrier RH, Chiles C, Wilkinson WE, Johnson GA,

Ravin CE. Effects of image processing on nodule de-

tection rates in digitized chest radiographs: ROC

study of observer performance. Radiology 1988;

166(2):447–50.

[13] Hoffmann KR, MacMahon H, Doi K, Metz CE, Yao L,

Abe K. Evaluation of an enhanced digital film-du-

plication system by receiver operating characteristic

analysis. Invest Radiol 1993;28(12):1134–8.

[14] Prokop M, Schaefer CM, Oestmann JW, Galanski M.

Improved parameters for unsharp mask filtering of digi-

tal chest radiograph. Radiology 1993;187(2):521–6.

[15] Schaefer-Prokop CM, Prokop M, Schmidt A, Neitzel

U, Galanski M. Selenium radiography versus storage

phosphor and conventional radiography in the detection

of simulated chest lesions. Radiology 1996;201(1):

45–50.

[16] Leppert AG, Prokop M, Schaefer-Prokop CM, Galan-

ski M. Detection of simulated chest lesions: compari-

son of a conventional screen-film combination, an

asymmetric screen-film system, and storage phosphor

radiography. Radiology 1995;195(1):259–63.

[17] Li F, Sone S, Kiyono K. Lung nodule conspicuity

using unsharp mask filters with storage-phosphor-

based computed radiography. Acta Radiol 1997;

38(1):99–103.

[18] van Heesewijk HP, van der Graaf Y, de Valois JC, Vos

JA, Feldberg MA. Chest imaging with a selenium de-

tector versus conventional film radiography: a CT-con-

trolled study. Radiology 1996;200(3):687–90.

[19] Muller RD, Von Koschitzki T, Hirche H, John V, Her-

ing K, Gocke C, et al. Frequency-filtered image post-

processing of digital luminescence radiographs in

pulmonary nodule imaging. Clin Radiol 1996;51(8):

577–86.

[20] Woodard PK, Slone RM, Sagel SS, Fleishman MJ,

Gutierez FR, Reiker GG, et al. Detection of CT-proved

pulmonary nodules: comparison of selenium-based

digital and conventional screen-film chest radiographs.

Radiology 1998;209:705–9.

[21] Krupinski EA, Evanoff M, Ovitt T, Standen JR, Chu

TX, Johnson J. Influence of image processing on

chest radiograph interpretation and decision changes.

Acad Radiol 1998;5(2):79–85.

[22] Yang ZG, Sone S, Li F, Takashima S, Maruyama Y,

Hasegawa M, et al. Detection of small peripheral lung

cancer by digital chest radiography. Performance of

unprocessed versus unsharp mask-processed images.

Acta Radiol 1999;40(5):505–9.

[23] Awai K, Komi M, Hori S. Selenium-based digital ra-

diography versus high-resolution storage phosphor ra-

diography in the detection of solitary pulmonary

nodules without calcification: receiver operating char-

acteristic curve analysis. AJR Am J Roentgenol 2001;

177(5):1141–4.

[24] Ho JT, Kruger RA. Comparison of dual-energy and

conventional chest radiography for nodule detection.

Invest Radiol 1989;24(11):861–8.

[25] Ishigaki T, Sakuma S, Ikeda M. One-shot dual-energy

subtraction chest imaging with computed radiography:

clinical evaluation of film images. Radiology 1988;

168(1):67–72.

[26] Kelcz F, Zink FE, Peppler WW, Kruger DG, Ergun

DL, Mistretta CA. Conventional chest radiography vs

dual-energy computed radiography in the detection and

characterization of pulmonary nodules. AJR Am J

Roentgenol 1994;162(2):271–8.

[27] Kido S, Ikezoe J, Naito H, Arisawa J, Tamura S,

Kozuka T, et al. Clinical evaluation of pulmonary

nodules with single-exposure dual-energy subtraction

chest radiography with an iterative noise-reduction

algorithm. Radiography 1995;194:407–12.

[28] Katsuragawa S, Doi K, MacMahon H. Image feature

analysis and computer-aided diagnosis in digital radi-


ography: classification of normal and abnormal lungs

with interstitial disease in chest images. Med Phys

1989;16(1):38–44.

[29] Doi K, Giger ML, MacMahon H, Hoffmann KR, Nish-

ikawa RM, Schmidt RA, et al. Computer-aided di-

agnosis: development of automated schemes for

quantitative analysis of radiographic images. Semin

Ultrasound CT MR 1992;13(2):140–52.

[30] Giger M, Doi K, MacMahon H. Image feature analysis

and computer-aided diagnosis in digital radiography. 3.

Automated detection of nodules in peripheral lung

field. Med Phy 1988;15:158–66.

[31] Giger M, Ahn N, Doi K, MacMahon H, Metz C. Com-

puterized detection of pulmonary nodules in digital

chest images: use of morphological filters in reducing

false-positive detections. Med Phys 1990;17:861–5.

[32] MacMahon H, Sanada S, Doi K, Giger M, Xu X, Yin

F, et al. Direct comparison of conventional and com-

puted radiography with a dual-image recording tech-

nique. Radiographics 1991;11:259–68.

[33] Iinuma T, Tateno Y, Matsumoto T, Yamamoto S, Mat-

sumoto M. Basic idea of lung cancer screening CT

(LSCT) and its preliminary evaluation. Jap J Radio

Med 1992;52:182–90.

[34] Xu X, Doi K, Kobayashi T, MacMahon H, Giger M.

Development of an improved CAD scheme for auto-

mated detection of lung nodules in digital chest

images. Med Phys 1997;24:1395–403.

[35] Krupinski EA, Nodine CF, Kundel HL. A perceptually

based method for enhancing pulmonary nodule recog-

nition. Invest Radiol 1993;28(4):289–94.

[36] Kobayashi T, Xu XW, MacMahon H, Metz CE, Doi K.

Effect of a computer-aided diagnosis scheme on radi-

ologists’ performance in detection of lung nodules on

radiographs. Radiology 1996;199(3):843–8.

[37] MacMahon H, Engelmann R, Behlen FM, Hoffmann

KR, Ishida T, Roe C, et al. Computer-aided diagnosis

of pulmonary nodules: results of a large-scale observer

test. Radiology 1999;213:723–6.

[38] Freedman MT, Lo SCB, Lure F, Xu XW, Lin J, Zhao

H, et al. Computer aided detection of lung cancer on

chest radiographs. Algorithm performance vs radi-

ologists’ performance by size of cancer. Proceedings

of SPIE medical imaging: visualization, display, and

image-guided procedures 2001;4319:150–9.

[39] Freedman M, Benedict Lo S, Osicka T, Lure F, Xu X,

Lin J, et al. Computer aided detection of lung cancer

on chest radiographs: effect of machine CAD false

positive locations on radiologists’ behavior. Proceed-

ings of the Society of Photo-Optical Instrumentation

Engineers 2002;4684:1311–9.

[40] Freedman M, Osicka T, Benedict Lo S, Lure F, Xu X,

Lin J, et al. Methods for identifying changes in radiol-

ogists’ behavioral operating point of sensitivity–speci-

ficity trade-offs within an ROC Study of the use of

computer aided detection of lung cancer. Proceedings

of the Society of Photo-Optical Instrumentation Engi-

neers 2002;4324:184–94.

[41] Freedman MT. Digital chest radiography. In: Boiselle

P, White C, editors. New techniques in thoracic imag-

ing. New York: Marcel Dekker, Inc; 2002. p. 315–48.

[42] Kido S, Nakamura H, Ito W, Shimura K, Kato H.

Computerized detection of pulmonary nodules by

single-exposure dual-energy computed radiography of

the chest (part 1). Eur J Radiol 2002;44(3):198–204.

[43] Kido S, Kuriyama K, Kuroda C, Nakamura H, Ito W,

Shimura K, et al. Detection of simulated pulmonary

nodules by single-exposure dual-energy computed

radiography of the chest: effect of a computer-aided

diagnosis system (part 2). Eur J Radiol 2002;44(3):

205–9.

[44] Kakeda S, Nakamura K, Kamada K, Watanabe H,

Nakata H, Katsuragawa S, et al. Improved detection

of lung nodules by using a temporal subtraction tech-

nique. Radiology 2002;224(1):145–51.

[45] Tsubamoto M, Johkoh T, Kozuka T, Tomiyama N,

Hamada S, Honda O, et al. Temporal subtraction for

the detection of hazy pulmonary opacities on chest

radiography. AJR Am J Roentgenol 2002;179(2):

467–71.

[46] Johkoh T, Kozuka T, Tomiyama N, Hamada S, Honda

O, Mihara N, et al. Temporal subtraction for detection

of solitary pulmonary nodules on chest radiographs:

evaluation of a commercially available computer-aided

diagnosis system. Radiology 2002;223(3):806–11.


State-of-the-art screening for lung cancer: (part 2):

CT scanning

David Yankelevitz, MD*, Claudia I. Henschke, PhD, MD

Department of Radiology, Weill Medical College, Cornell University, 525 East 68th Street,


Starting in 1993 two groups independently began On annual repeat 2.5% of patients had an abnormal-

exploring the use of CT scans as the initial test in

screening protocols for lung cancer. These groups

were the National Cancer Center in Tokyo, Japan [1]

and the Early Lung Cancer Action Project (ELCAP)

at Weill Medical College of Cornell University [2].

The Japanese group had an active clinical screen-

ing program using chest radiography. They intro-

duced CT for an additional cost of $350 in 1993.

The patients’ ages ranged between 38 and 83 and

most had a 20 pack-year smoking history, although

smoking history was not required. The results of

1369 baseline screens and 2088 semiannual repeat

screens were reported by Kaneko et al [1]. Among

the 3457 screens, positive findings were present in

20% (701) of patients, and 15 of these subjects had

malignant results. The overall yield of CT was 0.43%

(15/3457) compared with 0.12% (4/3457) with

chest radiography.

At Weill Medical College, a prospective study

called ELCAP was started in 1992. Starting in 1993

1000 high-risk subjects aged 60 and older who had at

least a 10 pack-year smoking history were enrolled

for baseline and annual repeat screening. The median

age at enrollment was 67 and the median pack-years

smoked was 47. Baseline results were published in

1999 [2], and annual repeat results were published in

2001 [3]. At baseline, 23% (233) of patients had an

abnormality, of which 27 were found to be malignant,

which yielded an overall rate of 2.7% (27/1000), of

which 0.7% (7/1000) were seen on chest radiography.


doi:10.1016/S1547-4127(04)00035-0



(D. Yankelevitz).

ity, of which seven were found to be malignant,

yielding an overall rate of 0.6% (7/1184). Chest

radiography was not performed for the repeat studies.

Starting in 1996 a third study, also in Japan, com-

pared CT with chest radiography [4]. Using a mobile

CT unit, Sone et al [4] performed baseline screening

on 5483 individuals from the general population in

Japan aged 40 to 74 years; 3967 also had miniature

flourophotography. They found that 5% (279/5483)

had a positive result, and the malignancy rate was

0.48% (19/5483) on CT compared with 0.3% on chest

radiography. A follow-up report on the results of

their annual repeat screening using only CT showed

that 3.8% (309/8303) had a positive result and that

the malignancy rate was 0.41% (34/8303) [5]. An

additional follow-up through 2001 showed that for

baseline screening the malignancy rate was 0.51%

(40/7847) and on annual repeat screening it was 0.4%

(40/10,045) [6].

These studies demonstrated that CT screening for

lung cancer was superior to the chest radiograph in

detecting lung cancer. They also showed that a pos-

itive result on baseline screening was more common

than on repeat screening. In the ELCAP study, which

had the highest median age and smoking history, the

lung cancer rate was also the highest, confirming that

age and smoking history are key risk indicators of

lung cancer.

Since these early studies several other groups have

reported their results. None of them used chest radi-

ography. These studies were done at University of

Muenster in Muenster, Germany [7], Japan [8], and,

more recently, in the Mayo Clinic in Rochester Min-

nesota [9] and Milan, Italy [10]. The Mayo group

reported that screening with sputum cytology and CT

s reserved.

D. Yankelevitz, C.I. Henschke / Thorac Surg Clin 14 (2004) 53–5954

scanning found that on baseline screening 76% of

non–small-cell lung cancer was stage I. On annual

repeat screening this proportion was 55%. The Milan

group found 55% that of their baseline cancers were

stage I, whereas 100% of the annual repeat cancers

were stage I. These studies showed a consistent pat-

tern of finding a high proportion of early-stage cancers

on baseline screening and annual repeat screening.

Screening regimen

To study CT screening for lung cancer mean-

ingfully, a regimen needs to be described, including

specification of the type of scanner, the scanning

protocol, the definition of a positive result, and the

Fig. 1. Starting in 1993 CT screening was performed using a 10 mm

were obtained while covering the same volume in a single breath-h

(A) 10 mm slice thicknesses. (B) 5 mm slice thickness. (C) 2.5 m

workup of the positive result (both for baseline and

annual repeat). It is important to think of screening as

the pursuit of early diagnosis with a view toward early

treatment. In this way screening is not merely the

application of a single test. In the case of CT screening

for lung cancer, CT is merely the initial test. Without

considering the regimen of subsequent diagnostic tests

that follow, the results of the initial test are not

meaningful. Entirely different results will be found

following the initial test when different algorithms

for workup are used. Thus, there is a need to specify

the entire regimen. Each feature of the regimen is

important. The ELCAP protocol for the diagnostic

workup is updated to incorporate increasing knowl-

edge about screening and technologies advances

as they occur [11].

slice thickness. With advances in technology, thinner slices

old. Effect of slice thickness on visibility of a stable nodule:

m slice thickness. (D) 1.25 mm slice thickness.

D. Yankelevitz, C.I. Henschke / Thorac Surg Clin 14 (2004) 53–59 55

As for the initial CT, technological advances have

come at a rapid pace. In 1993, when the initial CT

screening studies began, images were obtained with a

10 mm slice thickness. At that time only single slice

scanners existed. To scan the entire chest in a single

breath hold, a 10 mm thickness was necessary. With

the advent of multislice scanners, this practice has

changed dramatically. With the constraint of scan-

ning the entire chest in a single breath, slice thickness

has decreased progressively from 10 mm to 5 mm to

2.5 mm to current standards of 1.25 mm (Fig. 1). The

newest generation of scanners even allow for proto-

cols using a 0.675 mm slice thickness. The basic

principle is that use of thinner slice images allows for

the detection of smaller nodules. This trend toward

Fig. 2. (A) A 10 mm nodule was identified on CT. (B) Three-dimen

CT data. (C) One month later. The nodule has grown during thi

determined by use of an image processing technique, correspondin

thinner slices will continue; there are now prototype

units that can produce images with 0.1 mm resolu-

tion. Nevertheless, along with the improved resolu-

tion comes the additional burden of having the

radiologist interpret many more images per scan.

Currently, more than 300 images are obtained for

each study when using these thinner section images.

While this increase has become a concern for the

radiologist, it has opened the door for computer-

assisted techniques, which perform much better with

higher-resolution images. Computer-assisted tech-

niques include techniques that are used to measure

the volume of pulmonary nodules so that growth

rates can be determined and techniques that allow for

automated nodule detection (Fig. 2) [12].

sional volumetric reconstruction was performed on the initial

s short interval. An increase in volume was 22% has been

g to a malignant growth rate.


Diagnostic distribution

On the first baseline screen, it is expected that

more cancers will be found than on any single repeat

screening cycle. It is assumed that the malignancies

reported on baseline screening were those found in

nodules detected on the baseline screening, even

though the actual diagnosis of malignancy might be

made as much as several years later.

The diagnostic distribution is summarized by the

relevant prognostic categories, defined as determi-

nants of the long-term outcome of these cases. For

lung cancer, these categories include stage, size, and

histology. Given this definition it can be expected

that the diagnostic distribution will be differ from

baseline screening compared with annual repeat

screening. An additional important consideration is

that with each round of annual repeat screening the

diagnostic distribution should remain relatively con-

stant [13], which is of great importance because with a

large enough population of patients being studied, the

distribution of cancers can be determined with two

rounds of screening. It also means that the results of

each round of annual repeat screening can be pooled

to learn the actual diagnostic distribution with greater

confidence. An additional consideration in regard to

the diagnostic distribution relates to the relative risk

of the population being studied. It can be expected

that for a given risk there will be a different overall

frequency of detected cancer; however, it can also be

expected that even though the overall frequency of

cancers might be different, the proportion of the

various subtypes should remain the same. In other

words, in a high-risk population compared with a low-

risk population, many more cancers might be detected

with a particular regimen of screening; however, the

proportion of stage I cancers will remain constant.

Ultimately, this method allows for pooling of data

from various sources, providing that the same regimen

of screening is followed, which has been an underly-

ing principle in the ongoing International Early Lung

Cancer Action Project (I-ELCAP) study, in which

approximately 35 institutions around the world have

agreed to follow the same protocol and pooling of

data. To date, approximately 25,000 cases are in the

pooled database [6].

False-positive diagnosis

The issue of false-positive diagnosis has been

brought up as a concern in regard to CT screening

for lung cancer. One group reported that up to 99%

of the nodules identified represent false-positive

findings [14], but this is a misleading number. This

group defined screening cases with nodules less than

4 mm in diameter as being negative results, yet a

high proportion of their false-positives included these

4 mm or smaller nodules [15]. Thus, they interpreted

scans as being negative when they contained false-

positive findings.

A rational definition of a positive result of screen-

ing is to provide for sufficient sensitivity to not

miss too many of the cancers while not having too

many false-positive findings. For instance, one would

not consider the result of a stool guiac study to be

positive if a single red blood cell was found; rather,

there is some threshold at which the study is consid-

ered to be positive. Similarly, in regard to CT screen-

ing there is some small size threshold where every

person being scanned will have at least one nodule. Its

size can be less than 1 mm, but it would not make

sense to call each of these nodules false-positives

because they occur so frequently as to be noncontribu-

tory in terms of discriminating between subjects who

have or who do not have the target illness. In regard to

screening, the definition of a false-positive result

becomes a bit more complex because screening for

lung cancer can be thought of as a year-to-year

process. It is envisioned that when a person enrolls

in a screening program they will come back for an

annual repeat study. Under these circumstances, a

person who has a small nodule, say less than 5 mm

on baseline screening, might simply be told to return

for annual repeat screening without any intervening

workup. The reason for this might be that in nodules

this small it might be so difficult to make a diagnosis

in less than 1 year because growth determinations or

other diagnostic tests are so inaccurate that it is

impractical to pursue each of these nodules. Never-

theless, the nodules cannot be ignored. The patients

are merely told to return for their routinely scheduled

annual repeat study; they can, thus, be thought of as

perhaps representing a different type of positive

finding but not a false-positive finding in the sense

of leading to additional workup.

The ELCAP group reviewed their results recently

and found that on baseline screening it was not

practical to obtain a diagnosis of cancer in less than

1 year for nodules smaller than 5 mm [16]. ELCAP’s

current definition of a positive result of baseline

screening does not include subjects whose largest

nodule is less than 5 mm [11]. Using this definition

they have been able to reduce the number of subjects

who have positive baseline screening results to be-

low 15%.

It is also important to distinguish between the find-

ings of baseline screening and annual repeat screen-


ing. While a nodule less than 5 mm on baseline

screening does not prompt additional workup in the

new ELCAP protocol, a 5 mm nodule found on annual

repeat screening that was not present on the prior

study does prompt immediate further workup. In this

situation one now has the additional information

that because the nodule was not present previously,

it now is, and therefore it is growing. The growth of a

nodule from being invisible to visible in the range of

3 to 5 mm in 1 year is suggestive of a malignant

growth rate and needs to be thought of differently than

a nodule that is of a similar size and only found on

baseline screening, in which no additional information

regarding its growth rate is available.

Curability of early lung cancer

While CT screening for lung cancer provides for

early diagnosis (especially when compared with wait-

ing for symptom prompting), the ultimate goal is to

allow for early treatment. Thus, the issue in regard

to the benefit of lung cancer screening relates to

answering two component questions. First, how fre-

quently does a particular regimen of screening lead to

early diagnosis? Second, how curable are those can-

cers? When these component issues have been under-

stood, the benefit in terms of reducing death from lung

cancer can be derived.

Critical to the concept of curability of lung cancer

is learning what proportion of lung cancers are gen-

Fig. 3. Subsolid nodules. (A) Nonsolid nodule contains no solid elem

parenchyma, and vessels can still be identified. (B) Part-solid nod

uine cancers and not overdiagnosed ones. A genuine

cancer is one that would lead to death in the absence

of intervention. It is not reasonable to think in terms

of curing a cancer if in the absence of treatment it

was not fatal.

For traditional radiography, a great deal has been

learned in regard to issues of genuineness and cur-

ability of screen detected lung cancer. Flehinger et al

studied this issue directly in the chest radiography

screening studies performed as part of the National

Cancer Institute Cooperative Early Lung Cancer De-

tection Project (Mayo Lung Project, New York Lung

Project at Memorial Sloan-Kettering Cancer Center,

and the Johns Hopkins Lung Project) [17]. When

focusing on 45 untreated cases that were stage I, they

showed that 5-year fatality rates in the absence and

presence of treatment were 90%, which implies that

at least 90% are genuine with at most 10% being

indolent. Among the cases of stage I cancer that

underwent treatment (resection), the corresponding

fatality rate was only 30%. Therefore, the overall

curability rate for the genuine lung cancers was

(90�30)/90, or 67%. Because cases of unresected

malignancies might be understaged, Flehinger extend-

ed her evaluation to include patients who had sus-

pected mediastinal metastases and found qualitatively

similar results.

In further support of these data, a recently pub-

lished review of screen-detected cancers in stage I in

two of these studies (the Mayo Lung Project and the

New York Project) found that they fit the profile of

genuine (ie, fatal if not treated) cancers. The study

ents. The lesion does not completely obscure the underlying

ule, which contains solid and nonsolid elements.


evaluated the growth rates of these tumors and found

that they were typical of those found in usual clinical

practice [18].

In the context of radiographic screening (even at

4-month intervals), stage I diagnosis was achieved in

only 29% of the cases [19]. With CT-based screening

the proportion has increased markedly. It is now

approximately 80%. This shift to a higher frequency

of early diagnosis should translate to improved

curability because the tumors diagnosed under CT

screening are smaller than those found with radiog-

raphy. An important remaining question to be an-

swered in the context of CT screening is the

proportion of stage I lung cancers that are not

genuine. This is a more serious concern in regard

to CT screening because a new class of lung cancers

has now been described called subsolid nodules [20].

Subsolid nodules include the nonsolid and the part-

solid nodules (Fig. 3). While they are seen primarily

on baseline screening, there is strong evidence to

suggest that some of the nonsolid ones are indolent,

as manifested in their relatively slow growth rates,

their appearance nearly restricted to baseline screen-

ing, and the near 100% absence of fatality when they

are actually resected [21,22]. Future studies on this

topic should help clarfiy the overall proportion of

these lesions that are indolent, thus allowing for un-

derstanding the overall curability the CT screen-

detected lung cancer.

Summary

There have been dramatic improvements in tech-

nology in the past decade. In conjunction there have

also been advances in our clinical knowledge that

have led to changes in the screening regimen. These

changes are expected to continue in the future as CT

scanners continue to improve and knowledge about

screening accumulates, and computer-assisted techni-

ques are expected to play an ever more important role.

This dynamic process will lead to continued improve-

ments in the diagnostic distribution of lung cancers

detected under CT screening.

References

[1] Kaneko M, Eguchi K, Ohmatsu H, Kakinuma R,

Naruke T, Suemasu K, et al. Peripheral lung cancer:

screening and detection with low-dose spiral CT ver-

sus radiography. Radiology 1996;201(3):798–802.

[2] Henschke CI, McCauley DI, Yankelevitz DF, Naidich

DP, McGuinness G, Miettinen OS, et al. Early Lung

Cancer Action Project: overall design and findings from

baseline screening. Lancet 1999;354(9173):99–105.

[3] Henschke CI, Naidich DP, Yankelevitz DF, McGuin-

ness G, McCauley DI, Smith JP, et al. Early lung

cancer action project: initial findings on repeat screen-

ings. Cancer 2001;92(1):153–9.

[4] Sone S, Takashima S, Li F, Yang Z, Honda T, Maru-

yama Y, et al. Mass screening for lung cancer with

mobile spiral computed tomography scanner. Lancet

1998;351(9111):1242–5.

[5] Sone S, Li F, Yang ZG, Honda T, Maruyama Y, Taka-

shima S, et al. Results of three-year mass screening

programme for lung cancer using mobile low-dose

spiral computed tomography scanner. Br J Cancer

2001;84(1):25–32.

[6] International Collaboration to Screen for Lung Can-

cer. Proceedings of the International Conferences on

Screening for Lung Cancer. New York, NY. Available

at: http://ICScreen.med.cornell.edu. Accessed Decem-

ber 1, 2003.

[7] Diederich S, Wormanns D, Semik M, Thomas M, Len-

zen H, Roos N, et al. Screening for early lung cancer

with low-dose spiral CT: prevalence in 817 asympto-

matic smokers. Radiology 2002;222(3):773–81.

[8] Nawa T, Nakagawa T, Kusano S, Kawasaki Y, Suga-

wara Y, Nakata H. Lung cancer screening using low-

dose spiral CT: results of baseline and 1-year follow-up

studies. Chest 2002;122(1):15–20.

[9] Swensen SJ, Jett JR, Hartman TE, Midthun DE, Sloan

JA, Sykes AM, et al. Lung cancer screening with

CT: Mayo Clinic experience. Radiology 2003;226(3):

756–61.

[10] Pastorino U, Bellomi M, Landoni C, De Fiori E, Ar-

naldi P, Picchio M, et al. Early lung-cancer detection

with spiral CT and positron emission tomography in

heavy smokers: 2-year results. Lancet 2003;362(9384):

593–7.

[11] Protocol I-ELCAP. Available at: http://icscreen.med.

cornell.edu/ielcap.pdf. Accessed December 1, 2003.

[12] Yankelevitz DF, Reeves AP, Kostis WJ, Zhao B,

Henschke CI. Small pulmonary nodules: volumetri-

cally determined growth rates based on CT evaluation.

Radiology 2000;217(1):251–6.

[13] Henschke CI, Miettinen OS, Yankelevitz DF, Libby

DM, Smith JP. Radiographic screening for cancer. Pro-

posed paradigm for requisite research. Clin Imaging

1994;18(1):16–20.

[14] Swensen SJ. CT screening for lung cancer. AJR Am J

Roentgenol 2002;179(4):833–6.

[15] Contemporary Screening for the Detection of

Lung Cancer. Available at: http://acrin.org/pdf_file2.

html?file=protocol_docs/A6654protocol.pdf. p. 29.

Accessed December 1, 2003.

[16] Henschke CI, Yankelevitz DF, Miettinen OS. CT

screening for lung cancer: suspiciousness of nodules

at baseline according to size. Radiology2003; [in press].

[17] Flehinger BJ, Kimmel M, Melamed MR. The effect

of surgical treatment on survival from early lung can-

http:\\ICScreen.med.cornell.edu

http:\\icscreen.med.cornell.edu\ielcap.pdf

http:\\acrin.org\pdf_file2.html?file=protocol_docs\A6654protocol.pdf


cer. Implications for screening. Chest 1992;101(4):

1013–8.

[18] Yankelevitz DF, Kostis WJ, Henschke CI, Heelan RT,

Libby DM, Pasmantier MW, et al. Overdiagnosis in

chest radiographic screening for lung carcinoma: fre-

quency. Cancer 2003;97(5):1271–5.

[19] Fontana RS, Sanderson DR, Woolner LB, Taylor WF,

Miller WE, Muhm JR. Lung cancer screening: the

Mayo program. J Occup Med 1986;28(8):746–50.

[20] Henschke CI, Yankelevitz DF, Mirtcheva R, McGuin-

ness G, McCauley D, Miettinen OS, ELCAP Group.

CT screening for lung cancer: frequency and signifi-

cance of part-solid and nonsolid nodules. AJR Am J

Roentgenol 2002;178(5):1053–7.

[21] Aoki T, Nakata H, Watanabe H, Nakamura K, Kasai T,

Hashimoto H, et al. Evolution of peripheral lung

adenocarcinomas: CT findings correlated with histol-

ogy and tumor doubling time. AJR Am J Roentgenol

2000;174(3):763–8.

[22] Noguchi M, Morikawa A, Kawasaki M, Matsuno Y,

Yamada T, Hirohashi S, et al. Small adenocarcinoma

of the lung. Histologic characteristics and prognosis.

Cancer 1995;75(12):2844–52.


Imaging for esophageal tumors

Robert J. Korst, MD, Nasser K. Altorki, MD*

Division of Thoracic Surgery, Department of Cardiothoracic Surgery, Weill Medical College, Cornell University,

525 E. 58th Street, New York, NY 10021, USA

Carcinoma of the esophagus comprises the vast process, with specific imaging modalities being use-

majority of malignant esophageal tumors and repre-

sents the seventh most common malignancy world-

wide, with its incidence reaching endemic proportions

in specific geographic locations in Asia and Africa

[1]. Although esophageal cancer is presently respon-

sible for only approximately 13,000 deaths annually

in the United States [2], the incidence of adenocarci-

noma of the esophagus is rising faster than any other

malignant tumor in the United States [3]. Because

the majority of patients present with advanced dis-

ease, only roughly 12% of patients diagnosed with

this tumor will survive more than 5 years after diag-

nosis [2].

The treatment of carcinoma of the esophagus is

stage-dependent (Table 1). While patients who have

widely metastatic disease are not treated with curative

intent (ie, only palliative chemotherapy or supportive

care), most clinicians would agree that patients who

have early (superficial, node-negative) cancers should

undergo surgical resection for cure; however, the

ideal treatment of locally advanced (transmural,

node-positive) disease remains controversial, with

some clinicians advocating surgical resection alone,

others supporting preoperative neoadjuvant therapy

followed by surgery, and still others backing defini-

tive chemoradiation without surgery.

Given the stage dependency of therapeutic options

for patients who have esophageal cancer, it is essen-

tial to determine the extent of disease accurately be-

fore formulating the treatment plan. Imaging plays an

integral role in guiding the clinician in this staging


doi:10.1016/S1547-4127(04)00038-6



(N.K. Altorki).

ful for the evaluation of distant disease, locoregional

disease, or both. Certain imaging techniques have

proven to be useful in guiding biopsy procedures,

such as fine needle aspiration (FNA) of suspicious

lesions; however, the accuracy of some of these tech-

niques seems to rely, at least in part, upon the ex-

perience of the operator [4]. Finally, individual

imaging algorithms and the preference of one modal-

ity versus another varies with device availability,

individual experience, and geographic location.

Imaging of distant metastatic disease

In the United States, approximately 20% to 30%

of patients who have carcinoma of the esophagus

have distant metastatic disease at the time of presen-

tation [2,5]. The most common visceral metastatic

sites include, in decreasing order of prevalence, liver,

lung, bone, and adrenal glands [5,6]. As a result,

imaging for patients who have esophageal cancer

should evaluate these sites. The brain is an uncom-

mon site of metastases from esophageal cancer,

occurring in less than 2% of patients who have

metastatic disease [5,6]. Further, it is uncommon for

patients who have carcinoma of the esophagus to

present with solitary metastatic lesions; most possess

multiple numbers of metastases, albeit usually in a

single organ [5,6]. In these cases of metastatic disease

in a pattern consistent with esophageal cancer, often-

times histologic confirmation by means of biopsies is

not necessary; however, a second, corroborating

imaging study should be performed. In the uncom-

mon situation in which a patient presents with a

single metastatic lesion radiographically, or a pattern

inconsistent with that typically seen with esophageal

s reserved.

Table 1

Staging scheme for carcinoma of the esophagus

Stage Characteristics

Primary tumor (T)

TX Tumor cannot be assessed

T0 No evidence of tumor

Tis Carcinoma in situ/high grade dysplasia

T1 Confined to mucosa or submucosa, not

into muscularis proporia

T2 Invades into muscularis propria

T3 Invades through muscularis propria but not

into adjacent organs

T4 Invades adjacent structures/organs

Nodal status (N)

NX Regional nodes cannot be assessed

N0 No regional nodal metastases

N1 Regional nodal metastases

Distant metastases (M)

MX Distant metastases cannot be assessed

M1a Metastatic cervical nodes/upper thoracic

esophageal tumor

Metastatic celiac nodes/lower thoracic

esophageal tumor

M1b Any tumor location with visceral/bony

metastases

Any tumor location with nodal metastases

beyond N1 or M1a

Stage groupings

0 TisN0M0

I T1N0M0

IIA T2-3N0M0

IIB T1-2N1M0

III T3N1M0

T4 Any N

IVA Any T Any N M1a

IVB Any T Any N M1b

Fig. 1. Intravenous contrast-enhanced CT image of the liver

of a patient who had carcinoma of the esophagus. The

encircled region demonstrates a large, hypodense, irregu-

larly bordered lesion representing the typical appearance

of metastasis.

R.J. Korst, N.K. Altorki / Thorac Surg Clin 14 (2004) 61–6962

cancer, confirmatory biopsy should be performed

more routinely to ensure that the patient does not

have potentially curable (resectable) disease or an-

other distinct disease process. Nearly all potentially

metastatic foci can technically be assessed cytologi-

cally by means of image-guided FNA [4].

Because carcinoma of the esophagus is still an

uncommon disease relative to other tumor types in

the United States, little published data exist regarding

accuracy of many imaging modalities (eg, radio-

nuclide bone scan) exclusively for the detection of

distant metastases in patients who have esophageal

cancer; however, multiple published reports con-

cerning the accuracy of these imaging techniques

exist for carcinomatous tumors in general. Intrave-

nous contrast-enhanced CT remains the workhorse

for imaging patients who have carcinoma of the

esophagus to rule out distant metastatic lesions be-

cause it allows assessment of the three most common

sites of distant metastases. Scans should be obtained

from the base of the neck (thoracic inlet) through the

liver and adrenal glands in the upper abdomen.

Metastatic deposits in the liver usually appear as

hypodense, ill-defined lesions on contrast-enhanced

CT scans (Fig. 1) [7,8]. As with any liver imaging

modality, the sensitivity of the CT scan for detecting

metastatic liver disease depends on the size of the

lesion [7,8]. While the vast majority of lesions larger

than 1 cm are detected using CT scan, the sensitivity

drops precipitously for metastatic deposits less than

1 cm in diameter or if the scan is performed without

intravenous contrast. Similarly, if the lesions are of

adequate size ( > 1 cm), CT is useful for distinguish-

ing metastases from benign entities, most notably

cysts and hemangiomas, with the former possessing

the density of fluid and the latter demonstrating pe-

ripheral enhancement with delayed washout of intra-

venous contrast [7,8].

Other imaging modalities that are useful in assess-

ing the status of the liver include ultrasound (US) and

MRI. Although transabdominal US is inexpensive

and distinguishes between cystic and solid liver le-

sions accurately, its sensitivity in detecting metastatic

liver deposits in general is clearly inferior to that of

CT [7,8]. Laparoscopic US is potentially more sen-

sitive than the transcutaneous approach [9], but it is

an invasive procedure that tends to be especially user-

dependent, with published data suggesting only lim-

ited benefit for patients who have cancer of the

esophageal body [9,10]. MRI can be beneficial when

CT demonstrates liver lesions and further characteri-

R.J. Korst, N.K. Altorki / Thorac Surg Clin 14 (2004) 61–69 63

zation is needed. Gadolinium contrast agents might

enhance the sensitivity of MRI, which is an effective

modality for distinguishing metastases from benign

liver lesions, including cysts and hemangiomas [7,8].

Pulmonary metastases are also seen in patients

who have esophageal carcinoma. Suspicious pulmo-

nary nodules are usually round, smooth-bordered,

and noncalcified on CT scan. Given the high preva-

lence of incidental, benign pulmonary nodules seen in

smokers over the age of 60 [11], any suspicious lung

lesion should be biopsied using FNA or a thoraco-

scopic approach. Further, given the role of smoking

in carcinogenesis of the lung and esophagus and the

concept of field cancerization, primary lung cancer

also needs to be ruled out in these situations, particu-

larly if the pulmonary lesion is solitary [12].

Because bone is a common site for metastases

from carcinoma of the esophagus, routine radionu-

clide bone scanning can be performed in these pa-

tients. In general, in patients who have cancer, a scan

showing multiple areas of uptake strongly suggests

metastases; however, only 50% of solitary foci rep-

resent metastases, even in patients who have a history

of cancer [13]. Tracer accumulation can occur at any

skeletal site with an elevated rate of bone turnover.

As a result, corroborative studies are required in the

majority of cases of a positive bone scan, which

include MRI (which is especially useful for evalua-

tion of the spine), plain radiographs, and even a CT

scan. The radiographic evaluation of adrenal lesions

has been the subject of many reported studies involv-

ing the use of CT and MRI. While primary malignant

lesions of the adrenal glands are uncommon, the

prevalence of benign adrenal adenomas in the general

population is significant and might approach 7% by

age 70 [14]. Because of the high intracellular lipid

content in adenomas, thin-cut (3 mm), noncontrast

CT and MRI have been reported to possess specificity

rivaling that of FNA with cytologic examination for

distinguishing metastases from adenomas [15].

Positron emission tomography ([18F]2-flouro2-

deoxyglucose positron emission tomography [FDG]-

PET) is a new imaging modality that is gaining

popularity in staging patients who have many types

of malignant disease. Based on the finding that

malignant cells possess higher rates of glucose uptake

compared with normal cells, several small studies

have demonstrated that FDG-PET has been shown to

radiographically detect occult distant metastatic dis-

ease in approximately 20% of patients who have

esophageal cancer [16,17]. Given these encouraging

preliminary findings, this concept is presently being

evaluated in a large, multicenter, prospective study.

Drawbacks of FDG-PET are related to its lack of

sensitivity for detecting small (<1 cm) metastatic

lesions and its relative lack of anatomic detail. The

latter problem can be at least partially addressed by

the advent of newer PET/CT fusion scanners, in

which a composite image is generated incorporating

FDG-PET and CT images. It is important to note that

until larger, confirmatory studies are performed ex-

amining the utility of FDG-PET for detection of

metastatic disease, FDG-PET findings in patients

who have esophageal cancer should be confirmed

with a second imaging technique or a biopsy depend-

ing on the individual clinical scenario. This guideline

is especially true in the assessment of potentially

metastatic pulmonary lesions because although the

FDG-PET scan is frequently positive in pulmonary

metastases, a number of benign pulmonary lesions

(mainly inflammatory) can also be glucose avid [18].

Imaging of the primary tumor

Carcinoma of the esophagus originates in the

epithelial lining and spreads into and through the

wall of the esophagus and throughout the draining

lymphatics to lymph nodes. Esophageal carcinoma

readily disseminates hematogenously to distant sites.

Published data have confirmed that the presence of

lymph node metastases is a powerful predictor of

prognosis in these patients and is a marker for sys-

temic spread of the disease [19,20]. Similarly, the

depth of penetration of the primary tumor into the

esophageal wall predicts the presence or absence of

lymph node metastases, with approximately 85% of

T3 tumors being associated with lymphatic spread

[1]. Accurate imaging of the primary tumor in pa-

tients who have esophageal carcinoma is therefore

important, not only for determining resectability in

patients who have locally advanced disease but also

predicting prognosis in patients who have disease that

appears to be limited to the esophagus.

In past decades, primary tumors of the esophagus

were imaged using barium esophagography. Not only

could the location and longitudinal extent of the tumor

be determined, estimations of resectability could be

made based on the esophagram. In this regard, Aki-

yama and colleagues found that 74% of transmural

tumors caused distortion of the normal axis of the

esophagus [21]. This distortion is caused by tethering

of the esophagus in the region of the tumor.

The two most commonly used contemporary im-

aging procedures for assessing the primary tumor are

CT and endoscopic ultrasound (EUS). Given its lack

of anatomic detail, FDG-PET is unable to provide any

definition of the esophageal wall or periesophageal

Fig. 2. CT/PET fusion study depicting esophageal carci-

noma in the distal third of the esophagus. The lesion is

encircled in each panel. (A) Noncontrast CT image. (B)

FDG-PET image. (C) CT/PET fusion image.


tissues, making it of limited utility in assessing the

primary tumor (Fig. 2B). Similarly, CT does not

provide adequate resolution in distinguishing the

layers of the esophageal wall; however, information

can be gained concerning neighboring organ involve-

ment, or, more specifically, the lack thereof (Fig. 2A).

Preservation of fat planes surrounding the tumor has

been proposed and is supported as radiographic ex-

clusion of a T4 tumor [22,23]. Conversely, loss of fat

planes might indicate neighboring organ involvement.

When the tumor compresses the membranous left

main bronchus or trachea, bronchoscopy should be

performed to definitively establish airway invasion.

As with the airway, invasion of the descending tho-

racic aorta is difficult to predict using CT. Some

published evidence suggests that the greater the cir-

cumference of the aorta abutted by the tumor, the

more likely the tumor will be unresectable [24]. In

summary, although T4 tumors can be excluded reli-

ably by the preservation of peritumoral fat planes, the

definitive establishment of neighboring organ inva-

sion is difficult to predict with CT and on most

occasions operative exploration is required.

EUS is an imaging modality that is gaining popu-

larity in the preoperative assessment of patients who

have esophageal tumors. The great strength of EUS

lies in its ability to visualize the esophageal wall in

greater detail than any other imaging modality. The

esophageal wall is seen as four distinct layers using

EUS: mucosa, muscularis mucosa, submucosa, and

muscularis propria. A fifth layer corresponding to

periesophageal fat is also readily discernable using

EUS. A standard EUS examination usually involves

evaluation of the tumor with 7.5 MHz and 12 MHz

probes and is considered to be the most accurate

means by which to estimate tumor invasion. In this

regard, large review series place the accuracy of EUS

in determining the depth of invasion of esophageal

carcinoma at approximately 85% [25,26], with the

identification of T2 tumors being the least accurate

(Figs. 3, 4) [25,26].

Drawbacks of EUS include the relatively steep

learning curve [27] and the inability to pass the trans-

ducer completely through the tumor in up to 50% of

cases [25]. Newer probes are being developed con-

tinuously to address this problem, some being thin

enough to pass through the instrument channel of the

endoscope [28]. Other recent developments in EUS

technology include probes that allow for helical

scanning with subsequent three-dimensional recon-

struction of EUS images [29] and the use of high-

frequency transducers. These latter probes tend to be

useful in imaging superficial tumors of the esophagus

by providing more detail, and they can differentiate

Fig. 3. Elderly patient who had T1 adenocarcinoma of

the distal esophagus. (A) Endoscopic appearance. (B) EUS

image demonstrating lack of penetration into the muscularis

propria (MP).

Abbreviation: T, tumor.


between T1A and T1B successfully [30]. This dis-

tinction might be of importance in locations in which

esophageal cancer screening is performed and lesions

are detected at earlier stages more routinely.

Similar to EUS, preliminary data suggest that

investigational techniques such as endoluminal MRI

might be able to visualize the layers of the esophageal

wall accurately [31]. Whether or not this technique

will earn a role in the future of imaging for carcinoma

of the esophagus requires further investigation.

Fig. 4. EUS image of T2 squamous cell carcinoma of the

esophagus. Note the tumor (T) is indistinguishable from the

muscularis propria (MP).

Imaging of lymphatic metastases

It is generally agreed that the presence of lymph

node metastases (N1 disease) associated with resect-

able carcinoma of the esophagus is the strongest

known predictor of recurrence and mortality follow-

ing definitive therapy for this disease [19,20]. As

with some other types of malignancies, the degree of

lymph node involvement might also be of prognostic

value, with published studies demonstrating that pa-

tients who have less than three to five metastatic

nodes survive appreciably longer than those who

have more than 10 involved nodes following a

potentially curative resection [19,32]. Given this

information, the determination of lymph node status

before definitive therapy might be of importance be-

cause patients who have more advanced locoregional

disease could be enrolled in trials of novel or multi-

modal therapies.

Historically, clinicians have attempted to image

lymph node metastases using multiple modalities

with limited success. The accuracy of the CT scan

for staging this aspect of the disease has been well

described in multiple literature reports. Because the

detection of metastatic nodes using CT depends pri-

marily on size criteria, its sensitivity and specificity in

detecting metastatic disease in the lymph nodes varies

with the definition of an abnormally enlarged node.

Sensitivity is enhanced if smaller size criteria are

used, but specificity is sacrificed. Conversely, large

lymph nodes on CT are more likely to be metastatic;

however, many metastatic nodes are only mini-

mally—if at all—enlarged, which hampers sensitivi-

ty. Using the common size criterion of 1 cm to define

an enlarged node, most studies report that the sensi-

tivity of CT is poor (30–60%) [17,33] and does not

appear to be enhanced with helical scanning [34]. In

contrast, specificity tends to be somewhat better, but

still suboptimal (60–80%). In summary, if CT sug-

Fig. 5. CT/PET fusion study depicting malignant perieso-

phageal lymph node. The arrow indicates the malignant

node in each panel. (A) Noncontrast CT image. (B) FDG-

PET image. (C) CT/PET fusion image.


gests the presence of metastatic lymph nodes, tissue

confirmation should be obtained if the treatment plan

will be affected.

In recent years the role of FDG-PET has been

evaluated for the detection of lymph node metastases

in patients who have esophageal cancer. FDG-PET is

a physiologic examination that has poor anatomic

definition, which severely affects its ability to predict

N1 disease accurately in the peritumoral location

[33,35]. In this regard, most esophageal tumors are

intensely FDG avid, further inhibiting the resolution

of the study and making it easy to miss metastatic

nodes that are adjacent to the primary tumor. In

contrast, when metastatic lymph nodes are located

more remotely, the accuracy of FDG-PET increases

[33,35]. The differentiation of FDG-avid peritumoral

nodes from the primary tumor might be aided by the

development of CT/PET fusion scanners (Fig. 5), in

which the anatomic detail of CT is combined with the

physiologic nature of FDG-PET, but this scenario

remains to be seen.

Given these spatial limitations of FDG-PET, it is

not surprising that the sensitivity of this modality in

detecting peritumoral metastatic lymph nodes is poor

(20–50%) in most contemporary series [17,33,35];

however, sensitivities as high as 90% have been

reported in the detection of metastatic nodes in distant

locations such as the abdomen and the neck [33,35].

In distinct contrast, the specificity of FDG-PET in

lymph node evaluation tends to be high, exceeding

90% in many series [17,33,35].

US, transcutaneous and endoscopic, is used fre-

quently to stage the N descriptor in patients who have

esophageal carcinoma. US relies not only on size

criteria to determine metastasis but also on the in-

ternal echo characteristics of individual nodes. Well-

demarcated, larger, hypoechoic nodes with scattered

large, internal echoes are more likely to represent

metastases (Fig. 6) [36,37]. The use of transcutaneous

US to image cervical and supraclavicular lymph

nodes has become routine in some regions, especially

in Asia, where reported accuracy is approximately

70% to 80% [36,38]; however, other reports have not

been able to confirm these results [35].

The accuracy of EUS in detecting metastatic

lymph nodes in patients who have esophageal carci-

noma has also been investigated and reported in

many series (Fig. 6). Wide variations of sensitivity

and specificity have been reported in these series,

ranging from 40% to 100% [39]. Similar to the ability

to detect T stage, the ability of EUS to stage the N

descriptor effectively is highly user-dependent. Cen-

ters that perform large numbers of procedures report

higher accuracy rates [37], which have not been

reproducible in other studies [35], which leads one

to question the accuracy of EUS in routine practice

settings. To address this issue, EUS has been com-

bined with FNA of suspicious lymph nodes. The

addition of FNA to EUS has been shown by some

investigators to markedly enhance the specificity of

EUS alone, especially in the region of the celiac axis

[40,41]. Whether or not these excellent results can be

Fig. 6. EUS image of a typical metastatic lymph node in a

patient who had carcinoma of the esophagus. The metastatic

node is seen as a large, hypoechoic structure in the peri-

esophageal location (arrow).

Abbreviations: Ao, descending thoracic aorta; T, tumor.


achieved and reproduced routinely remains to be

determined and will influence the applicability of

this technique in routine practice situations.

Assessment of response to therapy

Given the relatively poor prognosis of patients

who have carcinoma of the esophagus who undergo

surgical resection alone for locally advanced disease,

preoperative (induction) chemotherapy or chemora-

diotherapy are being investigated as means to obtain

higher cure rates. Data from these clinical trials have

suggested that patients who are complete pathologic

responders to induction therapy seem to reap the most

benefit from multimodal treatment protocols [42,43],

so it might be advantageous to determine which

patients would benefit most from surgery before

resection. The accuracy of imaging modalities in this

capacity is now being investigated, with some pre-

liminary results published in recent literature.

Jones and colleagues compared the response to

preoperative chemoradiation as determined by repeat

CT scanning to pathological response rates pros-

pectively in 50 patients [44]. Using standard ra-

diographic response criteria, CT was found to be

ineffective for determining pathologic tumor response

or disease stage in this setting. Similarly, EUS was

unable to stage patients accurately after induction

therapy [45,46]; however, some evidence suggested

that measurements of tumor size using EUS might

correlate with response to chemoradiotherapy [47].

Some recent data suggest that a reduction in FDG

uptake by esophageal tumors after induction chemo-

radiotherapy might correlate with pathologic response

to therapy [48] and even improved survival in these

patients [49]. The use of imaging studies to assess the

response to therapy in patients who have esophageal

carcinoma is an emerging field, and it requires ex-

tensive investigation in future studies.

Summary

Carcinoma of the esophagus must be staged

accurately before a treatment plan is initiated, and

imaging studies play a major role in this process.

Imaging for esophageal carcinoma involves evalua-

tion of the locoregional extent of the tumor and

distant metastatic disease. A CT scan of the chest

and upper abdomen provides the most comprehensive

information about esophageal carcinoma; however,

accurate assessment of the depth of primary tumor

invasion and lymph node status remains limited, even

with newer generation scanners. Endoscopic US is a

user-dependent modality that has emerged as a highly

accurate technique in experienced hands to evaluate

the depth of penetration of esophageal tumors, but its

ability to detect metastatic lymph nodes is less im-

pressive, leading some investigators to perform con-

firmatory needle aspiration of suspicious nodes.

FDG-PET is a physiologic examination that is the

subject of intense investigation in patients who have

esophageal carcinoma. Preliminary studies have sug-

gested that FDG-PET can detect otherwise radio-

graphically occult distant metastatic disease in these

patients, and changes in FDG uptake might correlate

with the response to therapy. These findings need to

be confirmed in larger studies. More sophisticated

technology continues to be developed for imaging

carcinoma of the esophagus, which will more than

likely affect staging algorithms in the future.

References

[1] Korst RJ, Altorki NK. Esophageal cancer. In: Winches-

ter DP, Daly JM, Jones RS, Murphy GP, editors. Can-

cer surgery for the general surgeon. Philadelphia:

Lippincott-Raven; 1999. p. 155–72.

[2] Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer

statistics, 2000. CA 2000;50:7–33.

[3] Devesa SS, Blot WJ, Fraumeni Jr JF. Changing pat-


terns in the incidence of esophageal and gastric carci-

noma in the United States. Cancer 1998;83:2049–53.

[4] Long BW. Image-guided percutaneous needle biopsy:

an overview. Radiol Technol 2000;71(4):335–59.

[5] Quint LE, Hepburn LM, Francis IR, Whyte RI,

Orringer MB. Incidence and distribution of distant

metastases from newly diagnosed esophageal carci-

noma. Cancer 1995;76(7):1120–5.

[6] Christie NA, Rice TW, DeCamp MM, Goldblum JR,

Adelstein DJ, Zuccaro Jr G, et al. M1a/M1b esopha-

geal carcinoma: clinical relevance. J Thorac Cardio-

vasc Surg 1999;118(5):900–7.

[7] Paulson EK. Evaluation of the liver for metastatic dis-

ease. Semin Liver Dis 2001;21:225–36.

[8] Robinson PJA. Imaging liver metastases: current limi-

tations and future prospects. Br J Radiol 2000;73:

234–41.

[9] Bemelman WA, van Delden OM, van Lanschot JJ, de

Wit LT, Smits NJ, Fockens P, et al. Laparoscopy and

laparoscopic ultrasonography in staging of carcinoma

of the esophagus and gastric cardia. J Am Coll Surg

1995;181(5):421–5.

[10] van Dijkum EJ, de Wit LT, van Delden OM, Kruyt PM,

van Lanschot JJ, Rauws EA, et al. Staging laparoscopy

and laparoscopic ultrasonography in more than 400 pa-

tients with upper gastrointestinal carcinoma. J Am Coll

Surg 1999;189(5):459–65.



Cancer Action Project: overall design and findings from

baseline screening. Lancet 1999;354(9173):99–105.

[12] Margolis ML, Howlett P, Bubanj R. Pulmonary nod-

ules in patients with esophageal carcinoma. J Clin Gas-

troenterol 1998;26:245–8.

[13] Rosenthal DI. Radiologic diagnosis of bone metas-

tases. Cancer 1997;80(8 Suppl):1595–607.

[14] Reincke M. Mutations in adrenocortical tumors. Horm

Metab Res 1998;30(6–7):447–55.

[15] Lockhart ME, Smith JK, Kenney PJ. Imaging of adre-

nal masses. Eur J Radiol 2002;41(2):95–112.

[16] Luketich JD, Friedman DM, Weigel TL, Meehan MA,

Keenan RJ, Townsend DW, et al. Evaluation of distant

metastases in esophageal cancer: 100 consecutive

positron emission tomography scans. Ann Thorac Surg

1999;68(4):1133–6.

[17] Block MI, Patterson GA, Sundaresan RS, Bailey MS,

Flanagan FL, Dehdashti F, et al. Improvement in stag-

ing of esophageal cancer with the addition of positron

emission tomography. Ann Thorac Surg 1997;64(3):

770–6.

[18] Croft DR, Trapp J, Kernstine K, Kirchner P, Mullan B,

Galvin J, et al. FDG-PET imaging and the diagnosis of

non-small cell lung cancer in a region of high histoplas-

mosis prevalence. Lung Cancer 2002;36(3):297–301.

[19] Korst RJ, Rusch VW, Venkatraman E, Bains MS, Burt

ME, Downey RJ, et al. Proposed revision of the stag-

ing classification for esophageal cancer. J Thorac Car-

diovasc Surg 1998;115(3):660–9.

[20] Altorki N, Kent M, Ferrara C, Port J. Three-field lymph

node dissection for squamous cell and adenocarcinoma

of the esophagus. Ann Surg 2002;236(2):177–83.

[21] Akiyama H, Kogure T, Itai Y. The esophageal axis and

its relationship to the resectability of carcinoma of the

esophagus. Ann Surg 1972;176:30–6.

[22] Rice TW. Clinical staging of esophageal carcinoma.

CT, EUS, and PET. Chest Surg Clin N Am 2000;

10(3):471–85.

[23] Thompson WM, Halvorsen RA. Staging esophageal

carcinoma II. CT and MRI. Semin Oncol 1994;21:

447–52.

[24] Picus D, Balfe DM, Koehler RE, Roper CL, Owen JW.

Computed tomography in the staging of esophageal

carcinoma. Radiology 1983;146(2):433–8.

[25] Lightdale CJ. Staging esophageal carcinoma I. En-

doscopic ultrasonography. Semin Oncol 1994;21:

438–46.

[26] Rosch T. Endosonographic staging of esophageal can-

cer: a review of literature results. Gastrointest Endosc

Clin N Am 1995;5:537–47.

[27] Schlick T, Heintz A, Junginger T. The examiner’s

learning effect and its influence on the quality of en-

doscopic ultrasonography in carcinoma of the esopha-

gus and gastric cardia. Surg Endosc 1999;13:894–8.

[28] Silverstein FE, Martin RW, Kimmey MB, Jiranek GC,

Franklin DW, Proctor A. Evaluation of an endoscopic

ultrasound probe: in vitro and in vivo canine studies.

Gastroenterology 1989;96:1058–62.

[29] Tokiyama H, Yanai H, Nakamura H, Takeo Y, Yoshida

T, Okita K. Three-dimensional endoscopic ultrasonog-

raphy of lesions of the upper gastrointestinal tract using

a radial – linear switchable thin ultrasound probe.

J Gastroenterol Hepatol 1999;14(12):1212–8.

[30] Waxman I. Endosonography in columnar-lined esoph-

agus. Gastroenterol Clin N Am 1997;26(3):607–12.

[31] Ozawa S, Imai Y, Suwa T, Kitajima M. What’s new in

imaging? New magnetic resonance imaging of esoph-

ageal cancer using an endoluminal surface coil and

antibody-coated magnetite particles. Recent Results

Cancer Res 2000;155:73–87.

[32] Eloubeidi MA, Desmond R, Arguedas MR, Reed CE,

Wilcox CM. Prognostic factors for the survival of

patients with esophageal carcinoma in the US: the im-

portance of tumor length and lymph node status. Can-

cer 2002;95(7):1434–43.

[33] Kato H, Kuwano H, Nakajima M, Miyazaki T, Yoshi-

kawa M, Ojima H, et al. Comparison between positron

emission tomography and computed tomography in the

use of the assessment of esophageal carcinoma. Cancer

2002;94(4):921–8.

[34] Romagnuolo J, Scott J, Hawes RH, Hoffman BJ, Reed

CE, Aithal GP, et al. Helical CT versus EUS with fine

needle aspiration for celiac nodal assessment in patients

with esophageal cancer. Gastrointest Endosc 2002;

55(6):648–54.

[35] Lerut T, Flamen P, Ectors N, Van Cutsem E, Peeters M,

Hiele M, et al. Histopathologic validation of lymph

node staging with FDG-PET scan in cancer of the

esophagus and gastroesophageal junction: a prospec-


tive study based on primary surgery with extensive

lymphadenectomy. Ann Surg 2000;232(6):743–52.

[36] Natsugoe S, Nakashima S, Matsumoto M, Sakita H,

Sakamoto F, Okumura H, et al. Biologic and imaging

diagnosis of lymph node metastasis in esophageal car-

cinoma. J Surg Oncol 2002;81(1):25–32.

[37] CatalanoMF, Sivak Jr MV, Rice T, Gragg LA, Van Dam

J. Endosonographic features predictive of lymph node

metastasis. Gastrointest Endosc 1994;40(4):442–6.

[38] Natsugoe S, Yoshinaka H, Shimada M, Shirao K,

Nakano S, Kusano C, et al. Assessment of cervical

lymph node metastasis in esophageal carcinoma using

ultrasonography. Ann Surg 1999;229(1):62–6.

[39] Kelly S, Harris KM, Berry E, Hutton J, Roderick P,

Cullingworth J, et al. A systematic review of the stag-

ing performance of endoscopic ultrasound in gastro-

oesophageal carcinoma. Gut 2001;49(4):534–9.

[40] Eloubeidi MA, Wallace MB, Reed CE, Hadzijahic N,

Lewin DN, Van Velse A, et al. The utility of EUS and

EUS-guided fine needle aspiration in detecting celiac

lymph node metastasis in patients with esophageal can-

cer: a single-center experience. Gastrointest Endosc

2001;54(6):714–9.

[41] Parmar KS, Zwischenberger JB, Reeves AL, Waxman

I. Clinical impact of endoscopic ultrasound-guided fine

needle aspiration of celiac axis lymph nodes (M1a dis-

ease) in esophageal cancer. Ann Thorac Surg 2002;

73(3):916–20.

[42] Urba SG, Orringer MB, Turrisi A, Iannettoni M, For-

astiere A, Strawderman M. Randomized trial of preop-

erative chemoradiation versus surgery alone in patients

with locoregional esophageal carcinoma. J Clin Oncol

2001;19(2):305–13.

[43] Bosset JF, Gignoux M, Triboulet JP, Tiret E, Mantion

G, Elias D, et al. Chemoradiotherapy followed by sur-

gery compared with surgery alone in squamous-cell

cancer of the esophagus. N Engl J Med 1997;337(3):

161–7.

[44] Jones DR, Parker Jr LA, Detterbeck FC, Egan TM.

Inadequacy of computed tomography in assessing

patients with esophageal carcinoma after induction

chemoradiotherapy. Cancer 1999;85(5):1026–32.

[45] Beseth BD, Bedford R, Isacoff WH, Holmes EC,

Cameron RB. Endoscopic ultrasound does not accu-

rately assess pathologic stage of esophageal cancer af-

ter neoadjuvant chemoradiotherapy. Am Surg 2000;

66(9):827–31.

[46] Mallery S, DeCamp M, Bueno R, Mentzer SJ, Sug-

arbaker DJ, Swanson SJ, et al. Pretreatment staging

by endoscopic ultrasonography does not predict com-

plete response to neoadjuvant chemoradiation in pa-

tients with esophageal carcinoma. Cancer 1999;86(5):

764–9.

[47] Willis J, Cooper GS, Isenberg G, Sivak Jr MV, Levitan

N, Clayman J, et al. Correlation of EUS measurement

with pathologic assessment of neoadjuvant therapy re-

sponse in esophageal carcinoma. Gastrointest Endosc

2002;55(6):655–61.

[48] Kato H, Kuwano H, Nakajima M, Miyazaki T, Yo-

shikawa M, Masuda N, et al. Usefulness of positron

emission tomography for assessing the response of

neoadjuvant chemoradiotherapy in patients with esoph-

ageal cancer. Am J Surg 2002;184(3):279–83.

[49] Downey RJ, Akhurst T, Ilson D, Ginsberg R, Bains

MS, Gonen M, et al. Whole body 18FDG-PET and

the response of esophageal cancer to induction ther-

apy: results of a prospective trial. J Clin Oncol 2003;

21(3):428–32.


Fluorescent bronchoscopy

Sebastien Gilbert, MD, James D. Luketich, MD, Neil A. Christie, MD*

Division of Thoracic and Foregut Surgery, University of Pittsburgh Medical Center, 200 Lothrop Street,

Suite C800, Pittsburgh, PA 15213, USA

Long-term lung cancer survival in North America patients [5–7]. Based on these results, there are cur-

remains less than 15% and has not changed appre-

ciably over the past several decades [1]. It is antici-

pated that the number of patients who have lung

cancer will continue to rise in North America over the

next 15 years despite the gradual decrease in the

proportion of people who smoke because the risk of

lung cancer remains elevated in previous smokers. If

current smoking trends continue to show a decline,

the majority of cancers seen in the future will be in

previous smokers [2]. Given the ineffectiveness of

treatment for advanced cancer, early lung cancer

detection and treatment offer the greatest potential

for achieving a decrease in lung cancer mortality.

Detection of lung cancer at an earlier stage should

result in improved survival and the opportunity for

less invasive therapy such as thoracoscopic resection

of peripheral tumors and endobronchial ablative ther-

apy for central airway tumors. Experience with

screening and early diagnosis and treatment in other

epithelial organs such as the cervix have shown that

early detection and treatment of lesions can be

accomplished with improved cure rates [3]. Based

on calculations of estimated tumor doubling times, it

is estimated that a tumor will grow for months or

even years before reaching a size detectable with

standard imaging techniques [4], which should allow

a significant window of time in which to detect early

tumors in high-risk patients. Large lung cancer

screening trials evaluating sputum cytology and chest

radiography resulted in earlier diagnosis with im-

proved survival in identified patients, but no differ-

ence in overall survival when compared with control


doi:10.1016/S1547-4127(04)00041-6

* Corresponding author. Shadyside Medical Center,

5200 Center Avenue, Suite 715, Pittsburgh, PA 15232.

E-mail address: [email protected] (N.A. Christie).

rently no recommendations for lung cancer screen-

ing. There has been a resurgence of interest in

screening for lung cancer, however, with the advent

of more sensitive screening tests such as CT and

fluorescent bronchoscopy. While CT scans can iden-

tify subcentimeter parenchymal nodules accurately,

early endobronchial lesions and central tumors are

not seen well [8]. Sputum cytologic analysis offers

the detection of clinically occult lesions, but it can-

not localize the lesion in the airway. In situ and

microinvasive cancers might not produce visible ab-

normalities on standard white-light bronchoscopy

(WLB). Even with multiple bronchoscopies or se-

lective segmental bronchial brushing, the source

of cytologically abnormal cells can be difficult to

localize. Fluorescent bronchoscopy offers the poten-

tial for more accurate discovery and localization of

early tumors and premalignant epithelial changes

which are generally not well seen with WLB. In a

study by Woolmer [9], only 29% of carcinoma in situ

(CIS) detected by sputum cytology examination

could be localized by conventional WLB.

Fluorescent bronchoscopy

Fluorescent bronchoscopy uses the observation

that dysplastic tissue and areas of CIS demonstrate

weaker green fluorescence than normal tissues when

illuminated with blue light. Fluorescent properties of

human tissues have been the object of scientific

interest since the early 1920s [10]. Early attempts at

endobronchial surveillance used fluorescent dyes.

Current fluorescent bronchoscopy exploits the auto-

fluorescence characteristics of premalignant and ma-

lignant lesions of the bronchial mucosa, and no

exogenous dyes are required. The best-known fluo-

s reserved.

S. Gilbert et al / Thorac Surg Clin 14 (2004) 71–7772

rescent bronchoscope that works on this principle is

the lung imaging fluorescence endoscope (LIFE,

Xillix Technologies Corp, Vancouver, British Colum-

bia, Canada), which was developed by Dr. Stephen

Lam at the British Columbia Cancer Agency [11].

The LIFE system uses tissue fluorescence to localize

suspicious lesions in the tracheobronchial tree. A

helium–cadmium laser light source projects light at

a wavelength of 442 nm [12] to induce tissue fluo-

rescence. Two cameras, one with a red filter and one

with a green filter, capture the fluorescent signal. The

ratio between the red and green fluorescence is

used to distinguish benign from malignant tissue.

Real-time digitized images are constructed using

the relative intensities of red and green fluores-

cence, and a nonlinear analysis combines the red

and green fluorescence intensity values to create a

single number that discriminates between normal and

abnormal tissue sites. A computer-enhanced pseudo-

image is created, allowing the delineation of abnor-

mal areas when displayed on the monitor. Suspicious

areas appear reddish-brown and normal areas appear

green. Abnormal-appearing mucosa can then be

biopsied to identify dysplastic areas, CIS, or micro-

invasive cancers.

LIFE bronchoscopy represents the evolution and

refinement of existing fluorescence imaging concepts

and techniques rather than a novel diagnostic imaging

modality. Normal tissue produces significantly higher

fluorescence intensity than dysplastic lesions or CIS,

particularly in the green region of the emission

spectrum [11,12]. Decrease in the autofluorescence

in early cancer or dysplastic tissue is likely a result of

multiple factors. Most of the fluorescent signal origi-

nates from the submucosa. The loss of autofluores-

cence, as evidenced by a reddish-brown image on the

LIFE endoscope, might be related to destruction of

the extracellular matrix by metalloproteinases [13].

Increased bronchial microvascular density and im-

paired transmission of fluorescent signal through a

thickened malignant or dysplastic epithelium might

also contribute [14].

The SAFE 1000 (Pentax, Asahi Optical, Tokyo,

Japan) and the D-light (Storz, Tuttlingen, Germany)

are other commercially available systems.

Bronchoscopic technique

Bronchoscopy is performed on an outpatient basis

using local anesthesia with or without intravenous

sedation. It is combined with a conventional WLB

examination and adds approximately 15 minutes to

the overall procedure time [15]. An Olympus BF20

(Olympus America, Melville, New York) is used.

During a LIFE examination, areas of normal green

fluorescence are labeled class I (normal), whereas

areas of increased redness with indistinct borders are

labeled class II (abnormal). Class III lesions (suspi-

cious) show deeper red coloration and distinct bor-

ders. This classification scheme was described by

Lam in 1998 [15]. Most LIFE bronchoscopists would

agree that there is a learning curve of approximately

20 examinations, during which accuracy improves

consistently. The biopsy specimens obtained during

LIFE bronchoscopy should be interpreted by lung

pathologists according to defined criteria published in

the World Health Organization lung tumor classifica-

tion [16]. Preneoplastic lesions include squamous

dysplasia and CIS. Four grades of preneoplastic

lesions have been defined (ie, mild, moderate, severe

dysplasia, and CIS) based on the distribution of

atypical cells and mitotic figures. Although the re-

producibility of this system remains to be established,

it is part of a concerted effort to achieve a standard-

ized framework for classification [16]. Inconsistency

in pathologic classification can confound study re-

sults, and some studies have shown interobserver

variability on the pathologic classification.

Prebronchoscopy risk stratification

The success of any screening program depends on

the prescreening risk of cancer in the group being

evaluated. Although 80% of lung cancers are attrib-

uted to smoking, less than 20% of smokers develop

lung cancer in their lifetime. The yearly incidence of

lung cancer in the general population of the United

States is 0.05% to 0.09% [1]. Epidemiologic studies

show that an increased risk of lung cancer is seen in

patients who have more extensive smoking histories.

Presence of chronic obstructive pulmonary disease

has also been associated with an increased risk of

lung cancer. Previously treated primary lung cancer

also represents a risk factor for second primary lung

cancer. Clinically significant second primary lung

cancers are diagnosed in patients who have prior

non–small-cell lung cancer at a rate of 1% to 3%

per patient per year. Despite postoperative follow-up

care, only 50% of second primaries are resectable. At

the time of diagnosis 19% of these cancers are locally

advanced, 65% are associated with metastatic dis-

ease, and 20% occur in patients who are not surgical

candidates because of insufficient pulmonary reserve.

The 5-year survival rate after complete resection of a

second primary lung cancer is only 20% [17]. Sputum

screening with conventional cytology can identify

S. Gilbert et al / Thorac Surg Clin 14 (2004) 71–77 73

patients who are at high risk for endobronchial

neoplasia or dysplasia, and sputum immunostaining

promises even greater sensitivity. Experiments with

monoclonal antibody 703D4 have shown that over-

expression of an RNA binding protein, hnRNP

A2/B1, is a powerful predictor of early subclinical

cancer in high-risk groups [18].

Chemoprevention

Saccomanno observed in longitudinal studies that

abnormal bronchial epithelial cell changes predated

development of invasive lung cancer [19,20]. It is

now believed that lung cancers develop through a

series of sequential morphologic changes from meta-

plasia to dysplasia to CIS before the development of

invasive cancer. Bronchoscopic identification of these

premalignant lesions can be used to identify patients

for chemopreventative therapy or sequential monitor-

ing. Longitudinal monitoring of these patients should

identify early cancers when and if they appear and

allow clinicians to observe the natural history of these

lesions. Ten percent of patients who have moderate

dysplasia and 40% to 83% of patients who have

severe dysplasia progress to invasive cancer [21].

Based on autopsy studies of the tracheobronchial tree

of smokers performed in the 1960s and 1970s, the

incidence of CIS is probably between 2.2% and

22.5% [22]. The exact proportion of patients who

have CIS in whom disease will progress to invasive

cancer is not known. There have been reports that

some individuals continue to show malignant cells in

sputum for several years without symptoms or ab-

normality on chest radiograph. Frost showed that

only 43% of smokers who had marked dysplasia on

sputum cytology developed cancer over a 10-year

follow-up period. Saccomanno noted progression

from dysplasia to cancer in only three of 16 uranium

workers [23]. This variability underlines the lack of

knowledge of the natural history of premalignant

changes in the tracheobronchial tree. The variability

in incidence might reflect a lack of standardization in

the pathological definition of CIS over the years. It is

possible that a significant proportion of these lesions

were initially misclassified and that the prevalence of

CIS in the tracheobronchial tree of smokers has been

overestimated. This theory is corroborated by the

observed incidence of lung cancer, which is much

lower than the reported rate of CIS. The incidence of

second primary lung cancers should also be higher

than the reported 1% to 4% per year [17,24,25]. A

more recent observational study of bronchial CIS

suggested that such lesions almost uniformly prog-

ress to microinvasive carcinoma. LIFE should allow

localization, and longitudinal follow-up of these le-

sions which will allow clinicians to learn about the

natural history of these lesions and their risks of

progressing to invasive cancer.

Improved identification of dysplastic premalig-

nant lesions will provide an opportunity to intervene

with chemoprevention therapy to try to halt the

progression from dysplasia or CIS to invasive cancer

[26]. Chemoprevention, treatment directed at stop-

ping the progression of multistep lung carcinogene-

sis, will require a better understanding of the biology

of premalignant bronchial lesions and the develop-

ment of effective chemopreventative agents. Identifi-

cation of patients who have early endobronchial

lesions can be used to validate less invasive assays

on sputum or blood, which can be evaluated in

parallel to the pathologic and cytologic changes

occurring in bronchial epithelium and epithelial cells.

Clinical trials with lung imaging fluorescence

endoscope bronchoscopy

A large experience with LIFE bronchoscopy has

been reported in the literature. Generally, it attests to

the improved sensitivity of LIFE bronchoscopy over

standard WLB in detecting dysplasia and CIS. One

study involving 173 high-risk patients from seven

centers in the United States and Canada demonstrated

that the combination of WLB and LIFE bronchos-

copy improved clinicians’ ability to detect prema-

lignant and early-stage malignant bronchial lesions

endoscopically [15]. In that group of patients, WLB

had a sensitivity of 9% for detecting moderate/

severe dysplasia or CIS and a sensitivity of 65% for

detecting intraepithelial neoplasms and microinvasive

carcinoma. The addition of LIFE bronchoscopy to

WLB yielded sensitivity values of 56% and 95% for

preinvasive and invasive lesions, respectively, which

represents a 6.3-fold increase in the detection of

intraepithelial lesions and a 1.5-fold increase in the

detection of preneoplastic lesions. These results con-

firm the difficulties in detecting early neoplastic

lesions by conventional bronchoscopy alone. Thirty-

nine percent of patients who have abnormal sputum

cytology will require more than one WLB to identify

an associated neoplastic lesion, even if more than half

of these lesions have progressed beyond the CIS stage

[27]. LIFE bronchoscopy can help overcome this lack

of sensitivity and allow histologic follow-up of pre-

malignant lesions that would otherwise be undetect-

able with WLB. This information will be helpful in

characterizing the natural history of these lesions.


LIFE bronchoscopy has also been compared with

WLB in a randomized fashion [28]. The aim of the

trial was to assess the efficiency of each technique in

detecting premalignant lesions of the airways. It

included 55 patients who were considered to be at

high risk for lung cancer because of smoking history

(age >30; 7.1 pack-years mean smoking history),

documented airflow obstruction, and abnormal spu-

tum cytology (87%) or a past history of lung cancer.

Each patient was randomized to LIFE or WLB, and

each examination was performed by a different bron-

choscopist. The operator was blinded to the results of

the previous examination. A mean of seven biopsy

specimens was retrieved per patient. The sensitivities

of LIFE alone (68.8%) and that of LIFE combined

with WLB (81.3%) were significantly higher than

that of WLB alone (21.9%), but the combined exami-

nation was significantly less specific than WLB alone

(47.8% versus 78.3%). In this trial, neither the order

in which the procedures were performed nor the

bronchoscopist had a significant impact on sensitivity

and specificity.

Other published studies corroborate the enhanced

ability of LIFE to detect premalignant lesions [29,30].

On average, LIFE examination leads to more biopsies

because more areas of mucosa appear abnormal. An

argument can be made that the improved sensitivity

is merely related to the fact that more areas of the

airway are biopsied during LIFE. When specificity is

taken into account, however, and detection ratios of

LIFE and WLB are compared, the difference in

effectiveness remains significant [29]. Similar results

have been published by other groups [25,26,31–33].

The sensitivity of LIFE ranged from 73% to 89%,

and the specificity ranged from 46% to 61%. Once

again, the addition of LIFE to WLB improved the sen-

sitivity of the bronchoscopic examination [29–31,33].

One study failed to demonstrate increased sensitivity

with LIFE bronchoscopy [34]; however, this might be

explained by the selection of a relatively low-risk

population compared with other trials (ie, >20 pack-

years smoking history alone).

The performance of LIFE bronchoscopy is related

directly to the operator’s skill and experience with the

technique. Scope-induced trauma or other artifacts

can easily be mistaken for an area of abnormal fluo-

rescence. Although biopsy of these areas will un-

doubtedly help maintain a high sensitivity, it will

potentially increase the number of false-negatives

(ie, decrease sensitivity). In the authors’ experience,

the incidence of fluorescent anomalies might be

higher when LIFE is performed after WLB, which

is most likely related to scope-induced mucosal

trauma. Lastly, the problem of low specificity is not

unique to LIFE as a screening test. CT scanning for

early lung cancer, mammography for breast cancer,

and prostate specific antigen (PSA) for prostate can-

cer are noteworthy examples of relatively nonspecific

screening tests [8,31]. The result of decreased sensi-

tivity is the taking of additional biopsies that do not

pose a significant risk to the patient. The develop-

ment of quantitative fluorescence and the use of

nebulized photosensitizers and endobronchial ultra-

sonography might help overcome some of the speci-

ficity limitations.

Lung imaging fluorescence endoscope

bronchoscopy for cancer staging

A recent study found that LIFE bronchoscopy

was useful in staging early endobronchial lesions and

determining which lesions were amenable to endo-

bronchial therapy as opposed to more invasive therapy

[35]. Twenty-three patients who had radiologically

occult tumors who were referred for endobronchial

therapy were evaluated with LIFE bronchoscopy. On

high-resolution CT scanning, radiologically apparent

disease (lymph nodes or primary tumor) was detected

in four patients. The remaining 19 patients were

evaluated with fluorescent bronchoscopy. Six patients

had tumors less than 1 cm in diameter, and the distal

margin of the lesion could be seen bronchoscopically.

These patients were treated with endobronchial ther-

apy. The remaining patients had more extensive dis-

ease on LIFE bronchoscopy. Six of the 13 patients

underwent surgical resection of T1 or T2 node-nega-

tive tumors. On patient had stage II N1 disease. The

remaining patients were medically inoperable and

were treated with external beam radiation (n = 4) or

endoluminal therapy (n = 3). Of the localized tumors

treated with endoluminal therapy, no recurrence was

seen within a 30- to 50-month follow-up period.

High-resolution CT and fluorescent bronchoscopy

offer the ability to better stage patients who have

radiologically occult lung cancer, preventing and

identifying the subset of patients who can be treated

endobronchially with a good expectation of cure.

Another circumstance in which LIFE bronchos-

copy has been evaluated is in the preoperative

assessment of patients who had known lung carci-

noma to detect synchronous primary tumors [36].

Seventy-two patients who had known lung cancer

(69 non–small-cell; three limited-stage small-cell)

were evaluated with LIFE bronchoscopy and WLB.

Three synchronous cancers were detected, one by

WLB and LIFE and two by LIFE bronchoscopy

alone. Two of the three patients had squamous cell


primary cancer. The third tumor was not subclas-

sified histologically. Two of the three patients had

advanced cancers. One patient died of postobstruc-

tive pneumonia before the initiation of any therapy

and another patient had advanced nodal disease that

precluded resection. A third patient underwent right

pneumonectomy followed by endobronchial therapy

for a small lesion identified with LIFE bronchoscopy.

The authors recommended LIFE bronchoscopy to

evaluate for synchronous occult lung cancers, but

only immediately before surgery after all other stag-

ing procedures had been completed.

Fig. 2. LIFE bronchoscopy image corresponding to image

in Fig. 1.

University of Pittsburgh experience

At the University of Pittsburgh, LIFE bronchos-

copy was used to screen patients for the occurrence

of second primary lung cancer following pulmonary

resection for non–small-cell lung cancer between

1997 and 2002. The initial experience has been

reported [37]. Ninety-five patients participated in

the screening program. Fifty-five had resected adeno-

carcinomas and 40 had resected squamous cell car-

cinomas. The examination frequency was annually

if no abnormalities were identified. Seventy-four per-

cent of patients had stage I cancer, 18% had stage II

cancer, and 8% had stage III cancer as their initial

primary cancer. Seventy-two percent of patients had

undergone previous lobectomy, 6% had undergone

pneumonectomy, and 22% had undergone segmental

resection. Of the 12 abnormal areas identified patho-

Fig. 1. Conventional WLB.

logically (high-grade dysplasia, CIS, or microinva-

sive cancer), six were detected with the fluorescent

examination and four with WLB with sensitivities of

50% and 33%, respectively. Fig. 1 shows an endo-

scopic view of an area of CIS that was occult on

WLB. Fig. 2 shows the identical area on LIFE bron-

choscopy, in which the abnormality was visualized.

The specificity of fluorescent bronchoscopy was 76%

compared with 98% for WLB. Nine of 95 patients

(9%) had lesions for which treatment could be

considered. The poor sensitivity of WLB in this

group of patients might, in part, have been related

to the fact that two or three random biopsies were

taken in every patient, increasing the potential of

identifying bronchoscopically occult lesions.

Future directions

The evolution of LIFE technology toward a more

objective quantification of tissue fluorescence and the

addition of other complementary endoscopic tools

such as endobronchial ultrasound might improve the

specificity of the technique, which would ultimately

benefit patients by decreasing the number of biopsies

performed and the time requirement for the exami-

nation. With an emphasis on screening and early diag-

nosis, clinicians might see more patients who have

radiologically occult lesions who will be potential

candidates for endobronchial therapy for attempted

cure. One of the major challenges in achieving

widespread integration of this modality in clinical


practice is delineation of the subgroups of patients

who are at higher risk of developing lung cancer,

who are appropriate patients for this invasive and

relatively labor-intensive evaluation.

Endobronchial biopsy specimen evaluation with

methods other than histologic evaluation potentially

offers a fruitful opportunity. Molecular abnormalities

have been observed in patients who have histologi-

cally normal epithelium and might represent a more

suitable marker than histologic patterns, which do not

always correlate predictably with outcome [38]. Cel-

lular changes in gene copy number, gene expression,

and protein profiles might represent better predictors

of progression of dysplasia and CIS and act as

surrogate markers for efficacy in chemoprevention

studies [39]. That cytogenetic changes are occurring

with associated molecular alterations is compatible

with the current understanding of the molecular

pathogenesis of cancer [40]. The ability to sample

preneoplastic lesions accurately for molecular and

histologic analysis and to follow their progression

or regression longitudinally should prove to be valu-

able tools in outcomes research. Multiple chromo-

somal abnormalities have been noted and gene

mutations have been detected using sensitive molecu-

lar techniques such as polymerase chain reaction [39].

The results of these molecular studies have so far

failed to identify a single marker that is expressed

consistently in cancers or dysplastic lesions but not

seen in normal endobronchial cells, and no markers

seen in dysplastic cells have been reliably predictive

for progression to cancer. With the development of

comparative genomic hybridization techniques, clini-

cians have the potential to get a much broader picture

of total genomic damage patterns, which might be

more predictive [41]. New techniques of comparative

analysis of gene expression across thousands of genes

[42] and new methods of proteomic analysis might

also be adapted to these samples [43]. With this large

amount of additional information, a predictive mo-

lecular signature of lesions that are likely to progress

to cancer could be identified. The addition of these

newer methods of tissue analysis should stand to

improve the utility of LIFE bronchoscopy in the

future in clinicians’ attempts to decrease the mortality

from lung cancer and facilitate less invasive endo-

bronchial treatments.

Acknowledgment

The authors would like to acknowledge the

assistance of Jill Ireland in the preparation of

this manuscript.

References

[1] American Cancer Society. Cancer Facts and Figures

2003. No 5008.03. Atlanta: American Cancer Society;

2003. p. 1–48.

[2] Burns DM. Primary prevention, smoking, and smoking

cessation: implications for future trends in lung cancer

prevention. Cancer 2000;89:2506–9.

[3] Anderson GH, Boyes DA, Benedet JL, Le Riche JC,

Matisic JP, Suenk C, et al. Organization and results of

the cervical cytology screening program in British Co-

lumbia, 1955–1985. BMJ 1988;296:975–8.

[4] Geddes DM. The natural history of lung cancer:

a review based on rates of tumor growth. Br J Dis

Chest 1979;73:1–17.

[5] Flehinger BJ, Melamed MR, Zaman MB, Heelan RT,

Perchick WB, Martini N. Early lung cancer detection:

results of the initial (prevalence) radiologic and cyto-

logic screening in the Memorial Sloan-Kettering study.

Am Rev Respir Dis 1984;130(4):555–60.

[6] Fontana RS, Sanderson DR, Taylor WF, Woolner LB,

Miller WE, Muhm JR, et al. Early lung cancer detec-

tion: results of the initial (prevalence) radiologic and

cytologic screening in the Mayo Clinic study. Am Rev

Respir Dis 1984;130(4):561–5.

[7] Frost JK, Ball Jr WC, Levin ML, Tockman MS, Baker

RR, Carter D, et al. Early lung cancer detection: results

of the initial (prevalence) radiologic and cytologic

screening in the Johns Hopkins study. Am Rev Respir

Dis 1984;130(4):549–54.



Cancer Action Project: overall design and findings

from baseline screening. Lancet 1999;354(9173):

99–105.

[9] Woolner LB, Fontana RS, Cortese DA, Sanderson DR,

Bernatz PE, Payne WS, et al. Roentgenographically

occult lung cancer: pathologic findings and frequency

of multicentricity during a 10-year period. Mayo Clin

Proc 1984;59(7):453–66.

[10] Policard A. Etudes sur les aspects offerts par des

tumeurs experimentales examinees a la lumiere de

Woods. CR Soc Biol 1924;91:1423–5 [in french].

[11] Lam S, MacAulay C, Hung J, LeRiche J, Profio AE,

Palcic B. Detection of dysplasia and carcinoma in situ

with a lung imaging fluorescence endoscope device.

J Thorac Cardiovasc Surg 1993;105(6):1035–40.

[12] Hung J, Lam S, leRiche JC, Palcic B. Autofluores-

cence of normal and malignant bronchial tissue. Lasers

Surg Med 1991;11(2):99–105.

[13] Qu J, MacAulay C, Lam S. Optical properties of nor-

mal and carcinoma bronchial tissue. Appl Opt 1991;

11:99–105.

[14] Lam S, Becker HD. Future diagnostic procedures.

Chest Surg Clin N Am 1996;6(2):363–78.

[15] Lam S, Kennedy T, Unger M, Miller YE, Gelmont D,

Rusch V, et al. Localization of bronchial intraepithelial

neoplastic lesions by fluorescence bronchoscopy.

Chest 1998;113(3):696–702.


[16] Kerr KM. Pulmonary preinvasive neoplasia. J Clin

Path 2001;54(4):257–71.

[17] Johnson BE. Second lung cancers in patients after

treatment for an initial lung cancer. J Natl Cancr Inst

1998;90(18):1335–45.

[18] Tockman MS, Mulshine JL, Piantadosi S, Erozan YS,

Gupta PK, Ruckdeschel JC, et al. Prospective detection

of preclinical lung cancer: results from two studies of

heterogeneous nuclear ribonucleoprotein A2/B1 over-

expression. Clin Cancer Res 1997;3:2237–46.

[19] Saccomanno G, Saunders RP, Archer VE, et al. Cancer

of the lung: the cytology of sputum prior to the devel-

opment of carcinoma. Acta Cytol 1965;9:413–23.

[20] Saccomanno G, Archer VE, Auerbach O, Saunders RP,

Brennan LM, et al. Development of carcinoma of the

lung as reflected in exfoliated cells. Cancer 1974;33:

256–70.

[21] Frost JK, Ball WC, Levin ML, Jockman MS, Erozan

YS, et al. Sputum cytology: use and potential in mon-

itoring the workplace environment by screening for

biological effects of exposure. J Occup Med 1986;28:

692–703.

[22] Auerbach O, Hammond EC, Garfinkel L. Changes

in bronchial epithelium in relation to cigarette smok-

ing, 1955–1960 vs. 1970–1977. N Engl J Med 1979;

300(8):381–5.

[23] Band P, Feldstein M, Saccomanno G. Reversibility of

bronchial marked atypia: implications for chemopre-

vention. Cancer Detect Prev 1986;9:157–60.

[24] Gilbert S, Reid KR, Lam MY, Petsikas D. Who should

follow up lung cancer patients after operation? Ann

Thorac Surg 2000;69(6):1696–700.

[25] Venmans BJ, van Boxem TJ, Smit EF, Postmus PE,

Sutedja TG. Outcome of bronchial carcinoma in situ.

Chest 2000;117(6):1572–6.

[26] Arnold AM, Browman GP, Levine MN, D’Souza T,

Johnstone B, Skingley P, et al. The effect of the

synthetic retinoid etretinate on sputum cytology: re-

sults from a randomized trial. Br J Cancer 1992;65:

737–43.

[27] Bechtel JJ, Kelley WR, Petty TL, Patz DS, Sacco-

manno G. Outcome of 51 patients with roentgeno-

graphically occult lung cancer detected by sputum

cytologic testing: a community hospital program. Arch

Intern Med 1994;154(9):975–80.

[28] Hirsch FR, Prindiville SA, Miller YE, Franklin WA,

Dempsey EC, Murphy JR, et al. Fluorescence versus

white-light bronchoscopy for detection of preneoplas-

tic lesions: a randomized study. J Natl Cancr Instit

2001;93(18):1385–91.

[29] Hirsch FR, Prindiville SA, Miller YE, Franklin WA,

Dempsey EC, Murphy JR, et al. Fluorescence versus

white-light bronchoscopy for detection of preneoplas-

tic lesions: a randomized study. J Natl Cancr Inst 2001;

93(18):1385–91.

[30] Vermylen P, Pierard P, Roufosse C, Bosschaerts T,

Verhest A, Sculier JP, et al. Detection of bronchial

preneoplastic lesions and early lung cancer with fluo-

rescence bronchoscopy: a study about its ambulatory

feasibility under local anesthesia. Lung Cancer 1999;

25(3):161–8.

[31] Kennedy TC, Lam S, Hirsch FR. Review of recent

advances in fluorescence bronchoscopy in early local-

ization of central airway lung cancer. Oncologist 2001;

6(3):257–62.

[32] Kennedy TC, Hirsch FR, Miller YE, Prindiville S,

Murphy JR, Dempsey E, et al. A randomized study

of fluorescence bronchoscopy versus white-light bron-

choscopy for early detection of lung cancer in high-

risk patients. Lung Cancer 2000;29(1):244–5.

[33] Shibuya K, Fujisawa T, Hoshino H, Baba M, Saitoh Y,

Iizasa T, et al. Fluorescence bronchoscopy in the de-

tection of preinvasive bronchial lesions in patients with

sputum cytology suspicious or positive for malignancy.

Lung Cancer 2001;32(1):19–25.

[34] Kurie JM, Lee JS, Morice RC, Walsh GL, Khuri FR,

Broxson A, et al. Autofluorescence bronchoscopy in

the detection of squamous metaplasia and dysplasia in

current and former smokers. J Natl Cancr Instit 1998;

90(13):991–5.

[35] Sutedja TG, Codrington H, Risse EK, Breuer RH, van

Mourik JC, Golding RP, et al. Autofluorescence bron-

choscopy improves staging of radiographically occult

lung cancer and has an impact on therapeutic strategy.

Chest 2001;120(4):1327–32.

[36] van Rens MT, Schramel FM, Elbers JR, Lammers JW.

The clinical value of lung imaging fluorescence endos-

copy for detecting synchronous lung cancer. Lung

Cancer 2001;32:13–8.

[37] Weigel TL, Kosco PJ, Dacic S, Rusch VW, Ginsberg

RJ, Luketich JD. Postoperative fluorescence broncho-

scopic surveillance in non-small cell lung cancer

patients. Ann Thorac Surg 2001;71:967–70.

[38] Jett JR. Screening for lung cancer in high-risk groups:

current status of low-dose spiral CT scanning and spu-

tum. Sem Resp Crit Care Med 2000;21(5):385–92.

[39] Mitsudomi T, Lam S, Takayuki T, Gazdar AF. Detec-

tion and sequencing of p53 gene mutations in bron-

chial biopsy samples in patients with lung cancer.

Chest 1993;104:362–5.

[40] Vogelstein B, Kinzler KW. The multistep nature of

cancer. Trends Gent 1993;9:138–41.

[41] Forozan F, Mahlamaki EH, Monni O, Chen Y, Veld-

man R, Jiang Y, et al. Comparative genomic hybrid-

ization analysis of 38 breast cancer cell lines: a basis

for interpreting complementary DNA micro array data.

Cancer Res 2000;60:4519–25.

[42] Nacho M, Trachea T, Gao Y, Fujii T, Chen Y, Player A,

et al. Molecular characteristics of non-small cell lung

cancer. Proc Natl Acad Sci USA 2001;98:15203–8.

[43] Gharib TG, Chen G, Wang H, Huang CC, Prescott MS,

Shedden K, et al. Proteomic analysis of cytokeratin

isoforms uncovers association with survival in lung

adenocarcinoma. Neoplasia 2002;4(5):440–8.


Virtual bronchoscopy for evaluation of airway disease$

Steven E. Finkelstein, MDa, Ronald M. Summers, MD, PhDb,Dao M. Nguyen, MDa, David S. Schrump, MDa,*

aThoracic Oncology Section, Surgery Branch, Center for Cancer Research, National Cancer Institute, Building 10, Room 2B-07,

10 Center Drive, National Institutes of Health, Bethesda, MD 20892-1502, USAbDiagnostic Radiology Department, Warren Grant Magnuson Clinical Center, National Institutes of Health, Building 10,

Room 1c 660, 10 Center Drive, Bethesda, MD 20892, USA

A variety of medial conditions can cause airway tract; however, FB yields little information regard-

stenoses that require intervention by thoracic sur-

geons. For instance, patients who have primary lung

cancers or pulmonary metastases frequently develop

complete bronchial obstructions secondary to endo-

luminal tumors or extrinsic compression [1]. Patients

who have a variety of nonmalignant conditions can

also develop severe pulmonary complications includ-

ing fixed tracheobronchial stenoses, tracheomalacia,

or hemoptysis [2].

Typically, patients who have suspected airway

disease undergo diagnostic evaluation consisting of

chest radiographs and conventional CT scans fol-

lowed by fiberoptic bronchoscopy [3]. Conventional

CT generates two-dimensional (2D) cross-sectional

images of the thorax, which provide information re-

garding peribronchial anatomy. Standard CT scans

have a sensitivity of 63% to 100% and a specificity

of 61% to 99% for detection of major endobronchial

disease [4–6]. Occasionally, suboptimal scanning

techniques, inappropriate slice thickness, and other

artifacts might limit the accuracy of airway anatomy

defined by conventional CT scans [7].

In clinical practice, fiberoptic bronchoscopy (FB)

remains the gold standard for evaluation and surveil-

lance of endoluminal lesions within the respiratory

1547-4127/04/$ – see front matter Published by Elsevier Inc.

doi:10.1016/S1547-4127(04)00037-4

$Supplementary data associated with this article can

be found, in the online version, at doi:10.1016/S1547-4127

(04)00037-4.



(D.S. Schrump).

ing the extent of extraluminal disease or airway pat-

ency beyond a high-grade stenosis [8]. In addition, FB

might pose potential risks to patients who have ad-

vanced pulmonary disease (morbidity 0.8%) because

some degree of sedation might be required [9].

Recently, virtual bronchoscopy (also referred to as

CT bronchoscopy) has become available for noninva-

sive evaluation of the tracheobronchial tree [3]. Vir-

tual bronchoscopy (VB) uses three-dimensional (3D)

reconstruction of super high-resolution helical CT

(SHR-CT) images for delineation of the tracheobron-

chial tree. Perspective surface or volume rendering of

2D CT scan images are used to construct a virtual

airway. The natural contrast between the soft tissue of

the airway wall and air within the tracheobronchial

tree establishes a plane for generating the virtual

airway [6]. The viewer can navigate through the

virtual airway in a 3D manner analogous to standard

FB. VB also enables imaging of endoluminal and

extraluminal anatomy, which is not possible with FB.

The virtual airway can be manipulated in space and

evaluated from multiple angles (Fig. 1; Movie 1 in

online version of this article).

Technique

Two hundred to 300 contiguous images of the

thorax are obtained using a multislice helical CT scan-

ner [10]. The standard technique at the National Insti-

tutes of Health is 1.25 collimation, helical scan (HS)

mode (helical pitch 6; 7.5 mm table motion per

Fig. 1. Virtual bronchoscopy (VB) of normal anatomy. Viewpoint is above carina (A), looking into right mainstem bronchus (B),

a segmental bronchus (C). A three-dimensional movie of VB is available in the online version of this article found at doi:10.1016/

S1052-3359(03).

S.E. Finkelstein et al / Thorac Surg Clin 14 (2004) 79–8680

rotation, 120 kVp, 100 mAs, 0.8 sec tube rotation,

nonoverlapping reconstructions with a section inter-

val of 1.25 mm, and an effective z-axis resolution off1.6 mm). A standard algorithm is used to generate

the CT images [10]. The radiation dose with this

technique is the same or slightly less than that of a

conventional thoracic CT scan.

VB images can be viewed as standard CT scans or

reconstructed to 3D endoscopic views using commer-

cial software (ie, GE Navigator on a GE Advantage

Windows workstation, General Electric, Milwaukee,

Wisconsin). The radiologist or surgeon can review the

VB in a systemic manner. With the viewpoint placed

first in the proximal trachea, retrograde inspection of

the subglottis is done. Next, antegrade inspection of

the trachea is performed, followed by evaluation

of the right mainstem bronchus, right upper lobe

apical (B1), right upper lobe posterior (B2), right

upper lobe anterior (B3), bronchus intermedius, right

middle lobe, right middle lobe lateral (B4), right mid-

dle lobe medial (B5), right lower lobe superior (B6),

right lower lobe medial basal (B7), right lower lobe

anterior basal (B8), right lower lobe lateral basal (B9),

right lower lobe posterior basal (B10), left main stem

bronchus, left upper lobe apical posterior (B1 + 2),

left upper lobe anterior (B3), superior lingular (B4),

S.E. Finkelstein et al / Thorac Surg Clin 14 (2004) 79–86 81

inferior lingular (B5), left lower lobe superior (B6),

left lower lobe anteromedial basal (B7–8), left lower

lobe lateral basal (B9), and left lower lobe posterior

basal (B10) sequential bronchi. Using this technique,

airway abnormalities such as the presence or absence

of obstructive lesions (intrinsic or extrinsic), endo-

luminal masses, or mucosal abnormalities can be de-

fined precisely relative to bronchovascular anatomy.

Virtual bronchoscopy for thoracic malignancies

Accumulating data indicate that VB is extremely

useful for the detection of partial or complete bron-

chial obstructions secondary to endoluminal tumors

or extrinsic compression in cancer patients. In an

early study Fleiter et al [11] compared VB (performed

with a double-detector CT unit) and FB in 20 patients

who had thoracic malignancies. VB images were

created successfully in 19 of these patients; a strong

heart pulsation produced a motion artifact that pre-

vented accurate reconstruction in one individual.

Areas of high-grade stenoses were identified accu-

rately using both techniques; however, VB did not

detect discrete malignant infiltration and extraluminal

compression in five patients.

In a subsequent study Liewald et al [12] evaluated

30 lung cancer patients who had VB and FB. 3D im-

ages were created in all patients, and 13 obstructive

lesions were seen equally well by VB and FB. VB

demonstrated tracheobronchial anatomy beyond high-

grade stenoses in two patients; however, mucosal le-

sions were not visualized by VB. Rapp-Bernhardt

et al [13] observed no significant differences in the

location or extent of airway stenoses detected by VB

compared with FB in 21 patients who had esophageal

cancers infiltrating the respiratory tract. In a subse-

quent study these authors observed that conventional

Fig. 2. Endoluminal lesion in 51-year-old man who had metastatic

(B) FB. The lesion consists of large and small components in the

struction of the right upper lobe bronchus. (Courtesy of US gover

CT scans had a sensitivity of 92.9% and a specificity

of 100%, whereas VB had a sensitivity of 93.8% and

a specificity of 99.7% for detection of airway ste-

noses in lung cancer patients [14].

Hoppe et al [15] compared the efficacy of nonin-

vasive multidetector CT, which included VB images,

axial CT, coronal reformatted images, and sagittal

reformatted images, with that of FB. In their ex-

amination of 200 bronchial sections obtained from

20 lung cancer patients (15 patients had bronchial car-

cinoma and five did not have central airway disease),

these investigators observed that VB was a highly

accurate method for assessing the severity of tracheo-

bronchial stenoses; images from VB correlated ex-

tremely well with those obtained by FB (r = 0.91).

In a recent study the authors prospectively evalu-

ated VB and FB in 32 consecutive patients who had

suspected thoracic malignancies [16]. VB images

were obtained successfully in all patients during one

or two 17-second end-inspiration breath-holds (Fig. 2

shows data from representative patient). FB was

within normal limits in seven of 20 patients (35%),

and VB correlated with FB in these individuals. FB

revealed a total of 22 abnormalities in 13 patients; VB

detected 18 of these abnormalities, including 13 of

13 obstructing lesions (> 50% luminal occlusion) and

five of six endobronchial lesions with less than 50%

luminal obstruction. VB did not detect three mucosal

lesions identified by FB. Overall, the sensitivity of VB

was 82% for detection of any abnormality in the

respiratory tract, 100% for obstructing lesions, 83%

for endoluminal lesions, and 0% for mucosal lesions.

The specificity of VB was 100%.

In a subsequent prospective observer study the

authors evaluated SHR-CT, VB, and conventional CT

scans directly for detection of tracheobronchial ma-

lignancies in 44 patients [17]. Image acquisition and

simulation of tracheobronchial anatomy were suc-

renal cell carcinoma metastatic to the right hilum. (A) VB.

right mainstem bronchus (white arrows) with complete ob-

nment.)


cessful in all individuals. Thirty-two patients had

correlative FB within 1 month (Fig. 3 shows imaging

from representative patient). SHR-CT and VB corre-

lated with FB in nine patients who had normal

anatomy; however, CT demonstrated two false-posi-

tive obstructing lesions in one patient. Twenty-three

patients had a total of 35 abnormal FB findings.

SHR-CT and VB detected 29 (83%) of these abnor-

malities accurately, including 19 of 19 obstructing

lesions, 9 of 10 endoluminal masses, and one of six

mucosal lesions.

SHR-CT and VB failed to detect a small periph-

eral endobronchial mass in one patient, mucosal in-

flammation in two patients, and the presence of blood

without an identifiable source in three patients. It

is possible that endobronchial bleeding was intermit-

tent and not present at the time that SHR-CT and

VB were obtained in these three individuals. In con-

trast, SHR-CT and VB demonstrated 10 and 11 addi-

tional lesions, respectively, that were not identifiable

during FB because the size of the bronchoscope pre-

cluded evaluation of peripheral airways (in nine pa-

tients) or locations distal to high-grade stenoses (in

Fig. 3. Endoluminal lesion obstructing superior segment of left lo

FB (A), VB (B), and SHR-CT axial (D), coronal (E) and sagittal s

this lesion was not appreciated on consecutive conventional CT se

two patients). Because many patients in this study

underwent pulmonary resection, pathologic correla-

tion was possible in nine patients; six obstructive

lesions (67%) not detected by FB but visualized by

SHR-CT and VB were confirmed to be malignant.

Consistent with the authors’ previous report [16], the

sensitivities of SHR-CT and VB were 100% for ob-

structing lesions, 90% for endoluminal masses, and

17% for mucosal lesions. Specificities of SHR-CTand

VB were 100%.

In contrast to the excellent imaging obtained with

SHR-CT or VB, conventional CT scans were subop-

timal for identification of airway pathology. Twenty-

five patients who had bronchoscopic examinations

and SHR-CT and VB also underwent conventional

CT scanning. Seven patients (28%) had normal

examinations by FB; results of conventional CT scans

correlated with FB in six of these individuals (speci-

ficity 85%). Conventional CT scans depicted two

false-positive lesions in one patient. Eighteen patients

(72%) had a total of 29 abnormal FB findings.

Conventional CT detected 17 of these abnormalities

including 13 of 18 obstructive lesions, four of eight

wer lobe in 30-year-old man who had metastatic melanoma

ections (F) all visualized this lesion (white arrow); however,

ctions (C). (Courtesy of US government.)


endoluminal masses, and zero of three mucosal le-

sions. As with SHR-CT and VB, obstructive lesions

not visualized by FB were detected by conventional

CT because of the size limitation of the bronchoscope.

Three of six (50%) of these lesions were subsequently

confirmed to be positive by histologic evaluation. In

contrast to SHR-CT and VB, the sensitivity of con-

ventional CT was 72% for obstructive lesions, 50%

for endoluminal masses, 0% for mucosal lesions, and

59% overall. In no instance did conventional CT im-

prove upon the findings of SHR-CT or VB.

Virtual bronchoscopy for benign disease

VB has been used to evaluate airway stenoses

secondary to a variety of benign conditions [18–24].

Accumulating data indicate that VB has a sensitivity

of 94% to 100% for detection of benign airway

stenoses, and it is particularly useful for evaluation

of high-grade stenoses and delineation of airway

anatomy distal to these lesions, most of which are

located in the central airways. Ferretti et al [25] dem-

onstrated the utility of VB for evaluation of airway

obstruction in patients who had Mounier-Kuhn dis-

ease, tracheomalacia, post-tracheostomy strictures,

Wegener’s granulomatosis, tracheopathia osteoplas-

tica, and amyloidoses with tracheal wall involvement.

These investigators also demonstrated the use of VB

for the evaluation of airway compression by subster-

nal goiters, aneurysms of the great vessels, and me-

diastinal tumors. Burke et al [23] evaluated VB in

Fig. 4. Subglottic stenosis in patient who had Wegener’s granuloma

ability to reverse the viewing direction with the VB model is helpfu

airway from the posterior aspect (large arrow). The vocal cords (sm

in the subglottic region. (Courtesy of US government.)

21 patients who had primary tracheal strictures, eight

patients who had tracheomalacia, two patients who

had glottic webs, two patients who had tracheal

granulomas, seven patients who had vocal cord im-

mobility, and five patients who had innominate artery

aneurysms. The length and width of fixed airway

stenoses were demonstrated accurately by VB; steno-

sis-to-lumen ratios as determined by VB and FB

varied less than 10%. VB was particularly helpful in

evaluating high-grade airway stenoses that prevented

full bronchoscopic assessment. VB was less useful for

evaluation of dynamic airway obstructions, possibly

because images were obtained during breath-hold at

end-inspiration. Ferretti et al [25] obtained excellent

VB images of dynamic airway compromise second-

ary to tracheomalacia by scanning at end-inspiration

and end-expiration.

The authors’ group conducted a prospective ob-

server study recently comparing CT and VB to FB

for evaluation of airway stenoses in patients who had

Wegener’s granulomatous [26]. Helical CT scans

with 3D VB reconstruction of the trachea and bron-

chi were obtained in 11 patients. CT, VB, and FB

were performed and evaluated in a blinded manner.

Correlative FB was performed, on average, within

2 days of CT scans (Figs. 4, 5 show representative

data from two patients). VB visualized 188 of

198 bronchi (95%). Conventional CT scans detected

22 stenoses, whereas VB revealed 31 of 40 stenoses

identified by FB. Overall, this experience indicates

that VB can demonstrate anatomy down to the seg-

mental bronchi and that VB can detect the majority

tosis detected by way of (A) FB and (B) retrograde VB. The

l in locating the stenosis. The stenosis appears to narrow the

all arrows) are shown to indicate the location of the stenosis

Fig. 5. Bronchus intermedius stenosis in patient who had Wegener’s granulomatosis. (A) FB. (B) Endoscopic VB view.

(C) Exoscopic VB view. The viewpoint is in the distal trachea looking toward the main carina. This series of pictures shows the

left and right mainstem bronchus and the stenotic bronchus intermedius. The large arrow shows the extrastenotic region of

the bronchus intermedius that is only visible on the exoscopic VB view. The small arrows show the other stenotic region of the

bronchus intermedius that is visible on all views. (Courtesy of US government.)


of central airway stenoses in patients who have ad-

vanced Wegener’s granulomatosis.

Summary

The data presented above indicate that VB is a

novel and extremely useful modality for airway evalu-

ation in patients who have benign and malignant

disease. VB is noninvasive, with no additional radia-

tion exposure relative to standard CT scans of the

chest. Commercial software allows for the inter-

activity of 2D and 3D images. The ability to examine

2D and 3D anatomic detail from multiple direc-

tions enables precise assessment of intraluminal and

extraluminal pathology. The authors’ experience indi-

cates that VB is a superb modality for assessing the

length of airway stenoses and ascertaining airway

patency distal to these lesions (Fig. 6). As such, VB

has proven to be extremely useful for determining

the feasibility of endobronchial procedures such as

dilations, stent placements, and laser ablation of en-

dobronchial tumors. Ferretti et al [27] observed that

VB is an excellent noninvasive means for long-term

monitoring of tracheobronchial stents. Furthermore,

the authors have found VB useful for guiding the

bronchoscopic evaluation of patients who have inter-

mittent hemoptysis secondary to lesions in periph-

eral airways. The 3D anatomic detail provided by

VB has proven useful for assessing the feasibility of

lung-sparing procedures in patients who have limited

pulmonary reserve and for sequentially evaluating

treatment response in patients who have inopera-

ble disease.

Currently, the main limitation of VB pertains to its

inability to evaluate the mucosal surface of the

respiratory tract reliably. Although form can be de-

tected, mucosal color, irregularity, or friability cannot

Fig. 6. (A, B) Virtual bronchoscopy revealing high-grade obstruction of the proximal left mainstem bronchus in lung cancer

patient. (C) Fiberoptic bronchoscopy confirmed tumor recurrence.


be assessed. As such, VB cannot be used for routine

surveillance of patients at high risk of developing

airway malignancies. The development of novel aero-

solized contrast agents or spectroscopic techniques

that can discriminate benign versus malignant muco-

sal tissues might enhance the sensitivity and speci-

ficity of VB for the detection of preinvasive cancers

within the respiratory tract.

References

[1] Putnam Jr JB. New and evolving treatment methods

for pulmonary metastases. Semin Thorac Cardiovasc

Surg 2002;14(1):49–56.

[2] Sneller MC. Wegener’s granulomatous [clinical confer-

ence]. JAMA 1995;273:1288–91.

[3] McWilliams A, MacAulay C, Gazdar AF, et al. In-

novative molecular and imaging approaches for the

detection of lung cancer and its precursor lesions. On-

cogene 2002;21(45):6949–59.

[4] Colice GL, Chappel GJ, Frenchman SM, Solomon DA.

Comparison of computerized tomography with fiber-

optic bronchoscopy in identifying endobronchial ab-

normalities in patients with known or suspected lung

cancer. Am Rev Respir Dis 1985;131:397–400.

[5] Naidich DP, Lee JJ, Garay SM, McCauley DI, Aranda

CP, Boyd AD. Comparison of CT and fiberoptic bron-

choscopy in the evaluation of bronchial disease. AJR

Am J Roetgenol 1987;148:1–7.

[6] Harponik EF, Aquino SL, Vining DJ. Virtual bronchos-

copy. Clin Chest Med 1999;20:201–17.

[7] Dawn SK, Gotway MB, Webb WR. Multidetector-row

spiral computed tomography in the diagnosis of tho-

racic diseases. Respir Care 2001;46(9):912–21.

[8] Aquino SL, Vining DJ. Virtual bronchoscopy. Clin

Chest Med 1999;20:725–30.

[9] Pue CA, Pacht ER. Complications of fiberoptic bron-


choscopy at a university hospital. Chest 1995;107:

430–2.

[10] Summers RM, Sneller MC, Langford CA, Shelhamer

JH, Wood BJ. Improved virtual bronchoscopy using

a multi-slice helical CT scanner. In: Chen C-T,

Clough AV, editors. Medical imaging 2000: physi-

ology and function from multidimensional images.

Proceedings of SPIE, 3978. Bellingham,WA: SPIE;

2000. p. 117–21.

[11] Fleiter T, Merkle EM, Aschoff AJ, Lang G, Stein M,

Gorich J, et al. Comparison of real-time virtual and

fiberoptic bronchoscopy in patients with bronchial car-

cinoma: opportunities and limitations. AJR Am J

Roentgenol 1997;169:1591–5.

[12] Liewald F, Lang G, Fleiter TH, Sokiranski R, Halter G,

Orend KH. Comparison of virtual and fiberoptic bron-

choscopy. Thorac Cardiovasc Surg 1998;46:361–4.

[13] Rapp-Bernhardt U, Welte T, Budinger M, Bernhardt

TM. Comparison of three-dimensional virtual endos-

copy with bronchoscopy in patients with esophageal

carcinoma infiltrating the tracheobronchial tree. Br J

Radiol 1998;71:1271–8.

[14] Rapp-Bernhardt U, Welte T, Doehring W, Kropf S,

Bernhardt TM. Diagnostic potential of virtual bron-

choscopy: advantages in comparison with axial CT

slices, MPR, and MIP? Eur Radiol 2000;10:981–8.

[15] Hoppe H, Walder B, Sonnenschein M, et al. Multi-

detector CT virtual bronchoscopy to grade tracheo-

bronchial stenosis. AJR Am J Roentgenol 2002;178:

1195–200.

[16] Finkelstein SE, Summers RM, Nguyen DM, et al. Vir-

tual bronchoscopy for evaluation of malignant tumors

of the thorax. J Thorac Cardiovasc Surg 2002;123(5):

967–72.

[17] Finkelstein SE, Schrump DS, Nguyen DM, Hewitt

SM, et al. Comparative evaluation of super-high reso-

lution computed tomography and virtual bronchoscopy

for the detection of tracheobronchial malignancies.

Chest 2003(Nov);124(5):1834–40.

[18] LoCicero JR, Costello P, Campos CT, et al. Spiral CT

with multiplanar and three-dimensional reconstructions

accurately predicts tracheobronchial pathology. Ann

Thorac Surg 1996;62:811–7.

[19] Zeiberg AS, Silverman PM, Sessions RB, et al. Helical

(spiral) CT of the upper airway with three-dimensional

imaging: techniques and clinical assessment. AJR Am

J Roentgenol 1996;166:293–9.

[20] Ferretti GR, Vining DJ, Knoplioch J, et al. Tracheo-

bronchial tree: three-dimensional spiral CT with bron-

choscopic perspective. J Comput Assist Tomogr 1996;

20:777–81.

[21] Ferretti GR, Knoplioch J, Bricault I, et al. Central air-

way stenoses: preliminary results of spiral CT gener-

ated virtual bronchoscopy simulations in 29 patients.

Eur J Radiol 1997;7:854–9.

[22] Ferretti GR, Thony F, Bosson JL, et al. Benign abnor-

malities and carcinoid tumors of the central airways:

diagnostic impact of CT bronchography. AJR Am J

Roentgenol 2000;174:1307–13.

[23] Burke AJ, Vining DJ, McGuirt Jr WF, et al. Evaluation

of airway obstruction using virtual endoscopy. Laryn-

goscope 2000;110:23–9.

[24] Remy-Jardin M, Remy J, Artaud D, et al. Volume

rendering of the tracheobronchial tree: clinical evalua-

tion of bronchiographic images. Radiology 1998;208:

761–70.

[25] Ferretti GR, Bricault I, Coulomb M. Helical CT with

multiplanar and three-dimensional reconstruction of

nonneoplastic abnormalities of the trachea. JCAT

2001;25(3):400–6.

[26] Summers RM, Aggarwal NR, Sneller MC, et al. CT

virtual bronchoscopy of the central airways in Wegen-

er’s granulomatous. Chest 2002;121:242–50.

[27] Ferretti GR, Kocier M, Calaque O, Arbib F, Righini C,

Coulomb M, et al. Follow-up after stent insertion in the

tracheobronchial tree: role of helical computed tomog-

raphy in comparison with fiberoptic bronchoscopy. Eur

Radiol 2003;13(5):1172–8.


Chromoendoscopy and magnification endoscopy for

diagnosing esophageal cancer and dysplasia

Michael J. Connor, MDa, Prateek Sharma, MDa,b,*

aDivision of Gastroenterology and Hepatology, University of Kansas Medical Center, 3901 Rainbow Boulevard,

Kansas City, KS 66160, USAbDivision of Gastroenterology, Veterans Affairs Medical Center, 4801 E. Linwood Boulevard, Kansas City, MO 64128, USA

Early detection and classification of esophageal tries, primarily using exfoliative cytology methods,

cancer is an important task for the gastrointestinal

endoscopist. Two primary subtypes of esophageal car-

cinoma are commonly seen in the esophagus: squa-

mous cell carcinoma and adenocarcinoma. The

majority of esophageal malignancies are detected by

endoscopy at a late stage and are therefore cannot be

resected for cure. No obvious, endoscopically visible

premalignant stage exists for squamous cell carci-

noma of the esophagus; however, Barrett’s esophagus

is now recognized as an important risk factor for the

development of esophageal and esophagogastric junc-

tion adenocarcinoma.

Squamous cell carcinoma is the most common

esophageal malignancy in the world. Multiple envi-

ronmental and other factors have been shown to be

important in the pathogenesis of this carcinoma. In

industrialized countries, smoking, heavy alcohol in-

gestion, and achalasia are established risk factors.

Esophageal squamous cell carcinoma has also been

associated with head and neck cancer. Synchronous

or metachronous esophageal squamous cell carci-

noma has been reported in up to 15% of patients

who have head and neck carcinoma [1]. Widespread

screening for squamous cell carcinoma has been

attempted in Far Eastern and South American coun-


doi:10.1016/S1547-4127(04)00042-8

P. Sharma is supported by the Department of Veterans

Affairs Medical Center, Kansas City, Missouri, and the

American Gastroenterological Association (AGA) Castell

Esophageal Award.

* Corresponding author. Division of Gastroenterology,

Veterans Affairs Medical Center, 4801 E. Linwood Boule-

vard, Kansas City, MO 64128.

although the sensitivity and specificity of these tech-

niques are questionable. Identification of a target

population that would benefit from screening in the

United States is an important step in reducing mor-

bidity and mortality caused by this malignancy.

Barrett’s esophagus is defined as columnar-ap-

pearing mucosa of any length within the tubular

esophagus, with the histologic finding of intestinal

metaplasia [2]. The columnar-lined distal esophageal

mucosa can potentially contain three subtypes of

epithelium, including intestinal metaplasia, fundic,

and junctional. It has become clear that intestinal

metaplasia, with the presence of goblet cells by his-

tology, is the predominant premalignant epithelium

associated with dysplasia and adenocarcinoma. Cur-

rently, endoscopy with biopsy remains the gold stan-

dard for diagnosing Barrett’s esophagus. Standard

endoscopic techniques have been shown to be inac-

curate, with biopsies from short segments of colum-

nar-appearing mucosa generally revealing intestinal

metaplasia in only 40% to 60% of patients [3]. When

Barrett’s esophagus has been diagnosed, patients are

advised to enroll in a surveillance program. Current

guidelines suggest obtaining systematic four-quadrant

biopsies at 2 cm intervals from columnar-appearing

mucosa in the distal esophagus for the detection of

dysplasia or cancer [4]. Similar to the distribution of

metaplastic tissue, the presence of dysplasia or early

adenocarcinoma within a segment of Barrett’s esoph-

agus is patchy and focal. Standard endoscopy and

random biopsies might fail to detect these lesions [5].

Foci of unsuspected carcinoma have been found in up

to 73% of resected specimens when esophagectomy

is performed for high-grade dysplasia [6].

s reserved.

M.J. Connor, P. Sharma / Thorac Surg Clin 14 (2004) 87–9488

In squamous cell dysplasia, no visible endoscopic

lesions such as plaques, nodules, or ulcers are seen

regularly. Because of the patchy occurrence of dys-

plastic and cancerous lesions within the esophagus,

the sensitivity of standard biopsy techniques is low.

Because of these limitations, new techniques have

been used in an attempt to maximize the sensitivity

and overall accuracy of endoscopy and biopsy for the

diagnosis of squamous dysplasia, squamous cell

carcinoma, Barrett’s esophagus, and associated dys-

plasia/early cancer. Chromoendoscopy and magnifi-

cation endoscopy stand at the forefront of these

modalities because of their availability, ease of use,

and low cost. This article summarizes the basic chro-

moendoscopic and magnification techniques used for

the detection of metaplastic, dysplastic, and malig-

nant tissue in the esophagus and examines the current

literature regarding this subject.

Chromoendoscopy

Chromoendoscopy employs chemical staining

agents applied to the gastrointestinal mucosa to iden-

tify specific subtypes of epithelia or to highlight

surface characteristics of the epithelium. Chromoen-

doscopy has been used in several regions of the gas-

trointestinal tract including the esophagus, stomach,

duodenum, and colon to aid the characterization of

multiple disease states. Recently, the use of methyl-

ene blue-assisted chromoendoscopy was shown to

increase the yield of detecting dysplasia and cancer

in patients undergoing surveillance colonoscopy for

inflammatory bowel disease [7].

For squamous cell carcinoma, chromoendoscopy

is used to detect metachronous or synchronous le-

sions and to define the extent of dysplasia or cancer.

In the setting of Barrett’s esophagus, chromoendos-

copy is performed to allow targeting of biopsies to

increase the accuracy of detecting intestinal metapla-

sia and dysplasia. Two types of tissue staining are

used in the esophagus. Vital (absorptive) stains such

as Lugol’s solution and methylene blue are taken up

by esophageal mucosa actively. Contrast stains are

not absorbed, but they highlight the surface of the

mucosa, allowing for the identification of minute

lesions and subtle patterns. Contrast stains currently

used in the esophagus include indigo carmine, tolu-

idine blue, and dilute acetic acid solution.

Tissue staining is performed using multiple steps

with the goal of removing surface mucous and other

material before staining, which allows for maximal

contact of the agent with the epithelium. Tissue stains

are typically applied directly onto the mucosal sur-

face during endoscopy using a spray catheter [8].

After the stain is applied, water rinses are performed

to remove excess stain and allow for the most ac-

curate visualization of the mucosa.

Lugol’s solution

Lugol’s solution is an inexpensive, widely avail-

able solution comprising a mixture of iodine and

potassium iodide. This vital stain is absorbed by

glycogen-containing, nonkeratinized squamous epi-

thelium, the normal tissue type in the esophagus.

Lugol’s-stained tissue will characteristically turn

green–brown. The intensity is partly dependent upon

the amount of glycogen present within the epithelium.

This stain is used as a 1% or 2% solution in a volume

of 20 to 50 mL sprayed through endoscopic catheters.

Inflammatory or dysplastic squamous epithelium,

squamous cell carcinoma, and columnar epithelium

will not stain with Lugol’s solution. The most widely

accepted use of Lugol’s solution currently involves

screening for squamous cell carcinoma of the esopha-

gus in high-risk patients and in patients who have

documented squamous cell dysplasia/cancer to rule

out synchronous lesions (Fig. 1A, B).

Many investigators have used Lugol’s solution in

an attempt to identify early, treatable squamous cell

carcinomas of the esophagus. Muto et al used Lugol’s

chromoendoscopy of the esophagus in 389 patients

who had newly diagnosed squamous cell carcinoma

of the head and neck. In this population 54 patients

(14%) had synchronous squamous cell carcinoma of

the esophagus. Fifty-five percent of the patients who

had irregular, multiform regions of Lugol’s-voiding

mucosa had squamous cell carcinoma [1]. Fagunda

et al identified 190 asymptomatic patients who had

multiple risk factors (eg, prior head and neck carci-

noma, alcohol abuse, dietary factors, tobacco use) for

the development of squamous cell carcinoma of the

esophagus, then performed Lugol’s chromoendos-

copy. They found a higher rate of dysplastic mucosa

in biopsies taken from unstained areas than stained

areas, with a sensitivity of 46% and a specificity of

90%; however, the positive predictive value was only

26% [9]. Mori et al applied Lugol’s solution to

24 specimens of resected esophagus and attempted

to grade staining patterns into four types: (1) grade I,

hyperstaining; (2) grade II, normal green–brown

staining; (3) grade III, less intense staining; and (4)

grade IV, unstained. The authors established that

cancers and high-grade dysplasia tended to exhibit

the grade IV pattern, whereas low-grade dysplasia

tended to exhibit the grade III pattern. Margins

between normal squamous mucosa and carcinoma

Fig. 1. (A) Squamous cell carcinoma diagnosed in patient who had recent dysphagia and weight loss. (B) Use of Lugol’s solution

to highlight unstained areas in same patient representing flat dysplastic/cancerous lesions.

M.J. Connor, P. Sharma / Thorac Surg Clin 14 (2004) 87–94 89

tended to be sharp, whereas margins between normal

mucosa and low-grade dysplasia tended to be less

well demarcated [10]. Although some studies have

suggested low accuracy rates for screening, Lugol’s

solution appears to be a simple-to-perform, inexpen-

Fig. 2. Endoscopic picture of distal esophagus stained with

Lugol’s solution, highlighting the squamo–columnar junc-

tion. (From Conner MJ, Sharma P. Chromoendoscopy and

magnification endoscopy in Barrett’s esophagus. Tech

Gastrointest Endosc 2003;5:89–93; with permission.)

sive method of improving the endoscopic detection

and delineation of esophageal squamous cell dyspla-

sia and cancer in high-risk groups and defining the

extent and margin of the tumor in patients who have

known squamous cell cancer.

Because of the stain’s ability to differentiate esoph-

ageal from gastric mucosa, Lugol’s solution can also

be a valuable aid for identifying and highlighting

the squamo–columnar junction (Fig. 2) because

columnar mucosa will not absorb the stain. Stevens

et al used Lugol’s solution with indigo carmine and

35 � magnification endoscopy to identify Barrett’s

esophagus in 13 of 46 patients who had gastroesopha-

geal reflux symptoms. In this study Lugol’s solution

was used to identify the squamo–columnar junction

precisely, allowing for more accurate biopsies [11].

Several investigators have also used Lugol’s solution

to identify islands of residual columnar epithelium

after endoscopic ablation therapy has been performed

in patients who have Barrett’s esophagus [12].

Methylene blue

Methylene blue is a vital stain that is readily taken

up by absorptive epithelium, primarily that of the

small bowel and colon, but is not absorbed by normal

squamous or gastric epithelium. Metaplastic epithe-

lium, including intestinal metaplasia of the stomach

and esophagus, also absorb methylene blue. Methyl-


ene blue has been used successfully to aid in the

identification of gastric intestinal metaplasia and dys-

plasia [13]. Because of these properties, this stain can

be potentially beneficial in the distal esophagus. Be-

fore applying the stain, surface mucous must be re-

moved to expose as much surface area as possible for

staining. N-acetylcysteine solution is generally used

for this purpose. Next, depending on the length of

Barrett’s esophagus, 10 to 20 mL of 0.5% methylene

blue solution is sprayed onto the mucosa. The stained

area is then irrigated vigorously with water. Staining

becomes apparent within 2 to 3 minutes and generally

fades within 15 to 20 minutes (Fig. 3A, B) [14].

Several studies have evaluated the usefulness of

methylene blue staining for the identification of

intestinal metaplasia in the esophagus. Canto et al

compared methylene blue-directed biopsies with ran-

dom biopsies in 43 patients who had Barrett’s esopha-

gus. Intestinal metaplasia was found in 91% of

methylene blue-targeted biopsies versus 69% of ran-

dom biopsies (P = 0.0001). Using methylene blue-

targeted biopsies also enabled the endoscopist to

identify intestinal metaplasia using fewer overall

biopsies per patient (9.5 versus 14.1; P = 0.0001)

[15]. Sharma et al performed methylene blue-guided

target biopsies in 75 patients who had endoscopically

suspected short-segment Barrett’s esophagus. This

group was compared with a control group of 83 pa-

Fig. 3. (A) Short segment of columnar mucosa in the distal esophagu

methylene blue staining within the columnar mucosa after washing t

stained areas revealed intestinal metaplasia. (From Sharma P

chromoendoscopy for detection of short-segment Barrett’s esophagu

tients who had short-segment Barrett’s esophagus

who had undergone standard endoscopic random

biopsies. Intestinal metaplasia was detected in 61%

of the methylene blue group versus 42% of the con-

trol group (P = 0.016), and fewer biopsy specimens

were required in the methylene blue group [16]. This

study highlighted that methylene blue-targeted bi-

opsies might increase the diagnosis of short segments

of intestinal metaplasia in the distal esophagus.

Other studies have not demonstrated a significant

benefit of methylene blue staining in the identifica-

tion of intestinal metaplasia or dysplasia. In a non-

blinded study Dave et al performed methylene blue

staining with biopsies on nine patients who had Bar-

rett’s esophagus. Methylene blue staining was found

to have only 57% sensitivity and 32% specificity for

the detection of specialized intestinal metaplasia.

Furthermore, procedure times were longer and more

patient discomfort was recorded compared with

standard upper endoscopy [17]. Wo et al studied

47 patients who had columnar-lined esophagus in a

prospective, randomized crossover trial. They found

that the sensitivity and specificity of methylene blue

for the detection of specialized intestinal metaplasia

were 53% and 51%, respectively. No significant dif-

ferences were found in the detection of intestinal

metaplasia and dysplasia between methylene blue-

directed and standard biopsy methods [18]. Thus, use

s in the form of multiple tongue-like projections. (B) Areas of

he distal esophagus with water; target biopsies from the blue-

, Topalovski M, Mayo M, Weston A. Methylene blue

s. Gastrointest Endosc 2001;54(3):289–93; with permission.)


of methylene blue in patients who have Barrett’s

esophagus has yielded conflicting results, and its

general use remains controversial.

The use of methylene blue staining in surveillance

protocols to identify dysplasia also is controversial. In

a single-center study, Canto et al were able to diagnose

dysplasia and adenocarcinoma more accurately with

methylene blue-directed biopsies than with ran-

dom biopsies. The authors classified the degree tissue

staining according to pattern, intensity, and heteroge-

neity. High grades of dysplasia stained less intensely

with methylene blue, presumably because of the de-

creased number of goblet cells and the higher nuclear-

to-cytoplasmic ratio. Dysplastic regions also tended to

display a higher degree of stain heterogeneity than

nondysplastic regions [19]. Use of methylene blue in

this situation (ie, for detection of neoplastic lesions)

needs to be studied further.

High-resolution/high-magnification endoscopy

High-resolution imaging improves the ability of

the endoscopist to discriminate between two closely

approximated points. High-resolution endoscopes

provide magnified views of the gastrointestinal tract

with greater mucosal detail. These instruments are

capable of discriminating lesions 10 to 71 microns

apart, compared with the naked eye, which is only

capable of discriminating lesions 125 to 165 microns

apart. The technique of magnification is relatively

simple. A cap is fitted onto the distal tip of the en-

doscope, allowing the mucosa in contact with the cap

to be magnified without the motility of the esophagus

affecting visualization. Magnification is performed by

using a lever located next to the up/down knob of the

endoscope. When the lever is depressed fully, mag-

nification of up to 115 � can be achieved (Olympus

GIF-Q160Z Olympus, Melville, New York) [20].

Use of magnification endoscopy in Barrett’s

esophagus and dysplasia

The combination of chromoendoscopy with mag-

nification endoscopy has been used for more accurate

identification of Barrett’s esophagus and dysplasia.

Endo et al used 80 � magnification endoscopy with

methylene blue staining in 30 patients who had a

columnar-lined distal esophagus. Five discreet stain-

ing patterns were identified: (1) small/round (21 seg-

ments), (2) straight (8 segments); (3) long oval

(26 segments), (4) tubular (10 segments), and (5) vil-

lous (2 segments). The percentage of biopsy speci-

mens containing specialized columnar epithelium

from the long oval, tubular, and villous types were

40%, 100%, and 100%, respectively. Intestinal meta-

plasia was detected infrequently in specimens taken

from mucosa exhibiting the small/round or straight-

type patterns, but specimens from tubular and villous

patterns contained predominantly intestinal-type epi-

thelium [21]. This study showed that specific patterns

(ie, tubular and villous) observed under magnification

might help in identifying intestinal metaplasia.

Indigo carmine is a contrast stain that has been

shown to be useful in the detection and differentia-

tion of colon polyps. It has also been used in conjunc-

tion with magnification endoscopy to identify areas

of intestinal metaplasia and dysplasia within colum-

nar-lined esophageal mucosa. Sharma et al studied

80 patients who had columnar-lined distal esophagus

using indigo carmine dye and 115 � magnification

endoscopy. Three mucosal patterns were identified:

(1) ridged/villous, (2) circular, and (3) irregular/dis-

torted (Fig. 4A–C). Regions exhibiting the ridged/

villous pattern were found to have the highest yield of

intestinal metaplasia (97%) versus regions exhibiting

the circular pattern (17%). Six patients had the

irregular/distorted pattern, and all of these patients

were found to have histologic findings of high-grade

dysplasia. Low-grade dysplasia was detected in

18 patients, all of whom exhibited the ridged/villous

pattern. This technique proved useful for detecting

intestinal metaplasia and high-grade dysplastic le-

sions; however, it was unable to differentiate between

low-grade dysplastic lesions and nondysplastic epi-

thelium [22]. Stevens et al also used indigo carmine

with 35 � magnification endoscopy to identify short

segments of intestinal metaplasia. Identification of a

raised, villiform surface pattern correlated well with

the histologic finding of intestinal metaplasia in 13 of

46 patients who had gastroesophageal reflux dis-

ease [23].

By using magnification endoscopy with a contrast

stain such as indigo carmine, patterns are detected that

might suggest the presence of intestinal metaplasia

or dysplasia. Based on these studies, enhanced mag-

nification endoscopy appears to be a useful surveil-

lance tool for the detection of unsuspected dysplasia

or cancer and for screening for intestinal metaplasia of

the esophagus.

Acetic acid, another contrast agent, has been

studied extensively as an aid in the detection of small

lesions in the uterine cervical mucosa during colpos-

copy. It has recently been used in conjunction with

magnification endoscopy to improve screening for

Barrett’s esophagus. Five to 10 mL of 1.5% acetic

acid solution is sprayed onto the distal esophagus

using a spray catheter. Following application, the

Fig. 4. Three distinct patterns observed under magnification (115 �) after spraying indigo carmine in patients who had Barrett’s

esophagus. (A) Ridged villous. (B) Irregular/distorted. (C) Circular. (From Sharma P, Weston A, Topalovski M, et al. Mag-

nification chromoendoscopy for the detection of intestinal metaplasia and dysplasia in Barrett’s esophagus. Gut 2003;52:24–7;

with permission.)


esophageal and gastric mucosa turn white. Within 2 to

3 minutes the esophagus remains white and the

columnar epithelium turns reddish. Guelrud et al used

acetic acid to improve detection of residual islands

of Barrett’s esophagus after endoscopic ablation ther-

apy in 21 patients. In 11 patients, acetic acid demon-

strated small remnant islands of columnar epithelium

that were not visualized before acetic acid instillation

[24]. The authors later used acetic acid in conjunction

with magnification endoscopy to identify intestinal

metaplasia in 49 patients who had suspected short-

segment Barrett’s esophagus. In this study four mu-

cosal patterns were identified: (1) round, (2) reticular,

(3) villous, and (4) ridged. Mucosa exhibiting the

villous and ridged patterns yielded intestinal meta-

plasia in 87% and 100% of biopsy specimens, re-

spectively [25].

Summary

Based on preliminary reports, the use of chromo-

endoscopy and magnification endoscopy appears to

be a valuable adjunct to standard endoscopy for the

detection and classification of metaplastic and dys-

plastic lesions of the esophagus. Ideally, the use of

this technique would enable the endoscopist to rule in

or out the presence of intestinal metaplasia and


dysplastic/cancerous epithelium by obtaining only a

minimal number of targeted biopsy specimens—or

potentially taking no biopsies at all, which could

transform upper endoscopy into a much more effec-

tive screening and surveillance tool.

There are several problems with the use of chro-

moendoscopy and magnification endoscopy in the

esophagus. This technique is operator-dependent

(ie, dependent on the skill and experience of the en-

doscopist). Studies reporting the accuracy of chromo-

endoscopy remain mixed, especially for Barrett’s

esophagus and dysplasia, which is likely explained

by differences in techniques and materials used in the

investigations. Staining within the esophagus is often

patchy and uneven. Poor spraying technique can

exaggerate irregular uptake by the mucosa. There is

a high false-positive rate when staining gastric-type

epithelium or in the setting of inflammation. Areas of

dysplasia or cancer might take up stain in an irregular

manner or might not stain at all. Magnification only

allows the endoscopist to observe small areas of

mucosa at a time, increasing the overall difficulty of

the procedure and procedure length.

Currently, the greatest body of literature exists

concerning the use of Lugol’s solution for the diag-

nosis of squamous cell dysplasia/carcinoma of the

esophagus and methylene blue for diagnosing Bar-

rett’s esophagus. If used consistently by practicing

physicians, the accuracy of biopsies could be im-

proved. If endoscopic ablative therapy for high-grade

dysplasia and early carcinoma (eg, photodynamic

therapy and endoscopic mucosal resection) becomes

accepted, sensitive methods of detecting residual

metaplastic or dysplastic epithelium after ablation

will be needed to help guide additional endoscopic

therapy. Chromoendoscopy and magnification endos-

copy could prove helpful in this setting.

Further research in this field needs to be per-

formed. As a first step, a uniform classification sys-

tem for staining and magnification patterns should be

devised. Future studies could then be performed

using consistent terminologies. More controlled in-

vestigations with larger numbers of patients must be

performed before tissue staining and magnification

endoscopy become a part of day-to-day endoscopic

practice. Lugol’s chromoendoscopy is a simple tech-

nique for the detection of synchronous squamous

dysplasia and cancer, but a substantial amount of

work remains to be performed for the validation of

chromoendoscopy for the detection of Barrett’s

esophagus and dysplasia. The ultimate aim of chro-

moendoscopy and magnification endoscopy in the

esophagus is to show improved outcomes (ie, early

detection of cancer and improved survival). These

goals have not yet been realized and will require well-

designed studies in the future.

References

[1] Muto M, Hironaka S, Nakane M, Boku N, Ohtsu A,

Yoshida A. Association of multiple Lugol-voiding le-

sions with synchronous and metachronous esophageal

squamous cell carcinoma in patients with head and

neck cancer. Gastrointest Endosc 2002;56(4):517–21.

[2] Sharma P. Barrett’s esophagus: definition and diag-

nosis. In: Sharma P, Sampliner R, editors. Barrett’s

esophagus and esophageal adenocarcinoma. Malden

(MA): Blackwell Science; 2001. p. 1–7.

[3] Eloubeidi M, Homan R, Martz M, Theobald K, Pro-

venzale D. A cost analysis of outpatient care for

patients with Barrett’s esophagus in a managed care

setting. Am J Gastroenterol 1999;94:2033–6.

[4] Sampliner R. Practice guidelines for BE. Am J Gastro-

enterol 1998;93:1028–32.

[5] Cameron A, Carpenter H. Barrett’s esophagus, high-

grade dysplasia, and early adenocarcinoma: a patho-

logical study. Am J Gastroenterol 1997;92:586–91.

[6] Falk G, Rice T, Goldblum J, Richter J. Jumbo biopsy

protocol still misses unsuspected cancer in Barrett’s

esophagus with high-grade dysplasia. Gastrointest

Endosc 1999;49:170–6.

[7] Kiesslich R, Fritsch J, Holtmann M, Koehler H, Stolte

M, Kansler S, et al. Methylene blue-aided chromoen-

doscopy for the detection of intraepithelial neoplasia

and colon cancer in ulcerative colitis. Gastroenterology

2003;124:880–8.

[8] Fennerty MB. Tissue staining (chromoscopy) of the

gastrointestinal tract. Can J Gastroenterol 1999;13(5):

423–9.

[9] Fagunda RB, de Barros SG, Putten AC, Mello ES,

Wagner M, Bassi LA, et al. Occult dysplasia is dis-

closed by Lugol chromoendoscopy in alcoholics at

high risk for squamous cell carcinoma of the esopha-

gus. Endoscopy 1999;31(4):281–5.

[10] Mori M, Adachi Y, Matsushima T, Matsuda H, Sugi-

machi K. Lugol staining pattern and histology of

esophageal lesions. Am J Gastroenterol 1993;88(5):

701–5.

[11] Stevens P, Lightdale C, Green P, Siegel L, Garcia-Car-

rasqullo R, Rotterdam H. Combined magnification en-

doscopy with chromoendoscopy for the evaluation of

Barrett’s esophagus. Gastrointest Endosc 1994;40(6):

747–9.

[12] Overholt BF, Panjehpour M, Haydek JM. Photody-

namic therapy for Barrett’s esophagus: follow-up in

100 patients. Gastrointest Endosc 1999;49:1–7.

[13] Dinis-Ribeiro M, da Costa-Pereira A, Lopes C, Lara-

Santos L, Guilherme M, Moreira-Dias L, et al. Magni-

fication chromoendoscopy for the diagnosis of gastric

intestinal metaplasia and dysplasia. Gastrointest En-

dosc 2003;57(4):498–504.


[14] Fennerty MB. Tissue staining. Gastrointest Endosc

Clin N Am 1994;4(2):297–311.

[15] Canto MI, Setrakian S, Willis J, Chak A, Petras R,

Powe N, et al. Methylene blue-directed biopsies im-

prove detection of intestinal metaplasia and dyspla-

sia in Barrett’s esophagus. Gastrointest Endosc 2000;

51(5):560–7.

[16] Sharma P, Topalovski M, Mayo M, Weston A. Methyl-

ene blue chromoendoscopy for detection of short-seg-

ment Barrett’s esophagus. Gastrointest Endosc 2001;

54(3):289–93.

[17] Dave U, Shousha S, Westaby D. Methylene blue stain-

ing: is it really useful in Barrett’s esophagus? Gastro-

intest Endosc 2001;53(3):333–5.

[18] Wo J, Ray M, Mayfield-Stokes S, Al-Sabbagh G, Geb-

rail F, Slone S, et al. Comparison of methylene blue-

directed biopsies and conventional biopsies in the

detection of intestinal metaplasia and dysplasia in

Barrett’s esophagus: a preliminary study. Gastrointest

Endosc 2001;54(3):294–301.

[19] Canto MI, Setrakian S, Willis J, Chak A, Petras R,

Sivak M. Methylene blue staining of dysplastic and

nondysplastic Barrett’s esophagus: an in vivo and ex

vivo study. Endo 2001;33(5):391–400.

[20] ASGE Technology Committee. ASGE technology sta-

tus evaluation report: high-resolution and high-magni-

fication endoscopy. Gastrointest Endosc 2000;52(6):

864–6.

[21] Endo T, Awakawa T, Takahashi H, Arimura Y, Itoh F,

Yamashita K, et al. Classification of Barrett’s epithe-

lium by magnifying endoscopy. Gastrointest Endosc

2002;55(6):641–7.

[22] Sharma P, Weston A, Topalovski M, Cherian R, Bhat-

tacharrya A, Sampliner RE. Magnification chromo-

endoscopy for the detection of intestinal metaplasia

and dysplasia in Barrett’s esophagus. Gut 2003;52:

24–7.

[23] Stevens P, Lightdale C, Green P, Siegel L, Garcia-Car-

rasquillo R, Rotterdam H. Combined magnification

endoscopy with chromoendoscopy for the evaluation

of Barrett’s esophagus. Gastrointest Endosc 1994;

40(6):747–9.

[24] Guelrud M, Herrera I, Essenfeld H, Castro J. Enhanced

magnification endoscopy: a new technique to identify

specialized intestinal metaplasia in Barrett’s esopha-

gus. Gastrointest Endosc 2001;53(6):559–65.

[25] Guelrud M, Herrera I. Acetic acid improves identifica-

tion of remnant islands of Barrett’s epithelium after

endoscopic therapy. Gastrointest Endosc 1998;47(6):

512–5.


Radionuclide imaging of thoracic malignancies

Stanley J. Goldsmith, MDa,b,*, Lale A. Kostakoglu, MDa,b,Serge Somrov, MDb, Christopher J. Palestro, MDc,d

aWeill Medical College, Cornell University, 1300 York Avenue, New York, NY 10021, USAbDivision of Nuclear Medicine, New York Presbyterian Hospital, Weill Cornell Medical Center, 525 East 68th Street,

New York, NY 10021, USAcAlbert Einstein College of Medicine, Yeshiva University, 1300 Morris Park Avenue, Bronx, NY 10461, USA

dDivision of Nuclear Medicine, Long Island Jewish Medical Center, 270-05 76th Avenue, New Hyde Park, NY 11040, USA

Thoracic masses are usually detected by chest ated over time. Both approaches carry the risk of

radiograph or CT during a screening procedure, as

part of a routine physical examination, or in the

evaluation of some symptom or sign referable to the

thoracic structures such as chest pain, cough, hemop-

tysis, wheezing, or dyspnea. The most common

malignant tumor of the thorax is carcinoma of the

lung, specifically the non–small-cell type, which

includes adenocarcinoma and squamous cell carci-

noma. Other masses requiring different manage-

ment are also encountered, including small-cell lung

carcinoma; bronchial carcinoid (benign and malig-

nant); mediastinal masses, including thymoma, tera-

tomas, lymphomas, and metastases from carcinomas

such as breast, colon, head and neck tumors, thy-

roid carcinoma, and choriocarcinoma. In addition,

carcinoma of the lung might be present as a second

primary in patients known to have one of these

other malignancies.

Traditionally, when a pulmonary mass has been

identified a decision must be made regarding whether

to perform a biopsy or surgical resection to charac-

terize the lesion as a neoplasm versus granuloma or

other inflammatory lesion and to determine a suitable

course of management. In some instances surgical

intervention is deferred and the lesion is reevalu-


doi:10.1016/S1547-4127(04)00034-9

* Corresponding author. Division of Nuclear Medicine,

New York Presbyterian Hospital, Weill Cornell Medical

Center, 525 East 68th Street, New York, NY 10021.


(S.J. Goldsmith).

performing unnecessary surgery with the potential

attendant morbidity or delaying evaluation with the

associated risk of disease progression. This approach

to management of the patient who has a thoracic

lesion is rapidly changing with the development of

nuclear medical imaging procedures that are capable

of characterizing lesions according to their molecular

biology. Radionuclide imaging is based on tissue or

tumor function, metabolism, or other biochemical

characteristics that provide information that is com-

plementary to traditional diagnostic imaging tech-

niques in terms of assessing if a lesion is malignant

or not, and if malignant, determining the extent

of disease.

In recent years radionuclide imaging has made

great progress as a consequence of the development

of novel radiolabeled compounds, which identify

specific molecular processes and remarkable ad-

vances in the instrumentation used for acquisition

and display. Nuclear medicine imaging has pro-

gressed to the point where it can provide crucial

information about lesion biology and can thus play

an integral part in the evaluation and management

of the patient who has a suspected or known

pulmonary malignancy, including noninvasive char-

acterization of the solitary pulmonary nodule, as-

sessment of the extent of disease in the patient

who has a known malignancy, planning and optimiz-

ing radiation therapy, monitoring the response to

treatment, and even predicting prognosis. State-of-

the-art nuclear medicine imaging is clinically effica-

cious and cost-effective, leading to more accurate

s reserved.

S.J. Goldsmith et al / Thorac Surg Clin 14 (2004) 95–11296

diagnoses at less risk and lower cost to the patient

and to society.

Technical advances

Nuclear medicine instrumentation

Nuclear medicine images, or scintigraphs, are

generated by the external detection of emissions from

radioactive isotopes that localize in certain tissues,

organs, physiologic or pathophysiologic processes, or

lesions. In the past, conventional nuclear medicine

images were so-called planar images; data were re-

corded in multiple views: anterior, posterior, lateral,

and oblique. Each image compressed the data ob-

tained from the volume image into two dimensions,

resulting in the loss of object contrast caused by the

presence of background radioactivity (ie, radio-

activity surrounding the object of interest). More

recently, nuclear medicine has evolved toward

tomographic imaging. In recent years clinicians and

radiologists have become familiar with tomographic

images as a result of the broad application of CT

and, more recently, MRI (ie, transaxial slice images

derived from reconstructed transmission data). Data

are recorded in 360� geometry around the patient.

Initially, backprojection techniques were used to

create transverse, or transaxial, images (slices) that

revealed the distribution of radioactivity, or, in the

case of CT, absorption coefficient maps. Tomo-

graphic imaging is a more accurate representation of

the actual distribution of radioactivity in a patient

and results in improved image detail. Tomographic

radionuclide imaging can be performed with single-

photon or positron-emitting radionuclides.

Single-photon emission CT (SPECT) is the tomo-

graphic imaging technology employed with tra-

ditional radionuclides such as 99mTechnetium,67Gallium (67Ga), and 201Thallium (201Tl). SPECT

uses traditional collimated gamma camera position-

ing logic. Data are obtained at small angular inter-

vals as the camera revolves around the patient. A

gamma camera with a single detector must acquire

data over 360�, whereas a device with two or three

detector units requires that each head orbit only a

fraction of the full circumference. Multihead sys-

tems permit greater data acquisition over a shorter

period of time with a resultant improvement in image

quality. Data acquired by the gamma camera are

reconstructed into transaxial planes using sophis-

ticated processing algorithms such as filtered back-

projection and iterative reconstruction. In addition to

the transaxial images, images from the coronal and

sagittal planes are reconstructed readily. Modern

computer capacity also makes it feasible to view

three-dimensional, or volume, images.

Positron emission tomography (PET) is based

upon the unique decay characteristics of positrons.

A positron undergoes annihilation by combining

with a negatively charged electron. As a result of

this annihilation, two 511 keV gamma rays are

emitted 180� apart. Special electronics determine if

two recorded events are coincident, thus identifying

the axis along which the two photons were emitted,

which provides a significant advantage in terms of

reconstructing the position of an event and allowing

for the elimination of cumbersome lead collimators.

In contrast to SPECT, in which single events are

detected, PET makes use of two detector elements

on opposite sides of the subject to detect coincident

photons arising from the annihilation of a positron

and electron. Most PET radiopharmaceuticals have

short half-lives; consequently, until just a few years

ago PET imaging was limited to centers that had

cyclotron production facilities. After numerous

investigational studies confirmed the value and

cost-effectiveness of PET imaging with fluorine-

18-fluorodeoxyglucose (18FDG) in the management

of patients who have tumors, third-party insurers and

eventually governmental agencies approved the tech-

nique for reimbursement. Despite its short (2-hour)

half-life, 18FDG is now available from commercial

sources in most of the United States.

Until recently, PET imaging devices cost more

than $1 million and were available only at larger

centers. The contribution of this technology to patient

management, however, has been so significant that

this situation is changing rapidly. The increased clini-

cal demand for these studies has stimulated develop-

ment of less costly instrumentation, and a spectrum

of devices is now available including a $250,000 to

$350,000 upgrade of conventional dual detector

gamma camera systems, a 360� simultaneous acqui-

sition imaging system that uses six large curvi-

linear sodium iodide crystals (costing approximately

$1.3–1.5 million), and bismuth germanate multi-

crystal, multiring systems (costing $1.7–2.3 million).

Using a phantom in an experimental comparison

of a gamma camera-based coincidence imaging sys-

tem with a dedicated ring detector PET system, the

dedicated PET system identified nodules as small as

6 mm in diameter, whereas the camera-based system

resolved 1 cm and larger lesions [1]. There has been

no direct comparison between imaging with the dedi-

cated ring system and the less expensive devices in the

clinical milieu. A meta-analysis published in 2001

found that the performance of the camera-based

S.J. Goldsmith et al / Thorac Surg Clin 14 (2004) 95–112 97

system was comparable to that of dedicated PET in the

evaluation of lung nodules in patients who had lesions

greater than 1 cm in diameter [2]. Lesions as small as

7 mm in diameter can be detected on the dual detector

coincident camera system, but overall image quality

and lesion detection on a dedicated high-end system

are significantly better. The ability to detect a lesion is

based upon the resolution and sensitivity of the

systems. In this regard the dedicated ring systems will

regularly outperform (ie, show improved detection)

dual detector systems even though coincident dual

detector camera-based systems’ imaging of 18FDG are

frequently useful to characterize lesions greater than

1 cm.

In summary, the dedicated ring detector systems

represent the state-of-the-art in PET imaging with

greater sensitivity for lesion detection. Nevertheless,

dual detector camera-based systems provide access to18FDG imaging, and the positive predictive value is

probably equivalent to that of the more expensive

system. The negative predictive value of the dual

detector system is likely to be somewhat less than

that of the dedicated system because small lesions will

not be detected as a result of volume averaging and

reduced sensitivity.

As with most nuclear studies, PET images suffer

from a paucity of anatomic detail. To maximize the

accuracy of their interpretation, they should be read

together with anatomic cross-sectional studies such

as CT and MR, which has been accomplished by

viewing the studies side-by-side on viewboxes or

computer monitors or through the use of fusion

software that allows direct superimposition of the

images. Recently, instruments have been engineered

that acquire PET and CT images. Patients undergo

sequential PET and CT studies on the same instru-

ment during the same imaging session. Fused PET

and CT images and PET and CT images alone can

be viewed on a slice-by-slice basis. Though costly

and still new, these devices have already demon-

strated that they have advantages in terms of accu-

racy and confidence in interpretation, and they are

likely to eventually replace PET-only and CT-only

devices [3].

Radionuclides

In the past, nuclear medicine assessed thoracic

masses with 67Ga citrate and, more recently, with201Tl and 99mTechnetium (99mTc)-MIBI [4,5]. 67Ga

scintigraphy is positive in inflammatory and neoplas-

tic lesions. Despite this degree of nonspecificity, the

technique was useful but limited in application be-

cause of the comparatively poor resolution achieved

with this radionuclide. Tumor localization of 201Tl

and 99mTc-MIBI is a consequence of perfusion and

rapid extraction of these tracers from tumor tissue.99mTc-MIBI has the advantage of greater photon

flux than 201Tl because a larger dose can be given

because of the shorter (6-hour) half-life. The 140 keV

photon energy is more suitable for imaging than the

lower energy photons of 201Tl. Furthermore, 99mTc-

MIBI binds to intracellular elements, providing im-

proved target to background ratios.

These techniques, however, provide limited im-

provement over CT or MRI in terms of detection of

disease. 99mTc-MIBI could also be used to character-

ize tumor multiple-drug resistance by examining the

retention or washout of 99mTc-MIBI over time because99mTc-MIBI is eliminated from tissue by the same

p51 glycoprotein multiple drug resistance (MDR)

mechanism [6].

Any historical review should include Iodine-131

(131I), which is used to detect metastases from thyroid

carcinoma—even in patients who have a negative

chest radiograph or CT examination (Fig. 1). Thyroid

carcinoma frequently has a subtle micronodular

appearance, although it might occasionally appear

as single or multiple nodules. It is important to

correctly identify lung metastases from thyroid

carcinoma because they respond to radionuclide

therapy with 131I.

Radiolabeled peptides

Radiolabeled compounds that bind to receptors

present in normal and abnormal tissues form the

basis of receptor imaging. Tumor expressing recep-

tors can be visualized with radiolabeled antibodies or

radiolabeled messenger molecules. To date, the most

successful of these agents has been radiolabeled

analogs of regulatory peptides. Regulatory peptides

are small, easily diffuseable, naturally occurring

substances that possess a wide spectrum of recep-

tor-mediated actions. High-affinity receptors for

these peptides are present on many neoplasms.

These receptors offer molecular targets for diagno-

sis and therapy [7]. Currently, two radiolabeled pep-

tides, Octreoscan (Mallinkrodt, St. Louis, Missouri)

and Neotect (Diatide, Londonderry, New Hamp-

shire), both of which are somatostatin analogs, are

approved for diagnostic use in the United States.

Somatostatin is an endogenous neuropeptide that

exists in two forms: a 14 amino acid form and a 28

amino acid form. It is synthesized in the central

nervous system, the hypothalamopituitary axis, the

gastrointestinal tract, the pancreas, and the immune

system. Somatostatin receptors, of which there are

five subtypes, are present on many cells, particularly


those of neuroendocrine origin. These receptors have

also been identified on activated lymphocytes and

the vasa recta of the kidney. All five receptor sub-

types bind to naturally occurring somatostatin with

nearly identical affinity [8,9].

In addition to their presence on normal tissues,

somatostatin receptors are expressed on a wide variety

of human tumors. Three main groups of tumors have

been identified as having the highest density of

somatostatin receptors. Neuroendocrine tumors, in-

cluding islet cell tumors, gastrinomas, pheochromo-

cytomas, paragangliomas, and carcinoid tumors are

one group. Central nervous system tumors such as

astrocytomas and meningiomas represent another

group. The third group of tumors that possess somato-

statin receptors consists of lung carcinoma (small-cell

and non–small-cell), breast tumors, lymphomas, and

renal cell carcinoma.

The short biologic half-life of somatostatin

(f1 min) precludes its use for diagnostic or thera-

peutic purposes, which led to the development of

synthetic somatostatin analogs that had longer bio-

logic half-lives. Octreotide is an eight amino acid

analog with a high affinity for somatostatin subtype

receptors 2 and 5 but a decreased affinity for subtype

3 and no affinity for subtypes 1 and 4 [7]. Octreo-

scan is produced by radiolabeling diethylene tetra

amine penta acetic acid (DTPA)-pentetreotide (a de-

rivative of octreotide) with Indium-111 (111In) and is

used to image somatostatin receptor-bearing tumors.

Extensive studies in large numbers of patients have

shown that somatostatin receptor scintigraphy (SRS)

with 111In DTPA-pentetreotide is most useful in

detecting and staging neuroendocrine tumors [8–10].

In the thorax, SRS is especially useful in small-

cell lung carcinoma and bronchial carcinoid. In

small-cell lung carcinoma, the sensitivity of SRS is

more than 90% for the primary lesion. More than

half of metastatic lesions, however, lose their so-

matostatin receptor expression as a consequence of

dedifferentiation and increasing malignancy [11].

Visualization of metastatic small-cell lung carci-

noma lesions indicates that the tumor is relatively

Fig. 1. Twenty-year-old man post-thyroidectomy for differentiate

the right supraclavicular region. The mass was positive on diagnos

marrow radiation absorbed dose, the patient received 300 mCi of 13

131I uptake in the right supraclavicular mass and demonstrating un

had not been recognized on diagnostic imaging with a lower dose o

a dual detector system with a low-output CT device (GE Milleniu

Wisconsin, USA). Top row: CT images in the coronal, sagittal,

corresponding to CT slices. Bottom row: Fused CT plus 131I imag

negative; on the lower right is the anterior rendering of the 131I vo

Nuclear Medicine, Department of Radiology, New York Presbyteri

well differentiated, whereas nonvisualization is asso-

ciated with dedifferentiation and a poorer progno-

sis. Thus, it is possible, using scintigraphic imaging,

to not only localize lesions but also to determine

prognosis through in vivo tissue characterization.

Bronchial carcinoid is an uncommon neoplasm,

accounting for less than 5% of all lung tumors.

Thought at one time to be benign, this entity is, in

fact, a low-grade, slow-growing, malignant neoplasm

that has the potential for local invasion and distant

metastatic spread (Fig. 2). Several investigators have

reported on the role of SRS in bronchial carcinoid

[12–14]. In a series of 21 patients, SRS revealed

all eight primary lesions at the time of diagnosis,

demonstrated disease in all five patients who had

recurrent or metastatic disease (including two pa-

tients who were asymptomatic at the time of imag-

ing), and identified an increase in tumor size in two

patients who had unresectable disease [13]. In a se-

ries of 31 patients who had bronchial carcinoid,

six patients (nearly 20%) had lesions that were

identified only on SRS. Lesions identified only

with SRS included pulmonary, hepatic, and osseous.

In two patients who had inconclusive CT studies,

SRS correctly excluded recurrent disease. Only

two pulmonary lesions, both in the same patient,

which were detected with other modalities were not

detected with SRS [12].

The implications of the findings in these investi-

gations are important. Although sensitive for the de-

tection of neuroendocrine tumors, SRS cannot be

used for diagnosis because other lung tumors also

express somatostatin receptors. SRS is used to guide

patient management. For example, the exquisite

sensitivity of SRS can determine whether or not,

at the time of diagnosis, curative surgery is possible.

In patients who have recurrent disease, localized

surgical resection has met with some success. The

ability to identify recurrent disease in asymptomatic

patients suggests that SRS might be useful for

identifying individuals who have recurrent disease

when they are still amenable to surgery. This is of

value to determining if metastatic disease is limited to

d thyroid carcinoma was found to have a palpable mass in

tic 131I imaging. Following dosimetry to determine the bone1I. (A) Whole-body scan 1 week post 131I therapy confirming

expected diffuse uptake throughout both lung fields. Uptake

f 131I. (B) SPECT images of the same patient’s thorax using

m Hawkeye, General Electric Medical Systems, Milwaukee,

and transaxial plane. Middle row: 131I tomographic images

es. On the upper right, the scout radiograph of the chest is

lume (all images summed) display. (Courtesy of Division of

an Hospital, Weill Cornell Medical Center, New York, NY).


the liver, because in some cases current surgical

practice makes it possible to consider liver transplan-

tation. Extrahepatic metastatic disease is a contraindi-

cation, however, and SRS is useful for identifying or

excluding patients for this procedure. Finally, deter-

mining the presence or absence of somatostatin recep-

tors with SRS identifies patients who are likely to

respond to medical therapy.

Another radiolabeled somatostatin analog,99mTc-depreotide (Neotect), has been developed.99mTc-depreotide is a synthetic cyclic six amino acid

peptide labeled with technetium-99m and is ap-

proved for the differential diagnosis of the solitary

pulmonary nodule. This agent is a high-affinity

ligand for human somatostatin receptor subtype 3,

with in vitro characteristics that suggest it should

also be useful for imaging the extent of disease in

patients who have non–small-cell and small-cell

lung carcinoma. In a series of 30 patients who had

solitary pulmonary nodules at least 1 cm in diameter

who were at high risk for lung carcinoma but

had indeterminate CT criteria, the sensitivity of99mTc-depreotide for detecting malignancy was

93% (12/13), the specificity was 88% (15/17), and

the accuracy was 90% (27/30). The study was

falsely negative in one patient who had squa-

mous cell carcinoma and falsely positive in two pa-

tients who had necrotizing granulomas [15]. In a

114-patient multicenter trial, the sensitivity, speci-

ficity, and accuracy of 99mTc-depreotide was 97%

(85/88), 73% (19/26), and 91% (104/114), respec-

tively. The three false-negative lesions were adeno-

carcinomas; two were primary lung lesions and

one was thought to be metastatic colon carcinoma.

Six false-positive results were granulomatas; the

seventh was a hamartoma. The data suggest that99mTc-depreotide scintigraphy is a sensitive and

accurate method for the noninvasive evaluation of

the solitary lung nodule that is at least 1 cm in

diameter [16].

An analysis of the cost-effectiveness of 99mTc-

depreotide imaging in 114 patients who had indeter-

minate lung nodules found that in individuals who

Fig. 2. (A) 111In-DTPA-pentetreotide (Octreoscan) SPECT scintigra

carcinoid; status right upper lobe (RUL) resection 30 months earlier

GE Millennium dual detector camera system with Hawkeye con

column: Corresponding 111In images. Right column: Fused images

Note 111In-DTPA-pentetreotide-positive mass in region of right

(Courtesy of Division of Nuclear Medicine, Department of Radiolo

Center, New York, NY). (B) 111In-DTPA-pentetreotide (Octreoscan

old man who had small-cell lung carcinoma. Tumor foci are identi

left anterior cervical triangle. (Courtesy of Division of Nuclear

Hillside Medical Center, New Hyde Park, NY).

had a 50% probability of having a malignancy,

CT alone and CT followed by 99mTc-depreotide

scintigraphy showed an incremental cost-effective-

ness ratio of approximately $11,200 and $8600,

respectively, per year of life saved. Radiograph

follow-up is only cost-effective when the proba-

bility of malignancy is less than 0.14, whereas CT

alone is cost-effective when the probability of ma-

lignancy is 0.71 to 0.90. When the probability of

malignancy is greater than 0.90, thoracotomy is

the best choice. CT plus 99mTc-depreotide is the

most cost-effective strategy, resulting in a savings

of $68 to $1800 for the majority of patients, de-

pending on the risk, when the probability of ma-

lignancy is between 0.14 and 0.71. Based on a

Medicare reimbursement of approximately $900,99mTc-depreotide imaging of pulmonary nodules

that are indeterminate by CT criteria would result

in an annual savings of up to $54 million compared

with selecting patients for thoracotomy based on

CT results alone [17]. Another beneficial aspect of

this approach would be a decrease in the cost and

complications of unnecessary needle biopsies.

Currently, no data are available on the accuracy

of 99mTc-depreotide imaging for evaluating lesions

smaller than 1 cm in diameter, nor on its role in the

staging of lung carcinoma, monitoring response to

therapy, or detecting recurrent disease.

Fluorodeoxyglucose

Fluorodeoxyglucose (FDG) is a structural analog

of 2-deoxyglucose, which, like glucose, is transported

into cells and phosphorylated by a hexokinase to

FDG-6 phosphate. FDG accumulates intracellularly

in proportion to the glycolytic rate of the cell.

FDG uptake by tumor cells is also related to the

presence of increased glucose transporter molecule

expression at the tumor cell surface and to in-

creased levels of hexokinase in these cells. Labeled

with the positron emitter fluorine-18 (18F), FDG is

useful for detecting areas of normal and abnormal

glucose metabolism. Although it is filtered by the

glomerulus, FDG is not reabsorbed in the proxi-

phy in a 67-year-old woman who had a history of pulmonary

with negative follow-up scans. Left column: CT acquired on

figuration (transaxial, coronal, and sagittal slices). Middle

. Extreme right: Scout radiograph and 111In volume display.

hilum superimposed on superior portion of CT density.

gy, New York Presbyterian Hospital, Weill Cornell Medical

) planar scintigraphy of the thorax and abdomen in a 44-year-

fied in the right hilar area, the left paratracheal area, and the

Medicine, Department of Radiology, Long Island Jewish–


mal renal tubules, and the blood concentration of

this compound falls quickly, providing high contrast

between foci of increased glucose metabolism and

background activity within 1 hour of injection.

Many tumors are characterized by increased anaero-

bic glucose metabolism, and 18FDG provides a sen-

sitive tool for their detection. In lung cancer,18FDG-PET imaging provides important infor-

mation about the diagnosis, pretreatment staging,

and assessment of the effects of treatment in this

entity. Its potential role in predicting prognosis is

currently being assessed.

Fluorine-18-fluorodeoxyglucose–positron

emission tomography and lung carcinoma

Nearly 1 million new cases of lung cancer are

diagnosed annually, principally in developed nations.

At the time of diagnosis, the disease has already

spread to adjacent hilar or mediastinal lymph nodes

in about 25% of patients, and 35% to 45% of patients

have distant metastases [18,19]. A systematic ap-

proach to the diagnosis, staging, and treatment of lung

cancer optimizes therapy for each individual patient.

Diagnosis

The diagnosis of lung carcinoma, as for any other

tumor, is the first challenge with which the clinician

is faced when presented with a patient suspected

of having this entity. While morphologic imaging

studies such as planar radiographs, CT, and MRI

can detect a pulmonary lesion, they often cannot

determine whether it is benign or malignant. Only

about one third of pulmonary nodules can be diag-

nosed as benign or malignant on the basis of CT

criteria alone. In the other two thirds, diagnosis

depends on more invasive procedures such as bron-

choscopy and percutaneous CT-guided transthoracic

needle aspiration [20,21]. The overall sensitivity of

bronchoscopy in detecting malignancy is about

65%. If transbronchial biopsy is performed, the

sensitivity approaches 80% [22,23]. The sensitivity

Fig. 3. (A) Fifty-three-year-old woman who had a recently diagn

history of cigarette smoking. The 18FDG-PET images are entirely

five-year-old man who had a history of an right lower lobe (RLL) s

The nodule had increased in size recently. 18FDG-PET images dem

process. There is no evidence of regional lymph node involvement,

(Courtesy of Division of Nuclear Medicine, Department of Radio

ical Center, New York, NY).

of the CT-guided procedure is greater than 90% if

an adequate sample is obtained. The frequency of

sampling errors depends on the size and location of

the lesion and on operator expertise. The most

common complication of needle biopsy is pneu-

mothorax, which occurs in up to10% of patients [24].

The characterization of a pulmonary nodule as

benign or malignant with 18F-FDG-PET was one of

the earliest oncologic applications investigated, and

its value for this purpose is now well established

(Fig. 3). The sensitivity and specificity of 18FDG-

PET imaging in the evaluation of solitary lung nod-

ules ranges from 82% to 100% and 63% to 90%,

respectively [25–34]. A meta-analysis of 1474

pulmonary lesions found that the mean sensitivity

and specificity of 18FDG-PET was 96% and 74%,

respectively [2].

Several factors affect the sensitivity of 18FDG-PET

imaging for the diagnosis of malignancy. Lesion

visualization depends on the amount of 18FDG in-

corporated into the tumor. Abnormalities typically

present as areas of focally increased activity, collo-

quially referred to as hotspots. Images can be ana-

lyzed visually and semiquantitatively. In the chest,

mediastinal blood pool activity is often used as the

reference point. Uptake in a lesion that is more in-

tense than mediastinal blood pool activity is likely to

be malignant, whereas activity equal to or less than

mediastinal blood activity is likely to be benign. It

is also possible to quantify activity by calculating

the standardized uptake value (SUV), which reflects

the ratio of activity per estimated tumor volume

to the total activity administered to the patient, cor-

rected for the lean body mass. Although not abso-

lutely diagnostic, SUVs greater than 2.5 are often

associated with malignancy, and malignant lesions

generally have SUVs greater than 2.5. Fractional18FDG uptake is affected by specific tumor meta-

bolic activity. Consequently, tumors such as bronchi-

oalveolar cell carcinoma and bronchial carcinoid with

relatively low metabolic activity might not concen-

trate sufficient 18FDG to be identified as malignant.

Nevertheless, subsets of these tumor types (bron-

chioalveolar carcinoma and carcinoid or other neuro-

endocrine tumors) might be metabolically active and

identifiable as malignant on 18FDG imaging. Meta-

osed RUL pulmonary nodule. Patient had a 30 pack-year

normal. The patient will continue to be followed. (B) Sixty-

olitary pulmonary nodule that had been followed since 2000.

onstrate a hypermetabolic focus consistent with a malignant

indicating that the patient is an appropriate surgical candidate.

logy, New York Presbyterian Hospital, Weill Cornell Med-


static differentiated thyroid carcinoma can be posi-

tive or negative on 18FDG imaging depending, appar-

ently, on the degree of biologic aggressiveness at the

time of imaging. The degree of tumor aggressiveness

is reflected in the metabolic rate. Although some well-

differentiated adenocarcinomas might demonstrate

only modest accumulation of 18FDG, their SUVs are

nevertheless typically in the malignant range [35].

Sensitivity is also affected by lesion size. Lesions

below the limits of resolution of PET scanners (cur-

rently about 4–8 mm depending upon the system

hardware configuration) might not be detected

[36,37]. The lesion intensity and the measured SUV

will be blunted by the phenomenon known as volume

averaging, in which the absolute uptake in a lesion

below the spatial resolution of the system is distrib-

uted over the minimal resolution area, resulting in an

apparent lowering of the activity per pixel. Sensitivity

is also adversely affected by hyperglycemia. Presum-

ably, competitive inhibition results from elevated

serum glucose levels, reducing 18FDG uptake. In

addition to this direct competitive effect, the insulin

response to the glucose level is greatest in acute

hyperglycemia. This response promotes muscle and

hepatic uptake of glucose and 18FDG. Chronic hyper-

glycemia has a lesser effect on FDG uptake by tumors

[38]. In patients who are diabetic, control of the

disease should be optimized and serum glucose levels

checked before injecting 18FDG. In general, patients

who have serum glucose levels above 250 mg/dL

should probably not undergo 18FDG imaging until

serum glucose levels have been controlled.

Increased glycolysis is not unique to tumors,

however; it occurs in benign conditions such as

granulomas, histoplasmosis, coccidioidomycosis, and

pneumonia, in which false-positive findings are

observed [39– 42]. Some data suggest that the

specificity of the overall results can be improved

by performing dual time point imaging. 18FDG

uptake in tumor tends to increase over time,

whereas inflammation tends to remain constant or

decrease over time [43]. By acquiring a second set

of images about 1 hour after the first set, it might

be possible to distinguish 18FDG uptake in benign

inflammatory conditions from that in tumors.18FDG-PET obviates the need for invasive biopsy

in many patients who have lung nodules. To be used

for this purpose, the test must have a high negative

predictive value, which depends not only on sen-

sitivity and specificity but also on the pretest like-

lihood of malignancy. Using decision analysis

modeling, it has been shown that only patients who

have a 50% or lower pretest likelihood of cancer

should undergo 18FDG-PET imaging. If the pretest

likelihood of malignancy is more than 50%, the

posttest probability of disease will exceed 10%

even if the 18FDG images are negative for one

reason or another (ie, size, metabolic activity, blood

glucose), and histopathologic evaluation will be

necessary regardless of the 18FDG-PET results [44].

Because there is always the risk of a false-negative

result even when the negative predictive value is

high (eg, a negative 18FDG-PET study in a patient

who has <50% pretest probability), patients who

have lung nodules and negative 18FDG-PET studies

should undergo routine clinical and imaging follow-

up every 6 to 12 months (as with other potentially

malignant lesions) to monitor for any increase in

the lesion size.

Staging

Pretreatment staging of non-small cell lung carci-

noma (NSCLC) is necessary to assess prognosis and

to determine appropriate therapy (Figs. 4–6). For

example, patients who do not have mediastinal lymph

node or distant metastatic disease usually undergo

surgical resection of the tumor, whereas patients who

have mediastinal or distant disease can undergo in-

duction chemotherapy or radiotherapy before surgery.

CT imaging is used to anatomically define the extent

of the primary tumor and pleural or chest wall

involvement and is superior to FDG-PET for these

purposes because of its inherently better spatial reso-

lution and delineation of normal structures and ana-

tomic detail. CT identification of hilar and mediastinal

lymph node involvement is less than optimal, how-

ever, because it depends upon lesion size. Using a size

criterion of 1 cm as the threshold for identification of

malignant disease leads to under- and overstaging.

Normal-sized lymph nodes that are infiltrated by

tumor will not be recognized, whereas lymph nodes

that are enlarged secondary to benign processes will

be incorrectly interpreted as containing tumor. The

sensitivity, specificity, and accuracy of mediastinal

staging by CT, as reported in a meta-analysis, is

approximately 60%, 77%, and 65%, respectively

[45]. In a prospective study, the sensitivity and spec-

ificity of CT was 52% and 69%, respectively [46].

Mediastinoscopy has, consequently, been the refer-

ence technique for mediastinal lymph node staging.

The accuracy of 18FDG-PET for assessment of

mediastinal nodal involvement has been investigated

extensively. The sensitivity and specificity of the

procedure, when reported as positive or negative for

the ipsilateral or contralateral side, have ranged form

67% to 92% and 86% to 97%, respectively [47–52].

When analyzed by nodal stations, the reported results

Fig. 4. Selected transaxial slice demonstrating 18FDG-PET images in a 68-year-old woman who smoked 1 pack of cigarettes

per day for many years. She presented to her primary care physician with complaints of back pain but was otherwise in good

health. A chest radiograph revealed a hilar mass and lung nodules. Transbronchial biopsy was positive for poorly differentiated

non–small-cell lung carcinoma. The patient was referred for evaluation of the extent of disease. The so-called hilar mass was

actually the primary lung tumor adjacent to hilar structures with a nearby second and third focus. A metastatic lesion in the

vertebral body was also demonstrated. The accompanying CT image shows multiple tumor masses and evidence of a sclerotic

lesion in the vertebral body (lung CT window). 18FDG-PET indicates the extent of viable tumor. Recently, radiation treatment

plans using intensity modulated radiation therapy (IMRT) were designed to provide booster radiation doses to the well-

circumscribed viable tumor defined by 18FDG-PET as opposed to simply delivering the prescribed dose to the entire CT defined

tumor volume. (Courtesy of Jacqueline Brunetti, MD, Department of Radiology, Holy Name Hospital, Teaneck, NJ).


are similar. A study published in 1999 compared18FDG-PET and CT in 75 patients prospectively

[53]. 18FDG-PET imaging and CT were concordant

in 39 patients, correctly in 35 of the 39 patients but

overstaging in two patients and understaging in

two patients. The results of the two studies were

discordant in 36 patients; 18FDG-PET was correct

in 28 of these patients. Hence, 18FDG-PETwas correct

in 63 of 75 patients, whereas CT was correct in only

43 of 75 patients. In a meta-analysis of staging, the

mean sensitivity and specificity of 18FDG-PET was

79% (F3%) and 91% (F2%) respectively, versus 60%

(F2%) and 77% (F2%), respectively, for CT [45].

The anatomic–functional correlation of 18FDG-

PET and CT images using fusion imaging (in which

the two studies are obtained sequentially on the same

instrument) will undoubtedly further refine the clas-

sification of patients who have nodal or mediastinal

disease by separating the primary tumor from adjacent

lymph nodes, differentiating hilar from adjacent me-

diastinal nodes, and precisely identifying the medias-

tinal lymph node groups involved. It is especially

important to differentiate between N1 and N2 disease

because the former is directly operable and the latter

is not. These conclusions are based upon traditional

methods of staging. The identification of N1 disease

by 18FDG-PET at an earlier time than would have

been possible with CT provides a basis for modifying

surgical resection to include these positive nodes

rather than to conclude that there is no nodal involve-

ment based upon CT imaging alone.

Patients who have distant, or systemic, metastases

at the time of diagnosis cannot be cured by surgery

and are not likely to achieve a long-term remission.

Despite the fact that the incidence of distant recur-

rence after complete removal of the primary tumor

is at least 20%, conventional staging procedures

performed at the time of diagnosis are generally

unrewarding [54]. Because the diagnostic yield of

anatomic imaging is low, 18FDG-PET offers a rapid

method for whole-body imaging that identifies sys-

temic metastatic disease effectively. 18FDG-PET

detects distant disease in up to 15% of patients who

have negative conventional staging procedures

[52,55,56]. In addition to improving the detection

of disease, a negative study can also exclude disease

in patients who have false-positive or equivocal con-

ventional imaging results.

Adrenal masses are identified on CT in up to 20%

of patients who have NSCLC, and 18FDG-PET can


accurately characterize the lesion as benign or malig-

nant (Fig. 5). In one series of 27 patients, 18FDG-PET

was 100% sensitive and 80% specific for adrenal

metastases [57]. The high negative predictive value of

this technique can reduce the need for routine biopsy

of adrenal masses.

Lung carcinoma frequently metastasizes to bone

(Fig. 5). Radionuclide bone scintigraphy using 99mTc-

methylene diphosphonate (99mTc-MDP) had been

considered to be the procedure of choice for the

clinical assessment of possible skeletal involvement.

Bone metastases from NSCLC are often osteolytic,

Fig. 6. 18FDG-PET, CT, and fusion transaxial images in a patient presenting with a chest wall mass. No satellite lesions or lymph

node involvement was demonstrated; biopsy demonstrated chondrosarcoma. (Courtesy of Division of Nuclear Medicine,

Department of Radiology, New York Presbyterian Hospital, Weill Cornell Medical Center, New York, NY).


and 18FDG is reportedly more sensitive than conven-

tional radionuclide bone imaging for this type of

bone lesion. In addition, 18FDG-PET produces fewer

false-positive results in degenerative, inflammatory,

and posttraumatic bone disease [58,59]. False-positive

Fig. 5. (A) 18FDG-PET images from a 62-year-old woman admitted

50 years. A solitary pulmonary nodule on the chest radiograph

Brain metastases were present on MRI. A CT of the chest and ab

the primary pulmonary lesion. Coronal, sagittal, and transaxial 1

mary lesion (arrow 1). A metastatic ipsilateral hilar lymph node i

also seen (arrow 3), although the right hilar node and left adren

fusion images) demonstrating adrenal metastasis in a normal left

Nuclear Medicine, Department of Radiology, New York Presbyteri

18FDG-PET results have been reported with acute

fractures [60].

Liver metastases are readily detected by conven-

tional imaging studies. 18FDG-PET is most useful for

resolving abnormalities that are indeterminate on

with confusion who was a cigarette smoker, 1 pack/day for

was subsequently confirmed as adenocarcinoma on biopsy.

domen to the kidneys was interpreted as normal except for8FDG-PET images are triangulated (crosshairs) on the pri-

s identified (arrow 2), and a metastasis to the left adrenal is

al are normal on CT. (B) Transaxial slices (CT, PET, and

adrenal gland on CT examination. (Courtesy of Division of

an Hospital, Weill Cornell Medical Center, New York, NY).


conventional studies [61]. Although 18FDG-PET can

detect lung metastases, CT has higher resolution and

is less affected by respiratory motion than 18FDG-PET

images. For optimal detection of brain metastases, a

dedicated brain acquisition should be performed. This

additional study is probably not routinely warranted in

light of the low incidence of brain metastases in

asymptomatic patients and because of the excellent

results obtained with contrast-enhanced CT and MRI.

The effectiveness of 18FDG-PET in the staging of

NSCLC is a direct result of its ability to detect me-

tastases that are not apparent on conventional imaging

modalities and to clarify the etiology of indeterminate

lesions found on CT. It has been estimated that18FDG-PET imaging results in changes in patient

Fig. 7. 18FDG-PET, CT, and fusion images from a 73-year-old m

4 years earlier followed by a course of chemotherapy. The patient n

value and suspicion of a mediastinal mass. A hypermetabolic (incr

and a large mass is seen in the liver. These findings are metastatic

second primary neoplasm. (Courtesy of Division of Nuclear Me

Hospital, Weill Cornell Medical Center, New York, NY).

management in 20% to 40% of patients. Perhaps most

important is the exclusion of surgery in up to 15% of

patients as a result of the detection of distant metas-

tases [56,62–64].

Treatment and prognosis

In addition to assisting in the identification of

individuals who are suitable for curative surgery,18FDG-PET is also used for radiotherapy planning

by defining functional tumor volume and providing

an outline of the radiotherapy volume for inclusion of

tumor and sparing of adjacent, uninvolved structures.

In one series, changes in staging were made in 33%

of patients and changes in radiation treatment vol-

an who had a history of colon carcinoma that was resected

ow has an elevated serum carcino embryonic antigen (CEA)

eased 18FDG) mass is seen in the right anterior mediastinum

colon carcinoma. The chest mass is indistinguishable from a

dicine, Department of Radiology, New York Presbyterian

Fig. 8. 18FDG-PET, CT, and fusion images from a 50-year-old HIV+ man demonstrating a hypermetabolic mass in the right

lung and mediastinal lymphadenopathy and infradiaphragmatic disease. Diagnosis: non-Hodgkin’s lymphoma. (Courtesy of

Division of Nuclear Medicine, Department of Radiology, New York Presbyterian Hospital, Weill Cornell Medical Center,

New York, NY).


umes were made in 25% of patients as a direct result

of 18FDG-PET imaging [65]. In addition, 18FDG-PET

differentiates scarring from residual or recurrent

disease accurately. In one study it was more sensi-

tive than, and as specific as, other modalities em-

ployed for this purpose. In a study of 63 patients

suspected of NSCLC relapse, results of 18FDG-PET

and conventional evaluation methods were discordant

in 43 patients. In 39 patients (91%), 18FDG-PET was

correct, resulting in major changes in the diagnosis

in 25 patients (59%) [66]. To maximize the accuracy

of the study, 18FDG-PET should be performed

2 months after surgery and 4 to 6 months after ra-

diotherapy [67].

Although prognosis in NSCLC is determined pri-

marily by disease stage, tumor aggressiveness and

invasiveness—and even metabolic activity—might

also be important factors. Some data indicate that

patients who have more intense uptake of 18FDG

have a shorter survival time. Other data have shown

that patients who have persistent or recurrent abnor-

malities have shorter survival times than patients who

have negative follow-up studies [66,68].

Fluorodeoxyglucose–positron emission

tomography and other thoracic tumors

Increased anaerobic glucose metabolism, which

is the basis for 18FDG identification of carcinoma

of the lung, is a feature of other malignant tumors of

the thorax (Figs. 6–8). Accordingly, identification


of an 18FDG-avid mass does not exclude metastatic

foci from other adenocarcinomas, lymphoma, thyroid

carcinomas, or even active necrotizing granulomas.

The nuclear medicine physician should be provided

with pertinent patient clinical history to be able to

fully assess the likely etiology of the findings on the

PET images. Likewise, the nuclear medicine physi-

cian should evaluate the 18FDG images from the neck

to the mid-thigh to fully assess the extent of disease

and to identify other clinical conditions that might

be present.

Summary

Over the past decade a variety nuclear medicine

imaging studies have become available that are of

considerable value to patients who have pulmonary

malignancies. By far the greatest impact on the man-

agement of patients who have thoracic malignancy

has been the availability of 18FDG-PET imaging. In

the patient who has newly diagnosed lung carcinoma,18FDG-PET improves the accuracy of staging the

disease by identifying or excluding mediastinal dis-

ease and distant metastatic foci. 18FDG-PET is supe-

rior to anatomic methods for evaluating the response

to therapy and for distinguishing recurrent disease

from posttreatment changes. Studies are in progress

to evaluate the role of 18FDG-PET imaging in assess-

ing prognosis.

In patients who have bronchial carcinoid, somato-

statin receptor imaging with 111In-DTPA-pentetreotide

(Octreoscan) can help identify patients who are

candidates for curative surgery, detect unsuspected

metastatic spread, and identify patients who might

benefit from certain types of medical therapy. Al-

though it was initially speculated that 18FDG-PET

imaging would not be sensitive for tumor detection

in patients who have neuroendocrine tumors be-

cause of the usual slow metabolism and biology of

these tumors, many neuroendocrine tumors are

positive on 18FDG-PET imaging. Nevertheless, there

has been no direct comparison of 18FDG-PET imag-

ing and somatostatin receptor imaging, nor does

a positive or negative 18FDG-PET image exclude

neuroendocrine tumor.18FDG-PET imaging and somatostatin receptor

imaging with 99mTc-depreotide (Neotect) are safe,

cost-effective methods that are valuable in the diag-

nosis and management of patients who have sus-

pected or known lung cancer. 18FDG-PET and99mTc-depreotide imaging have a high degree of

sensitivity, specificity, overall accuracy, and positive

and negative predictive values in the evaluation of

the solitary pulmonary nodule. These agents provide

noninvasive, cost-effective methods for selecting

patients for aggressive intervention without contrib-

uting to increased morbidity. Both methods have

incremental value over CT imaging in selecting

patients who have solitary pulmonary nodules for

invasive biopsy or for thoracotomy.

References

[1] Coleman RE, Laymon CM, Turkington TG. FDG

imaging of lung nodules: a phantom study comparing

SPECT, camera-based PET, and dedicated PET. Radi-

ology 1999;210:823–8.

[2] Gould MK, Maclean CC, Kuschner WG, et al. Accu-

racy of positron emission tomography for diagnosis

of pulmonary nodules and mass lesions: a meta-analy-

sis. JAMA 2001;285:914–24.

[3] Lardinois D, Weder W, Hany TF, et al. Staging of non-

small-cell lung cancer with integrated positron-emis-

sion tomography and computed tomography. NEJM

2003;348:2500–7.

[4] Goldsmith SJ, Bar-Shalom R, Kostakoglu L, Weiner R,

Neumann R. Gallium-67 scintigraphy in tumor detec-

tion. In: Sandler M, Coleman R, et al, editors. Diag-

nostic nuclear medicine. Philadelphia: Lippincott

Williams & Wilkins; 2002. p. 911–9.

[5] Waxman AD. Thallium-201 and technetium-99m me-

thixyisobutyl isonitrile (MIBI) in nuclear oncology.

In: Sandler M, Coleman R, et al, editors. Diagnostic

nuclear medicine. Philadelphia: Lippincott Williams &

Wilkins; 2002. p. 931–47.

[6] Kostakoglu L, Goldsmith SJ. Imaging multidrug resist-

ance in hematological malignancies. Hematology 2001;

6:111–24.

[7] Behr TM, Gotthardt M, Barth A, Behe M. Imaging

tumors with peptide-based radioligands. Q J Nucl

Med 2001;45:189–200.

[8] Reubi JC. Regulatory peptide receptors as molecular

targets for cancer diagnosis and therapy. Q J Nucl Med

1997;41:63–70.

[9] Krenning EP, Kwekkeboom DJ, Pauwels S, Kvols LK,

Reubi JC. Somatostain receptor scintigraphy. In: Free-

man LM, editor. Nuclear medicine annual 1995. New

York: Raven Press; 1995. p. 1–50.

[10] Krenning EP, Kwekkeboom DJ, Bakker WH. Somato-

statin receptor scintigraphy with [111In-DTPA-D-Phe1]-

and [123I-Tyr3]-octreotide: the Rotterdam experience

with more than 1000 patients. Eur J Nucl Med 1993;

20:716–31.

[11] Bohuslavizki KH, Brenner W, Gunther M, et al. So-

matostatin receotor scintigraphy in the staging of

small cell lung cancer. Nucl Med Coomun 1996;17:

191–6.

[12] Fanti S, Farsard M, Battista G, et al. Somatostatin

receptor scintigraphy for bronchial carcinoid follow-

up. Clin Nucl Med 2003;28:548–52.


[13] O’Byrne KJ, O’Hare NJ, Freyne PJ, et al. Imaging of

bronchial carcinoid tumors with indium-111 pentetreo-

tide. Thorax 1994;49:284–6.

[14] Musi M, Carbone RG, Bertocchi C, et al. Bronchial

carcinoid tumors: astudy on clinicopathological fea-

tures and role of octreotide scintigraphy. Lung Cancer

1998;22:97–102.

[15] Blum JE, Handmaker H, Rinne NA. The utility of a

somatostatin-type receptor binding peptide in the eval-

uation of solitary pulmonary nodules. Chest 1999;115:

224–32.

[16] Blum JE, Handmaker H, Lister-James J, et al. A multi-

center trial with a somatostatin analog 99mTc-depreotide

in the evaluation of solitary pulmonary nodules. Chest

2000;117:1232–8.

[17] Gambhir SS, Shephard JE, Handmaker H, Blum J.

Analysis of the cost effectiveness of a somatostatin ana-

log-Tc99m-Depreotide (Neotect) in the scintigraphic

evaluation of solitary pulmonary nodules (SPN). J Nucl

Med 1999;40(Suppl):57P.

[18] Boring CC, Squires TS, Tong T. Cancer statistics,

1992. CA Cancer J Clin 1992;42:19–38.

[19] Parker SL, Tong T, Bolden S, Wingo PA. Cancer sta-

tistics, 1997. CA Cancer J Clin 1997;47:5–27.

[20] Ost D, Fein A. Evaluation and management of the soli-

tary pulmonary nodule. Am J Respir Crit Care Med

2000;162:782–7.

[21] Ward HB, Pliego M, Diefenthal HC, Humphrey EW.

The impact of phantom CT scanning on surgery for the

solitary pulmonary nodule. Surgery 1989;106:734–8.

[22] Salathe M, Soler M, Bolliger CT, et al. Transbronchial

needle aspiration in routine fiberoptic bronchoscopy.

Respiration (Herrlisheim) 1992;59:5–8.

[23] Schenk DA, Bower JH, Bryan CL, et al. Transbron-

chial needle aspiration staging of bronchogenic carci-

noma. Am Rev Respir Dis 1986;134:146–8.

[24] Santambrogio L, Nosotti M, Bellaviti N, Pavoni G,

Radice F, Caputo V. CT-guided fine- needle aspiration

cytology of solitary pulmonary nodules: a prospective,

randomized study of immediate cytologic evaluation.

Chest 1997;112:423–5.

[25] Patz EJ, Lowe VJ, Hoffman JM, et al. Focal pulmonary

abnormalities: evaluation with F-18 fluorodeoxyglu-

cose PET scanning. Radiology 1993;188:487–90.

[26] Lowe VJ, Fletcher JW, Gobar L, et al. Prospective

Investigation of positron emission tomography in lung

nodules. J Clin Oncol 1998;16:1075–84.

[27] Kubota K, Matsuzawa T, Fijiwara T, et al. Differential

diagnosis of lung tumor with positron emission tomog-

raphy: a prospective study. J Nucl Med 1990;31:

1927–32.

[28] Gupta NC, Maloof J, Gunel E. Probability of malig-

nancy in solitary pulmonary nodules using fluorine-18-

FDG and PET. J Nucl Med 1996;37:943–8.

[29] Patz Jr EF, Lowe VJ, Hoffman JM, et al. Persistent

or recurrent bronchogenic carcinoma: detection with

PET and 2-[F-18]-deoxy-D-glucose. Radiology 1994;

191:379–82.

[30] Duhaylongsod FG, Lowe VJ, Patz Jr EF, et al. Detec-

tion of primary and recurrent lung cancer by means of

F-18 fluorodeoxyglucose positron emission tomogra-

phy (FDG PET). J Thorac Cardiovasc Surg 1995;

110:130–9.

[31] Bury T, Dowlati A, Paulus P, et al. Evaluation of the

solitary pulmonary nodule by positron emission to-

mography imaging. Eur Respir J 1996;9:410–4.

[32] Sazon DA, Santiago SM, Soo Hoo GW, et al. Fluoro-

deoxyglucose-positron emission tomography in the

detection and staging of lung cancer. Am J Respir Crit

Care Med 1996;153:417–21.

[33] Prauer HW, Weber WA, Romer W, et al. Controlled

prospective study of positron emission tomography

using the glucose analogue [18f] fluorodeoxyglucose

in the evaluation of pulmonary nodules. Br J Surg

1998;85:1506–11.

[34] Bury T, Rigo P. Contribution of positron emission

tomography for the management of lung cancer. Rev

Pneumol Clin 2000;56:1506–11.

[35] Higashi K, Ueda Y, Seki H, et al. Fluorine-18-FDG

PET imaging is negative in bronchiolo-alveolar lung

carcinoma. J Nucl Med 1998;39:1016–20.

[36] Farquar TH, Llacer J, Sayre J, et al. ROC and LROC

analyses of the effects of lesion contrast, size, and sig-

nal-to-noise ratio on detectability in PET images. J Nucl

Med 2000;41:745–54.

[37] Hoffman EJ, Huang SC, Phelps ME. Quantitation

in positron emission computed tomography: 1. Ef-

fect of object size. J Comput Assist Tomogr 1979;3:

299–308.

[38] Wahl RL. Targeting glucose transporters for tumor

imaging: ‘‘sweet’’ idea, ‘‘sour’’ result [editorial]. J Nucl

Med 1996;37:1038–41.

[39] Roberts PF, Follette DM, von Haag D, et al. Factors

associated with false positive staging of lung cancer by

positron emission tomography. Ann Thorac Surg 2000;

70:1154–9.

[40] Kapucu LO, Meltzer CC, Townsend DW, et al. Fluo-

rine-18-fluorodeoxyglucose uptake in pneumonia.

J Nucl Med 1998;39:1267–9.

[41] Brudin LH, Valind SO, Rhodes CG, et al. Fluorine-18-

deoxyglucose uptake in sacroidosis measured with

positron emission tomography. Eur J Nucl Med 1994;

21:297–305.

[42] Bakheet SM, Saleem M, Powe J, et al. F-18 fluoro-

deoxyglucose chest uptake in lung inflammation and

infection. Clin Nucl Med 2000;25:273–8.

[43] Matthies A, Hickeson M, Cuchiara A, Alavi A. Dual

time point 18F-FDG PET for the evaluation of pulmo-

nary nodules. J Nucl Med 2002;43:871–5.

[44] Gambhir SS, Sheperd JE, Shah BD, et al. Analytical

decision model for the cost-effective management

of solitary pulmonary nodules. J Clin Oncol 1998;16:

2113–25.

[45] Dwamena BA, Sonnad SS, Angobaldo JO, et al. Metas-

tases from non-small cell lung carcinoma: mediastinal

staging in the 1990s—meta-analytic comparison of

PET and CT. Radiology 1999;213:530–6.

[46] Webb WR, Gatsonis C, Zerhouni EA, et al. CT and


MR imaging in staging non-small cell bronchogenic

carcinoma: report of the Radiologic Diagnostic Oncol-

ogy Group. Radiology 1991;178:705–13.

[47] Steinert HC, Hauser M, Allemann F, et al. Non-small

cell lung cancer: nodal staging with FDG PET versus

CT with correlative lymph node mapping and sam-

pling. Radiology 1997;202:441–6.

[48] Vansteenkiste JF, Stoobants SG, De Leyn PR, et al.

Mediastinal lymph node staging with FDG PET scan

in patients with potentially operable non-small cell

lung cancer: a prospective analysis of 50 cases. Leuven

Lung Cancer Group. Chest 1997;112:1480–6.

[49] Saunders CA, Dussek JE, O’Doherty MJ, et al. Eval-

uation of fluorine-18 fluorodeoxyglucose whole body

positron emission tomography imaging in the staging

of lung cancer. Ann Thorac Surg 1999;67:790–7.

[50] Kerstine KH, Stanford W, Mullan BF, et al. PET, CT,

and MRI with Combidex for mediastinal staging in

non-small cell lung cancinoma. Ann Thorac Surg

1999;68:1022–8.

[51] Weng E, Tran L, Rege S, et al. Accuracy and clinical

impact of mediastinal lymph node staging with FDG

PET imaging in potentially resectable lung cancer. Am

J Clin Oncol 2000;23:47–52.

[52] Pieterman RM, van Putten JW, Meuzelaar JJ, et al.

Preoperative staging of non-small cell lung cancer with

positron-emission tomography. N Engl J Med 2000;

343:254–6.

[53] Bury T, Paulus P, Dowlati A, et al. Staging of the me-

diastinum: value of positron emission tomography

imaging in non-small cell lung cancer. Eur Respir J

1996;9:2560–4.

[54] Martini N, Bains MS, Burt ME, et al. Incidence of local

recurrence and second primary tumors in resected stage

1 lung cancer. J Thorac Cardiovasc Surg 1995;109:

120–9.

[55] Weder W, Schmid RA, Bruchhaus H, et al. detection of

extrathoracic metastases by positron emission tomogra-

phy in lung cancer. Ann Thorac Surg 1998;66:886–92.

[56] Bury T, Dowlati A, Paulus P, et al. Staging of non-

small cell lung cancer by whole- body fluorine-18 de-

oxyglucose positron emission tomography. Eur J Nucl

Med 1996;23:204–6.

[57] Erasmus JJ, Patz Jr EF, McAdams HP, et al. Evaluation

of adrenal masses in patients with bronchogenic carci-

noma using 18F-fluorodeoxyglucose positron emission

tomography. Am J Roentgenol 1997;168:1357–60.

[58] Bury T, Barreto A, Daenen F, et al. Fluorine-18 deox-

yglucose positron emission tomography for the detec-

tion of bone metastases in patients with non-small cell

lung cancer. Eur J Nucl Med 1998;25:1244–7.

[59] Schmirreister H, Guhlmann A, Eisner K, et al. Sensi-

tivity in detecting osseous lesions depends on anatomic

localization: planar bone scintigraphy versus 18F PET.

J Nucl Med 1999;40:1623–9.

[60] Meyer M, Gast T, Raja S, Hubner K. Increased F-18-

FDG accumulation in an acute fracture. Clin Nucl Med

1994;19(1):13–4.

[61] Hustinx R, Paulus P, Jacquet N, et al. Clinical evalua-

tion of whole-body 18F-fluorodeoxyglucose positron

emission tomography in the detection of liver metasta-

ses. Ann Oncol 1998;9:397–401.

[62] Valk PE, Pounds TR, Hopkins DM, et al. Staging non-

small cell lung cancer by whole-body positron tomo-

graphic imaging. Ann Thorac Surg 1995;60:1573–81.

[63] Marom EM, McAdams HP, Erasmus JJ, et al. Staging

non-small cell lung cancer with whole-body PET. Ra-

diology 1999;212:803–9.

[64] Lewis P, Griffin S, Marsden P, et al. Whole-body 18F-

fluorodeoxyglucose positron emission tomo-graphy in

preoperative evaluation of lung cancer. Lancet 1994;

334:1265–6.

[65] Hicks RJ, Kalff V, MacManus MP, et al. 18F-FDG PET

provides high-impact and powerful prognostic stratifi-

cation in staging newly diagnosed non-small cell lung

cancer. J Nucl Med 2001;42:1596–604.

[66] Hicks RJ, Kalff V, MacManus MP, et al. The utility of18F-FDG PET for suspected recurrent non-small cell

lung cancer after potentially curative therapy: impact

on management and prognostic stratification. J Nucl

Med 2001;42:1605–13.

[67] Erasmus JJ, Patz Jr EF. Positron emission tomogra-

phy imaging in the thorax. Clin Chest Med 1999;20:

715–24.

[68] Patz Jr EF, Connolly J, Herndon J. Prognostic value of

thoracic FDG PET imaging after treatment for non-

small cell lung cancer. Am J Roentgenol 2000;174:

769–74.


Imaging of acute pulmonary emboli

Arfa Khan, MD, FACRa,b,*, Aaron Darius Cann, PhD, MDc,Rakesh D. Shah, MD, FCCPc,d

aAlbert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461-1602, USAbThoracic Radiology, Long Island Jewish Hospital, 270-05 76th Avenue, New Hyde Park, NY 11040, USA

cDepartment of Radiology, North Shore University Hospital, 300 Community Drive, Manhasset, NY 10461-1602, USAdDepartment of Radiology, New York University School of Medicine, 550 First Avenue, New York, NY 10016, USA

Pulmonary embolism (PE) is a significant cause The physiology of PE forms the basis for its de-

of surgical morbidity and mortality after surgical

procedures. Venous stasis caused by immobilization,

endothelial damage, and malignancy is a physiologic

factor that predisposes to thromboembolism [1] and is

common in the surgical patient. Because clinical signs

and symptoms such as chest pain, dyspnea, and

tachycardia are notoriously nonspecific, radiologic

imaging is the mainstay of diagnosis. In this article

the authors discuss the various methods of imaging

PEs in surgical patients.

The incidence of PE in surgical patients is high

and occurs throughout the spectrum of surgical

patients. In one large series of trauma patients, the

development of PE was found to confer an overall

tenfold increase in mortality, to 26% [2]. Perhaps the

best numerical description of the importance of PE to

surgeons is found in the extensive pathology series of

Lindblad, who found that 31.7% of all surgical pa-

tient autopsies from 1981 to 1988 had PE. Twenty-

nine percent of these autopsy-proven emboli were

considered to be fatal [3]. Other studies of surgical

mortality in inpatients show similar results [4]. Mod-

ern laparoscopic procedures also carry a risk of fatal

PE [5]. Pulmonary embolization has even been

reported during such minimally invasive procedures

as percutaneous discoplasty [6–8]. It is therefore

important to maintain a high index of suspicion in

all surgical patients.

1547-4127 /04/$ – see front matter D 2004 Elsevier Inc. All righ

doi:10.1016/S1547-4127(04)00032-5

* Corresponding author. Thoracic Radiology, Long

Island Jewish Hospital, 270-05 76th Avenue, New Hyde

Park, NY 11040.

E-mail address: [email protected] (A. Khan).

tection and has been reviewed extensively [1]. PEs

usually begin as thrombosis of the calf veins. They

typically propagate to the deep venous system of the

leg and thigh (popliteal vein, superficial femoral

veins, or common femoral veins), although in recent

years an increasing number of catheter-related ve-

nous thromboses have been seen that originated at

sites of central venous catheter placements. When in

the deep venous system, thrombi can dislodge or

fragment and travel to the lungs. DVTs are usually

asymptomatic [9] and are therefore usually not sus-

pected before PE [10]. DVT or PE can be imaged at

any of these stages: as DVT in the legs by way of ul-

trasonography or venography, directly in the pulmo-

nary arterial tree by way of conventional angiography

(CA), CT pulmonary angiography (CTPA), or MR

angiography (MRPA), or by way of its end-effects on

pulmonary perfusion and ventilation (V/Q scanning)

or the lung parenchyma by way of chest radiograph

(CXR). Large central emboli can even be identified

by transesophageal echocardiography (TEE). The

availability of so many different tests, each with its

own strengths and weaknesses, can be somewhat

perplexing. In this article the authors attempt to

provide a framework for the diagnosis of PE.

Chest radiograph

A plain CXR is an essential part of early diagnostic

investigation because it has a valuable role in the

exclusion of alternative pathology. By itself it is of

ts reserved.

Fig. 1. CXR of a patient who had known PE. Right lower

lobe pleural-based opacities (Hampton’s hump) represent

infarcts (arrows). Note dilated pulmonary arteries and

bilateral small pleural effusions.

A. Khan et al / Thorac Surg Clin 14 (2004) 113–124114

limited value in the diagnosis of PE because of poor

sensitivity and specificity.

The Prospective Investigation of Pulmonary Em-

bolism Diagnosis (PIOPED) study was a landmark in

the diagnosis of PE, although it was done before the

era of CTPA or MRPA. The most common radio-

Fig. 2. V/Q scan with high probability of pulmonary embolus—bi

ventilation. PE was confirmed by CTPA (see Fig. 6).

graphic abnormality is atelectasis, although this find-

ing was equally prevalent in patients who did not

have PE. Other signs of PE include enlargement of

the main pulmonary artery, pleural effusion (usually

small and often unilateral), regional oligemia, and

elevated hemidiaphragm (indicating volume loss).

The most specific sign was found to be a Hampton’s

hump, which is an uncommon 3 to 5 cm, pleural-

based, pyramid-shaped opacity that usually indicates

pulmonary infarction (Fig. 1). Thus, the radiographic

signs of PE are highly nonspecific (eg, atelectasis or

pleural effusion) or even absent altogether. In this

study 12% of 383 patients who had PE had normal

CXRs [11].

Thus, the main role of the CXR is to exclude

obviously unrelated causes of similar symptoms such

as pneumothorax, displaced endotracheal tube (ET)

tube, mucous plugs, and so forth. A high-quality

postero-anterior (PA) and lateral study is always

preferred when possible. The CXR also stratifies the

patient’s potential suitability for a V/Q scan.

Ventilation–perfusion scintigraphy

Until recently, V/Q scans were used extensively as

the primary imaging method for evaluation of sus-

pected PE. In this test, radiolabeled albumin aggre-

lateral multiple perfusion defects, all of which mismatch on

A. Khan et al / Thorac Surg Clin 14 (2004) 113–124 115

gates are injected and carried to capillary beds in the

lung, where they lodge. Their absence from a particu-

lar portion of lung suggests that the pulmonary artery

branch to that region might be occluded. Similarly,

images of an inhaled radioactive gas provide ventila-

tion imaging, giving the interpreting physician a view

of abnormally ventilated lung regions (Fig. 2). Clas-

sically, a PE will manifest as a V/Q mismatch with a

segmental region of low perfusion but normal venti-

lation. A normal or near-normal V/Q scan by itself has

a high negative predictive value (NPV), essentially

excluding PE (<5% probability); a high probability

scan is also widely regarded as diagnostic ( > 90%

positive predictive value). It is also widely available,

and most radiologists have had extensive experience

with it. A current CXR is required for interpretation of

a V/Q scan; however, V/Q scanning is most often

nondiagnostic (in 73% of patients in the PIOPED

study) [11]. As many as 90% of patients who have

underlying lung disease have neither normal nor high

probability studies. An indeterminate V/Q scan is

nondiagnostic (probability 10–90%). A low proba-

bility V/Q scan does not rule out PE. In the PIOPED

study, 14% of patients who had low probability V/Q

scans had angiographic evidence of PE. Moreover, not

every surgical patient can cooperate with the ventila-

tion portion of the study, and critically ill patients

might not be well served by the 1 hour or more of

imaging in the nuclear medicine department that is

needed for the test.

AV/Q scan is an appropriate first test for evaluat-

ing a patient who has suspected PE only when the

baseline CXR is normal and the pretest clinical sus-

picion is low or moderate. In these patients a normal

result will frequently be obtained without the use of

iodinated contrast or the somewhat higher radiation

dose of CTPA; however, in many patients—especially

when the baseline CXR is abnormal or there is history

of significant underlying pulmonary disease—a V/Q

scan might not provide the necessary information to

diagnose or exclude PE.

Fig. 3. Combined CTPA–CT venography at groin level

shows a thrombus in the right femoral vein (arrow).

Conventional angiography

CA gives nondiagnostic results in only 3% of pa-

tients and it has been shown to have a 99.4% NPV by

clinical follow-up; however, it is invasive and carries

a 0.5% mortality rate associated with the study itself,

most commonly in intensive care unit patients [12].

The traditional reliance on CA as the gold standard

has recently been questioned because of its inability

to detect subsegmental emboli consistently. Interob-

server agreement for diagnosing subsegmental em-

boli using the supposed gold standard, pulmonary

angiography, is about 66% [11].

Venous imaging

Deep vein thrombosis (DVT) and PE are separate

manifestations of the same disease process. DVT can

be detected in 50% of patients who have angiographi-

cally proven PE. Because the treatment of these

conditions is similar, the presence of DVT justifies

anticoagulant therapy and therefore obviates a search

for a pulmonary artery clot [13]. This test does not

directly detect PE itself; the source of PE can be

identified pathologically in the lower extremity ve-

nous tree in only 59.4% of patients [14]. In patients

suspected of having PE, only 29% of duplex ultraso-

nography (US) will be abnormal at the time of pre-

sentation [15]. At autopsy, no thromboembolic source

could be detected in 28% of patients who died of PE,

suggesting complete dislodgement of thrombus from

an unknown source [14]. A non-lower extremity or

completely dislodged embolus would lead to a nega-

tive lower extremity venous study despite the pres-

ence of PE of any size.

Duplex US [16] is the imaging method of choice

for evaluating DVT. It has a sensitivity of 91% and

specificity of 99% [17]. The sensitivity for diagnosis

of femoral DVT approaches 100%. Duplex scans are

less sensitive for isolated calf vein thrombosis and will

not detect iliac vein thrombosis. Duplex US should

always be done in both legs in patients who have sus-

pected DVT because of the high incidence of asymp-

Fig. 5. Bilateral central PE. CTPA at the level of right

pulmonary trunk shows filling defects in right main and left

descending arteries (arrows).


tomatic DVT in the contralateral leg, even when the

ipsilateral leg has no DVT by duplex US [18].

At some institutions, cross-sectional imaging of

the venous system is performed (Fig. 3). More com-

monly, indirect CT venography is performed imme-

diately after CTPA using venous enhancement from

the pulmonary artery contrast bolus itself. Loud and

colleagues [19] demonstrated a sensitivity of 99%

and a specificity of 100% for femoropopliteal DVT

using CT venography. In another study comparing

indirect CT venography to US, all 15 cases of DVT

identified on US were detected on CT, plus four ad-

ditional cases not identified on US [20]. Other studies

have also shown excellent results [21]. Moreover,

these methods can study the iliac system, most of

which is inaccessible to US. Indirect techniques

represent one-step imaging for PE and DVT and

only require a few extra minutes of imaging time [22].

The former gold standard test for lower extremity

DVT, conventional venography, is an invasive proce-

dure and is now rarely used for the primary evaluation

of DVT. The many potential complications of venog-

raphy include development of DVT.

CT pulmonary angiography

With the introduction of spiral CT scans in the

early 1990s, and then with the introduction of multi-

detector scanners in the late 1990s, it has become

possible to image the entire chest in a short time and in

a single breath-hold. CTPA visualizes PE directly as

filling defects within contrast-opacified pulmonary

arteries. Unlike other techniques for visualizing PE,

it also provides an excellent study of the lung paren-

Fig. 4. Massive central PE. CTPA at the level of main

pulmonary artery shows a large filling defect extending to

the left and right pulmonary trunks.

chyma and pleura. Like CA, the test involves the use

of a moderate radiation dose and the exposure to

iodinated contrast media, but it does not require

invasive catheterization; however, the technique can

be quite sensitive to respiratory motion during imag-

ing, often an issue with patients who are dyspneic

or ventilated.

The first major comparison of CTPA to the gold

standard of CA sparked tremendous interest in the

technique [23]. Remy-Jardin and colleagues com-

pared spiral CTPA to CA in 42 patients. In the

Fig. 6. Bilateral lobar and segmental PE. CTPA shows filling

defects in right middle lobe artery and multiple lower lobe

segmental arteries (arrows). For V/Q scan on the same

patient see Fig. 2.

Fig. 7. Segmental PE. CTPA at the level of lower lobes

shows a filling defect in right posterior segmental ar-

tery (arrow).

Fig. 8. Isolated subsegmental PE. CTPA at the level of lung

bases shows a filling defect in a small subsegmental artery

of medial basal segment of right lower lobe (arrow).


19 patients who showed central PE, 18 were con-

firmed by pulmonary angiography, with an overall

sensitivity of 100% and specificity of 96%. In 1993

Teigen et al [24] used electron beam CT to evalu-

ate PE. They studied 86 patients and found similar

results, with a sensitivity of 95% and specificity of

80%. These early reports showed the ability of CT to

demonstrate emboli in the main, lobar, and segmental

branches of the pulmonary arteries (Figs. 4–8, re-

spectively); however, the accuracy of detecting sub-

segmental clots was considerably low. In 1995

Goodman et al [25] found that the sensitivity and

specificity for detecting thrombus in the central ves-

sels using helical CT were 86% and 92%, respective-

ly; however, when subsegmental arteries were

included the sensitivity was only 63%. Other studies

have found similar results [26].

Although spiral CT is quite accurate in detecting

more central PEs, it has demonstrated limited value in

the diagnosis of subsegmental emboli. The preva-

lence, detection, and significance of these emboli are

controversial. In a prospective study that included

130 patients who had PE, 22% of patients had no

larger than a subsegmental clot [27]. Other studies

showed a 5.9% [28] or a 30% [29] prevalence of PE

limited to subsegmental vessels. Baile et al found no

difference between spiral CT and pulmonary angiog-

raphy for detection of subsegmental PEs when they

injected methacrylate beads in pigs [30]. Rapid tech-

nical advances in CT techniques and machinery are

increasing the detectability of smaller clots. Using

1.25 mm CT sections, one group found that 94% and

74% of subsegmental fourth order and fifth order

vessels, respectively, could be evaluated adequately

[31]. Other studies have shown benefit from thinner

sections [32] and multidetector machinery [33]. These

techniques and equipment are rapidly propagating

through radiology departments.

Even after diagnosis of isolated subsegmental

emboli is established (Fig. 8), the clinical significance

of the finding is unclear. Patients who have limited

cardiopulmonary reserve might be at increased risk

from even a small PE. Also, a small PE can be

significant when it is a sentinel event preceding a

larger embolus (ie, when there is a large residual

DVT burden at the originating site of the embolus).

In patients who have PE there is a correlation be-

tween patient outcome and residual clot burden at US.

The practical concern is that missed subsegmental

emboli could result in a poor outcome in patients who

have false-negative CTPAwho are not anticoagulated.

This issue has been studied extensively with out-

come-based studies of patients not anticoagulated

after a negative CTPA. These reports, summarized

in Table 1, have consistently found an NPVof greater

than 94% when measured against clinical follow-up in

the absence of anticoagulation. Most of these studies

have concluded that terminating the imaging sequence

after an adequate negative CTPA appears to be safe

[34–39], although some of these patients also had

negative Doppler studies. Only one large study did not

agree with these conclusions, at least for high-risk

patients, although they also had an NPV of at least

94.7% [40]. This same study also found that in a

subset of 12 patients who had negative Dopplers and

isolated subsegmental PE on CTPA, nine patients had

negative V/Q scans or CA and did well clinically

without anticoagulation [40]. Finally, a large prospec-

tive comparison study found that the NPV of a

negative CTPA (99% in that study) was statistically

Table 1

Studies withholding anticoagulation after negative CT pulmonary angiography

No. patients NPV (%) Clinical follow-up (mo) Other patient qualifications Ref.

100 100 6 [34]

71 96 6 Nondiagnostic V/Q scans [35]

215 98.6 3 [36]

993 99.5 3 Patients studied by electron beam CT,

not helical CT; retrospective

[37]

81 95.1–97.5 21 (avg) Negative Dopplers [38]

198 99 3 [39]

507 98.2 3 Negative Dopplers [40]

Low or intermediate clinical suspicion

75 94.7 0a Negative Dopplers [40]

High clinical suspicion [41]

a These patients were studied immediately by V/Q or CA.


similar to a normal V/Q study (100% NPV) [39]. The

large preponderance of evidence suggests that with-

holding anticoagulant therapy after negative CTPA

appears to be safe. In retrospect this might not be

surprising given the excellent NPV of CA itself

despite its poor interobserver agreement regarding

subsegmental clots. CA should continue to be used

for the approximately 10% of CTPA studies that

are nondiagnostic.

Another advantage of CTPA over other studies is

that it provides excellent evaluation of secondary

Fig. 9. Infarction versus atelectasis. CTPA at the level of

lung bases in a patient who had PE shows a pleural-based,

nonenhancing opacity in right lower lobe (infarct) and an

enhancing opacity in the left lower lobe (atelectasis).

signs of PE such as infarction, pulmonary artery

dilatation, atelectasis, and pleural effusion. Infarction

can be differentiated from atelectasis by its lack of

enhancement (Fig. 9).

In addition to its usefulness for diagnosing or ex-

cluding PE, CT gives unparalleled evaluation of the

lung parenchyma. Studies of CTPA reported alterna-

tive diagnoses by CTPA in 39% to 67% of patients

who did not have PE [41,42]. The alternative diag-

noses found in these studies included pneumonia,

cardiac or pericardiac disease, interstitial lung disease,

Fig. 10. Alternate diagnosis: lung cancer. Sixty-five-year-old

man who had shortness of breath and hemoptysis. V/Q scan

showed high probability for PE. CTPA showed a left hilar

mass and mediastinal adenopathy with no evidence of PE.

Fig. 11. Alternate diagnosis: pulmonary artery angiosar-

coma. Sixty-eight-year-old woman who had shortness of

breath and cold substernal sensation while playing tennis.

V/Q scan showed intermediate probability for PE. CTPA

showed a filling defect in left pulmonary artery with dis-

tension of the lumen. Biopsy of left lower lobe nodule (not

shown) revealed angiosarcoma.


malignancy, pleural effusion, and mediastinal mass

(Figs. 10–12).

The ever-increasing speed of multidetector CT

now allows for repeated scanning through the hila

during the course of the contrast bolus. A graph of

enhancement versus time allows for measurement of

functional tissue perfusion is reminiscent of nuclear

perfusion imaging but with specific anatomic detail

[43] that might, in turn, give important information

regarding patient management.

Fig. 12. Alternate diagnosis: pleural and pericardial effusion.

Sixty-year-old woman who had shortness of breath and

chest pain. rule out PE. CTPA showed no evidence of PE.

Magnetic resonance pulmonary angiography

MRPA, the ‘‘other’’ cross-sectional imaging tech-

nique, can be of use as an adjunctive technique in

selected surgical patients. Like CTPA, the test directly

detects the presence of emboli as filling defects in

contrast-labeled pulmonary arteries; however, MRPA

relies on completely different physical principles

(nuclear magnetic resonance rather than x-ray scatter-

ing) for image formation. Advantages of MRPA

include lack of ionizing radiation and the relatively

low incidence of renal and allergic complications

from gadolinium chelates [44], which facilitates im-

aging of particularly radiation-sensitive patients such

as pregnant women and patients who are allergic to

iodinated contrast or who suffer from renal insuffi-

ciency; however, MRPA is a longer, more complex,

and expensive test than CTPA, and its availability and

practicality vary widely.

Efficacy of modern gadolinium-based MRPA is

comparable to CT for segmental and larger emboli. In

a porcine model study involving 42 PEs, MRPA and

CTPA were found to have similar sensitivities (82%

versus 76%, respectively) and positive predictive

values (94% versus 92%) using a pathologic gold

standard [45]. Human studies comparing gadolinium-

based MRPA to a digital subtraction angiography

(DSA) gold standard are summarized in Table 2.

False-positives are uncommon, with specificity

reported as greater than 95% in all studies. Sensitivity

ranged from 68% to 100%; sensitivity for smaller

emboli was lower in all studies in which this discrim-

ination was made. The largest single study [46]

reported a sensitivity of only 40% for isolated sub-

segmental PEs compared with 84% for segmental and

100% for lobar or central PEs. Furthermore, outcome-

based studies following patients who had negative

MRPA (eg, those summarized in Table 1 for CTPA)

have not yet been performed for MRPA. Like CTPA, a

positive MRPA is a solid basis for treatment; however,

a negative MRPA does not exclude the possibility of

small PEs, and no adequate trials have demonstrated

the safety of withholding anticoagulation on the basis

of a negative MRPA.

Other barriers arise frequently when considering

MRPA. Critically ill patients who have many lines and

monitors can be difficult to place and adequately

monitor inside a magnet bore. Patients who have sur-

gical materials such as ferromagnetic aneurysm clips

or pacemakers must not enter an MRI facility. These

and other patients are not candidates for MRPA

[47,48]. Also, unlike CTPA, MRPA provides little

information about the lung parenchyma, so it has a

Table 2

Comparison studies of magnetic resonance angiography and digital subtraction angiography (DSA)

Emboli Sensitivity (%) Specificity (%) Distal evaluation Ref.

22 100 95 No subsegmental emboli reported on DSA [52]

19a 68 99.7 Missed 4/6 subsegmental PEs [53]

61 82 98 Missed 6/16 subsegmental PEs [48]

19 70 100 All 6 distal PEs missed [50]

Range 68–100 95–100

Each study compared gadolinium-enhanced MRA with conventional DSA, using conventional DSA as the gold standard.a All patients in this study had initial nondiagnostic V/Q scans.


greatly reduced ability to provide alternative diag-

noses. Health system barriers such as availability of

magnet time, state-of-the-art MRI hardware, local

radiologist expertise, and the considerable cost of

the study must be considered in the decision to order

the test.

MRPA, like CTPA, continues to evolve and im-

prove. Early methods used mostly spin-echo or time-

of-flight techniques and were limited by flow artifacts

and long imaging times, precluding breath-hold

images (see [49] for a review). These techniques are

often still used as supplemental imaging sequences,

but MRPA in nonpregnant patients now usually

involves administration of intravenous gadolinium

as a contrast agent. These methods, pioneered in

humans by Loubeyre and colleagues [50], increase

visibility of the distal arterial tree, especially when

done using breath-hold methods. Using state-of-the-

art equipment, some MRPA techniques can now be

accomplished in as little as 4 seconds [51].

In summary, MRPA is a useful method for ruling

in a PE in selected patients who have contraindica-

tions to CTPA. A positive MRPA is specific for PE

and thus appears to be sufficient basis for treatment;

however, a negative MRPA does not fully preclude

PE in the context of high clinical suspicion. The

practicality, cost, and poor visualization of the lung

parenchyma with MRPA must also be factored into

the decision to order the test.

Imaging algorithm

Imaging studies form the mainstay of the diag-

nosis of PE. The official position of the American

Thoracic Society, as adopted in 1999, states the

matter well: ‘‘The history, physical examination,

chest radiograph, electrocardiogram, and arterial

blood gas analysis. . .by itself. . .is inadequate to

confirm or exclude the diagnosis of PE’’ [52].

The natural history of PE combined with knowl-

edge of available modalities’ strengths and weak-

nesses forms the basis for an imaging algorithm that

is appropriate for surgical patients. In all patients,

the initial imaging study should be a high-quality

CXR. A study including PA and lateral views is

preferred, although often only an antero-posterior

(AP) view is possible. The presence of an obviously

unrelated nonthrombotic explanation for the patient’s

symptoms (eg, pneumothorax, mucus plug, dislo-

cated ET tube, and so forth) should lead to appro-

priate treatment. Further imaging directed toward

embolic disease is then only pursued if symptoms

unexpectedly persist.

Most surgical patients should then be considered

for CTPA as shown in the authors’ proposed algo-

rithm (Fig. 13). CTPA has repeatedly been shown to

be an effective first-line test for PE. It can be obtained

rapidly at most centers and it is diagnostic much more

often than V/Q scanning [53,54]. CTPA has been

found to be more sensitive and specific overall in at

least one direct prospective comparison of the two

procedures as the initial test [53]. When CTPA is

nondiagnostic, consideration should be given to con-

ventional pulmonary angiography, keeping in mind

the additional cumulative load of contrast dye.

V/Q scanning is appropriate as the initial test in

low- or intermediate-risk patients who have scrupu-

lously normal PA and lateral CXRs. In these patients

the V/Q scan is often normal, which excludes the

presence of PE without the risk of iodinated contrast

material. If findings are equivocal, the algorithm

should continue with CTPA or MRPA. Patients

who had abnormal CXRs who are not candidates

for CTPA (eg, patients who have renal insufficiency

[typically Cr >1.5 mg/dL] or severe allergy to iodin-

ated contrast media) should be considered for MRPA.

Although this test has limitations and contraindica-

tions as mentioned previously, a positive MRPA is a

solid basis for treatment (Table 2) and allows the

diagnosis of PE to be made without ionizing radia-

Fig. 13. Imaging algorithm.


tion or iodinated contrast media. MRPA also often

allows for the establishment of alternative diagnoses

and simultaneous venous study, although it is not as

effective at evaluating lung parenchyma as CT. A

negative MRPA is not an adequate basis for with-

holding treatment.

The role of venous imaging is somewhat com-

plex. Doppler imaging of the legs does not detect PE

directly, and it is neither sensitive nor specific for the

condition; however, the presence of DVT puts the

patient at risk for PE even if one has not yet occurred,

and anticoagulation is indicated for the DVT alone.

Anticoagulation usually obviates further imaging for

detection of PE itself. If the clinical question is ‘‘has

the patient suffered a pulmonary embolus?’’ then

lower extremity imaging is indicated only when direct

tests for PE cannot be done, and this is how the

authors’ algorithm (Fig. 13) was designed. If the

question is ‘‘would the patient benefit from antico-

agulation?’’ then it would be reasonable to perform a

venous imaging study even when a PE has been ruled

out by CTPA, V/Q, or CA because even a DVT that

has not embolized generally deserves anticoagulation.

If desired, CT or MR venography can be performed at

the same time as CTPA (MRPA) without additional

contrast media and with little additional radiation

(in the case of CT).

Caution should be used when relying on venous

imaging to guide treatment. A negative Doppler

alone does not exclude PE. The decision to treat

inpatients who cannot leave the Surgical Intensive

Care Unit (SICU) for V/Q, CTPA, or MRPA but who

have negative Dopplers must be based on d-dimer,

CXR findings, and clinical judgment alone. Patients

treated for PE on the basis of venous imaging alone

should be reevaluated, when possible, to confirm the

diagnosis, thus ensuring that the patient’s symptoms

are not caused by another undetected condition

masquerading as PE. Under these circumstances a

negative confirmatory study for PE would not negate

the need for anticoagulation for the DVT.

The exception to the rule of imaging diagnosis of

PE is the case of massive PE. Eleven percent of pa-

tients who have PE die within 1 hour of presentation

from systemic collapse caused by increased right heart

strain and acute pulmonary hypertension leading to

cardiovascular collapse [55]. Presentation of these

patients is often dramatic, and treatment (thrombolysis

or thrombectomy) is different than for the majority of

patients who have submassive PEs. TEE can be used


to investigate patients who have sudden acute shock

and appropriate physical signs, followed immediately

by thrombolysis if a central PE is found [56]. TEE can

also be used to assess for indirect signs (see [55] for a

review of these cases and their management).

Thus, each imaging modality has a role in the

diagnosis of PE. Normal V/Q, CA, or CTPA appear to

be adequate for withholding treatment, whereas high-

probability V/Q or positive MRPA, CTPA, or CA

appear to be specific. MRPA can be used when CTPA

is contraindicated, and V/Q scanning is still useful for

low- or intermediate-risk patients or patients who have

contrast allergy and contraindication to MRPA. Lower

extremity venous studies are neither sensitive nor

specific but can be done portably, whereas TEE can

detect some large central emboli quickly. For most

surgical patients, however, CTPA appears to be the

first and only advanced imaging modality needed to

diagnose or exclude PE.

References

[1] Dalen JE. Pulmonary embolism: what have we learned

since Virchow?: treatment and prevention. Chest 2002;

122(5):1801–17.

[2] Tuttle-Newhall JE, Rutledge R, Hultman CS, Fakhry

SM. Statewide, population-based, time-series analy-

sis of the frequency and outcome of pulmonary embo-

lus in 318,554 trauma patients. J Trauma 1997;42(1):

90–9.

[3] Lindblad B, Eriksson A, Bergqvist D. Autopsy-verified

pulmonary embolism in a surgical department: analysis

of the period from 1951 to 1988. Br J Surg 1991;78(7):

849–52.

[4] Gillies TE, Ruckley CV, Nixon SJ. Still missing the

boat with fatal pulmonary embolism. Br J Surg 1996;

83(10):1394–5.

[5] Abad C, Caceres JJ, Alonso A. Fatal pulmonary throm-

boembolism after laparoscopic cholecystectomy. Surg

Endosc 2001;15(11):1360–4.

[6] Jang JS, Lee SH, Jung SK. Pulmonary embolism of

polymethylmethacrylate after percutaneous vertebro-

plasty: a report of three cases. Spine 2002;27(19):

E416–8.

[7] Chen HL, Wong CS, Ho ST, Chang FL, Hsu CH, Wu

CT. A lethal pulmonary embolism during percutaneous

vertebroplasty. Anesth Analg 2002;95(4):1060–2.

[8] Tozzi P, Abdelmoumene Y, Corno AF, Gersbach PA,

Hoogewoud HM, von Segesser LK. Management of

pulmonary embolism during acrylic vertebroplasty.

Ann Thorac Surg 2002;74(5):1706–8.

[9] White RH, Henderson MC. Risk factors for venous

thromboembolism after total hip and knee replacement

surgery. Curr Opin Pulm Med 2002;8(5):365–71.

[10] Hansson PO, Welin L, Tibblin G, Eriksson H. Deep

vein thrombosis and pulmonary embolism in the gen-

eral population. The study of men born in 1913. Arch

Intern Med 1997;157(15):1665–70.

[11] Value of the ventilation/perfusion scan in acute pul-

monary embolism. Results of the prospective investi-

gation of pulmonary embolism diagnosis (PIOPED).

The PIOPED Investigators. JAMA 1990;263(20):

2753–9.

[12] Stein PD, Athanasoulis C, Alavi A, Greenspan RH,

Hales CA, Saltzman HA, et al. Complications and va-

lidity of pulmonary angiography in acute pulmonary

embolism. Circulation 1992;85(2):462–8.

[13] Stein PD, Hull RD, Saltzman HA, Pineo G. Strategy

for diagnosis of patients with suspected acute pulmo-

nary embolism. Chest 1993;103(5):1553–9.

[14] Diebold J, Lohrs U. Venous thrombosis and pulmonary

embolism. A study of 5039 autopsies. Pathol Res Pract

1991;187(2–3):260–6.

[15] Turkstra F, Kuijer PM, van Beek EJ, Brandjes DP, ten

Cate JW, Buller HR. Diagnostic utility of ultrasonog-

raphy of leg veins in patients suspected of having pul-

monary embolism. Ann Intern Med 1997;126(10):

775–81.

[16] Kearon C, Ginsberg JS, Hirsh J. The role of venous

ultrasonography in the diagnosis of suspected deep

venous thrombosis and pulmonary embolism. Ann In-

tern Med 1998;129(12):1044–9.

[17] Lensing AW, Prandoni P, Brandjes D, Huisman PM,

Vigo M, Tomasella G, et al. Detection of deep-vein

thrombosis by real-time B-mode ultrasonography. N

Engl J Med 1989;320(6):342–5.

[18] Naidich JB, Torre JR, Pellerito JS, Smalberg IS, Kase

DJ, Crystal KS. Suspected deep venous thrombosis: is

US of both legs necessary? Radiology 1996;200(2):

429–31.

[19] Loud PA, Katz DS, Klippenstein DL, Shah RD, Gross-

man ZD. Combined CT venography and pulmonary

angiography in suspected thromboembolic disease: di-

agnostic accuracy for deep venous evaluation. AJR

Am J Roentgenol 2000;174(1):61–5.

[20] Cham MD, Yankelevitz DF, Shaham D, Shah AA,

Sherman L, Lewis A, et al. Deep venous thrombosis:

detection by using indirect CT venography. The Pul-

monary Angiography-Indirect CT Venography Coop-

erative Group. Radiology 2000;216(3):744–51.

[21] Garg K, Kemp JL, Wojcik D, Hoehn S, Johnston RJ,

Macey LC, et al. Thromboembolic disease: comparison

of combined CT pulmonary angiography and venogra-

phy with bilateral leg sonography in 70 patients. AJR

Am J Roentgenol 2000;175(4):997–1001.

[22] Goodman LR. 1999 plenary session: Friday imaging

symposium: CT diagnosis of pulmonary embolism and

deep venous thrombosis. Radiographics 2000;20(4):

1201–5.

[23] Remy-Jardin M, Remy J, Wattinne L, Giraud F. Cen-

tral pulmonary thromboembolism: diagnosis with

spiral volumetric CT with the single-breath-hold tech-

nique—comparison with pulmonary angiography. Ra-

diology 1992;185(2):381–7.


[24] Teigen CL, Maus TP, Sheedy PF, Johnson CM, Stanson

AW, Welch TJ. Pulmonary embolism: diagnosis with

electron-beam CT. Radiology 1993;188(3):839–45.

[25] Goodman LR, Curtin JJ, Mewissen MW, Foley WD,

Lipchik RJ, Crain MR, et al. Detection of pulmonary

embolism in patients with unresolved clinical and scin-

tigraphic diagnosis: helical CT versus angiography.

AJR Am J Roentgenol 1995;164(6):1369–74.

[26] van Rossum AB, Pattynama PM, Ton ER, Treurniet

FE, Arndt JW, van Eck B, et al. Pulmonary embolism:

validation of spiral CT angiography in 149 patients.

Radiology 1996;201(2):467–70.

[27] de Monye W, van Strijen MJ, Huisman MV, Kieft GJ,

Pattynama PM. Suspected pulmonary embolism: preva-

lence and anatomic distribution in 487 consecutive pa-

tients. Advances in New Technologies Evaluating the

Localisation of Pulmonary Embolism (ANTELOPE)

Group. Radiology 2000;215(1):184–8.

[28] Stein PD,Henry JW. Prevalence of acute pulmonary em-

bolism in central and subsegmental pulmonary arteries

and relation to probability interpretation of ventilation/

perfusion lung scans. Chest 1997;111(5):1246–8.

[29] Oser RF, Zuckerman DA, Gutierrez FR, Brink JA.

Anatomic distribution of pulmonary emboli at pulmo-

nary angiography: implications for cross-sectional

imaging. Radiology 1996;199(1):31–5.

[30] Baile EM, King GG, Muller NL, D’Yachkova Y,

Coche EE, Pare PD, et al. Spiral computed tomogra-

phy is comparable to angiography for the diagnosis

of pulmonary embolism. Am J Respir Crit Care Med

2000;161(3 Pt 1):1010–5.

[31] Ghaye B, Szapiro D, Mastora I, Delannoy V, Duhamel

A, Remy J, et al. Peripheral pulmonary arteries: how

far in the lung does multi-detector row spiral CT allow

analysis? Radiology 2001;219(3):629–36.

[32] Schoepf UJ, Holzknecht N, Helmberger TK, Crispin

A, Hong C, Becker CR, et al. Subsegmental pulmonary

emboli: improved detection with thin-collimation

multi-detector row spiral CT. Radiology 2002;222(2):

483–90.

[33] Raptopoulos V, Boiselle PM. Multi-detector row spi-

ral CT pulmonary angiography: comparison with sin-

gle-detector row spiral CT. Radiology 2001;221(3):

606–13.

[34] Lomis NN, Yoon HC, Moran AG, Miller FJ. Clinical

outcomes of patients after a negative spiral CT pulmo-

nary arteriogram in the evaluation of acute pulmonary

embolism. J Vasc Interv Radiol 1999;10(6):707–12.

[35] Ost D, Rozenshtein A, Saffran L, Snider A. The neg-

ative predictive value of spiral computed tomography

for the diagnosis of pulmonary embolism in patients

with nondiagnostic ventilation–perfusion scans. Am J

Med 2001;110(1):16–21.

[36] Gottsater A, Berg A, Centergard J, Frennby B, Nirhov

N, Nyman U. Clinically suspected pulmonary embo-

lism: is it safe to withhold anticoagulation after a nega-

tive spiral CT? Eur Radiol 2001;11(1):65–72.

[37] Swensen SJ, Sheedy PF, Ryu JH, Pickett DD, Schleck

CD, Ilstrup DM, et al. Outcomes after withholding

anticoagulation from patients with suspected acute pul-

monary embolism and negative computed tomographic

findings: a cohort study. Mayo Clin Proc 2002;77(2):

130–8.

[38] Bourriot K, Couffinhal T, Bernard V, Montaudon M,

Bonnet J, Laurent F. Clinical outcome after a negative

spiral CT pulmonary angiographic finding in an inpa-

tient population from cardiology and pneumology

wards. Chest 2003;123(2):359–65.

[39] Goodman LR, Lipchik RJ, Kuzo RS, Liu Y, McAuliffe

TL, O’Brien DJ. Subsequent pulmonary embolism:

risk after a negative helical CT pulmonary angio-

gram—prospective comparison with scintigraphy. Ra-

diology 2000;215(2):535–42.

[40] Musset D, Parent F, Meyer G, Maitre S, Girard P,

Leroyer C, et al. Diagnostic strategy for patients with

suspected pulmonary embolism: a prospective

multicentre outcome study. Lancet 2002;360(9349):

1914–20.

[41] Tillie-Leblond I, Mastora I, Radenne F, Paillard S,

Tonnel AB, Remy J, et al. Risk of pulmonary embo-

lism after a negative spiral CT angiogram in patients

with pulmonary disease: 1-year clinical follow-up

study. Radiology 2002;223(2):461–7.

[42] Kim KI, Muller NL, Mayo JR. Clinically suspected

pulmonary embolism: utility of spiral CT. Radiology

1999;210(3):693–7.

[43] Schoepf UJ, Bruening R, Konschitzky H, Becker CR,

Knez A, Weber J, et al. Pulmonary embolism: compre-

hensive diagnosis by using electron-beam CT for de-

tection of emboli and assessment of pulmonary blood

flow. Radiology 2000;217(3):693–700.

[44] Runge VM. Safety of approved MR contrast media for

intravenous injection. J Magn Reson Imaging 2000;

12(2):205–13.

[45] Reittner P, Coxson HO, Nakano Y, Heyneman L, Ward

S, King GG, et al. Pulmonary embolism: comparison

of gadolinium-enhanced MR angiography with con-

trast-enhanced spiral CT in a porcine model. Acad Ra-

diol 2001;8(4):343–50.

[46] Oudkerk M, van Beek EJ, Wielopolski P, van Ooijen

PM, Brouwers-Kuyper EM, Bongaerts AH, et al. Com-

parison of contrast-enhanced magnetic resonance an-

giography and conventional pulmonary angiography

for the diagnosis of pulmonary embolism: a prospec-

tive study. Lancet 2002;359(9318):1643–7.

[47] Kanal E, Borgstede JP, Barkovich AJ, Bell C, Bradley

WG, Felmlee JP, et al. American College of Radiology

white paper on MR safety. AJR Am J Roentgenol 2002;

178(6):1335–47.

[48] Shellock FG. Magnetic resonance safety update 2002:

implants and devices. J Magn Reson Imaging 2002;

16(5):485–96.

[49] Erdman WA, Clarke GD. Magnetic resonance imaging

of pulmonary embolism. Semin Ultrasound CT MR

1997;18(5):338–48.

[50] Loubeyre P, Revel D, Douek P, Delignette A, Baldy C,

Genin G, et al. Dynamic contrast-enhancedMR angiog-

raphy of pulmonary embolism: comparison with pul-


monary angiography. AJR Am J Roentgenol 1994;

162(5):1035–9.

[51] Goyen M, Laub G, Ladd ME, Debatin JF, Barkhausen

J, Truemmler KH, et al. Dynamic 3D MR angiography

of the pulmonary arteries in under four seconds.

J Magn Reson Imaging 2001;13(3):372–7.

[52] Tapson VF, Carroll BA, Davidson BL, Elliott CG, Fe-

dullo PF, Hales CA, et al. The diagnostic approach to

acute venous thromboembolism. Clinical practice

guideline. American Thoracic Society. Am J Respir

Crit Care Med 1999;160(3):1043–66.

[53] Blachere H, Latrabe V, Montaudon M, Valli N, Couf-

finhal T, Raherisson C, et al. Pulmonary embolism re-

vealed on helical CT angiography: comparison with

ventilation – perfusion radionuclide lung scanning.

AJR Am J Roentgenol 2000;174(4): 1041–7.

[54] Crawford T, Yoon C, Wolfson K, Beller M, Emerick A,

Goldin JG, et al. The effect of imaging modality on

patient management in the evaluation of pulmonary

thromboembolism. J Thorac Imaging 2001;16(3):

163–9.

[55] Wood KE. Major pulmonary embolism: review of a

pathophysiologic approach to the golden hour of he-

modynamically significant pulmonary embolism. Chest

2002;121(3):877–905.

[56] Krivec B, Voga G, Zuran I, Skale R, Pareznik R, Pod-

bregar M, et al. Diagnosis and treatment of shock due

to massive pulmonary embolism: approach with trans-

esophageal echocardiography and intrapulmonary

thrombolysis. Chest 1997;112(5):1310–6.


Computer-aided diagnostics

Anthony P. Reeves, PhDa,*, Bryan M. Kressler, BSb

aSchool of Electrical and Computer Engineering, Cornell University, 331 Rhodes Hall, Ithaca, NY 14853, USAbBiomedical Engineering Program, Cornell University, 357 Rhodes Hall, Ithaca, NY 14853, USA

In this paper the authors provide an overview of Each of these domains is considered in turn. The

recent work in using computer analysis of CT images

of the lungs to aid the physician in diagnosing dis-

eases of the lung and planning treatments, with par-

ticular attention to lung cancer. Computers can provide

such aid in several domains:

1. Image visualization: by providing different

viewing options, the computer can present

image data to the radiologist in a more con-

venient form for diagnosis or to a surgeon in

such a way that anatomic relationships can be

recognized more easily

2. Detection: the computer can be used to detect

lung abnormalities automatically, especially in

typical situations in which whole-lung scans

consist of hundreds of images and the abnor-

mality is small and might only be visible on

one image

3. Characterization: the computer can make mea-

surements on a pulmonary nodule to determine

its malignancy status; currently, the most ac-

curate measurement for predicting malig-

nancy is growth rate, which is determined

from the change in nodule size in two time-

separated scans

4. Abnormality documentation and treatment

evaluation: if many nodules are present, the

computer is ideal for the tedious cataloging and

documenting task. Furthermore, for treatment

evaluation the computer can be used to quan-

titatively measure the difference through the

whole-lung region before and after surgical or

nonsurgical treatments


doi:10.1016/S1547-4127(04)00030-1


E-mail address: [email protected] (A.P. Reeves).

lung is a particularly convenient organ for CT image

analysis because many abnormalities show up as

brighter image regions on the dark lung parenchyma

background, which is in contrast to other organs, in

which the contrast is much less and the delineation

of abnormal tissue is typically more difficult. The

lung also presents a unique challenge to accurate mea-

surement because of its compressible nature. Image

analysis is complicated, for example, by change in

patient position, degree of inspiration, heart motion

image artifacts, and body movement image artifacts.

CT technology has made considerable advances

over the past decade, providing better information

and greater challenges for the radiologist. Newer

multislice scanners can capture many more images

in a single breath-hold. Consequently, the radiologist

is confronted with the task of examining several

hundred images for a single whole-lung scan rather

than the tens of images characteristic of older scan-

ners. Furthermore, these images are now often re-

corded using a low-dose protocol, which means that

there is much more image noise, making the reading

more difficult. However, for computer analysis the

thinner slices provide a tremendous opportunity for

considering the CT scan as single three-dimensional

(3D) image rather than the traditional viewpoint of a

set of individual two-dimensional images. Computer

methods can use true 3D geometric analysis, which

is much simpler and more direct than the two-

dimensional counterpart.

For 3D geometric techniques to be used, the 3D

image must have a close to isotropic voxel size. The

resolution in the axial direction (slice thickness) must

be similar to the in-plane resolution (pixel size). For

example, a typical whole-lung image has a pixel size

of about 0.6 � 0.6 mm. If the slice thickness is

s reserved.

A.P. Reeves, B.M. Kressler / Thorac Surg Clin 14 (2004) 125–133126

10 mm, there is an anisotropic mismatch of 10 to 0.6

or about 18 to 1. For a 1 mm slice thickness the ratio

is 1:0.6 or about 1.8:1. The most recent multislice

scanners offer 0.5 mm slice thickness, in which the

ideal 1:1 ratio is achievable.

The main benefits of computer-aided image analy-

sis are realized when quantitative methods are used

for measuring and classifying image characteristics.

Image visualization, in which the computer provides

a more human-convenient presentation of image

data, has been the more traditional use of computer

assistance; however, such a qualitative approach

leaves the image analysis and decision making en-

tirely up to the radiologist. In contrast, quantitative

data analysis can provide three major benefits to

the radiologist:

1. More accurate and repeatable measurements.

The computer does not suffer from fatigue and

will consistently use the same measurement

algorithm with the same parameters every time

it is applied. In contrast, humans use a num-

ber of subjective judgments in making mea-

surements and there are many sources of

measurement variation. However, the computer

sometimes makes mistakes in locating the cor-

rect boundary for a region of interest. Therefore,

a good strategy is to have the human observe the

decisions made by computer and to manually

override the measurement process when any

incorrect computer decisions are observed.

2. Large database for diagnosis. Diagnosis by

computer involves comparing the quantitative

information from an image with knowledge of

all previous examples that the computer has in

its database. This knowledge can be recorded

in the computer in many ways, from a set of

actual images to a set of derived measurement

parameters. However the data are organized,

the general result is that the larger the knowl-

edge database (number of cases), the better the

quality of the computer diagnosis. Given that

memory to store the cases is no longer a major

consideration for modern computer technology,

clinicians can anticipate that performance of

computer diagnosis methods will continue

to improve with time. Compare this scenario

to the physician, who must typically make a

judgment based on a lifetime personal experi-

ence of only a few hundred cases.

3. Management of large data sets. The computer

is an excellent data manager ideally suited to

the tedious task of documenting all abnormali-

ties that might be present in a single whole-

lung scan. Furthermore, it is equally well suited

to matching two time-separated whole-lung

scans and documenting the changes that have

occurred in the period between them. For

radiologists this is an arduous, time-consuming

task that is difficult to perform consistently for

long periods of time.

Recent advances in qualitative image visualization

are considered in the next section, followed by the

advantages of using quantitative methods for de-

tection, characterization, and general documentation.

Visualization

Unlike chest radiographs (the original standard for

chest radiography), the visualization of CT images

has always required computer assistance in the form

of digital reconstruction algorithms, even when the

images are presented to the radiologist on a film

base. To acquire CT image data, several parameters,

including the dose and slice thickness, need to be pre-

established; however, to view a CT image a radi-

ologist must specify a number of post hoc parameters

when the raw CT image data has been acquired.

These parameters include brightness and contrast

(level and widow), spatial enhancement, and the field

of view (magnification). The use of film fixes these

parameters while the use of a soft-copy computer dis-

play device permits the radiologist to modify these

parameters in real time while viewing the image data.

The standard radiology soft-copy viewing station

is designed to accommodate a range of imaging

modalities. Constraints on projective images such as

chest radiographs and mammograms require high

resolution and a well-controlled viewing environment;

hence, most such soft-copy systems are costly and

involve a special high-resolution grayscale monitor.

The viewing requirements for high-contrast lung

CT scans are, in general, less stringent than for pro-

jective images. Many standard PCs offer adequate

quality viewing characteristics for viewing CT image

data, especially for the characterization of previously

identified nodules. Furthermore, they offer color as

one means of drawing attention to regions of interest

such as lung abnormalities. Standard computer

graphics methods coupled with simple computer

analysis offer alternative modes for viewing CT image

data, and vendors are introducing such methods.

Acceptance of these methods by radiologists has been

rather slow; possibly one factor is the difficulty in

Fig. 1. Axial CT image with a small pulmonary nodule

outlined.

A.P. Reeves, B.M. Kressler / Thorac Surg Clin 14 (2004) 125–133 127

incorporating such techniques on the traditional high-

resolution, grayscale, soft-copy workstations.

To illustrate some of the visualization options, the

visualization of a single lung nodule is presented.

In Fig. 1, a conventional axial CT image of the lung

is shown with a 10 mm nodule outlined in the left

lung. All scanners and soft-copy workstations sup-

port this conventional visualization. Fig. 2 shows the

image slices through that nodule at the same time

in a montage display. Fig. 3 shows an alternative

Fig. 2. Consecutive 1 mm image slices th

method of viewing these data using standard com-

puter graphics techniques. A visualization is gene-

rated to resemble how the nodule might look if

it were perfectly extracted from the lung, if it had a

perfectly reflecting matt surface, and if it was illu-

minated by a single, simple light source with some

uniform background illumination. Because this

method involves creating a 3D model of the nodule

(which is made possible by the isotropic property of

the thin slice scan), the nodule can be viewed from

different viewpoints than just the conventional axial

direction. On a viewing workstation the nodule can

be rotated and viewed from any arbitrary direction. In

Fig. 3, visualizations from the three canonical view-

points, axial, sagittal, and coronal, are provided.

A second conventional nodule image is shown

in Fig. 4. The visualization of that nodule using

standard ray tracing techniques is shown in Fig. 5.

In this case the body tissue is given an opacity related

to its radiograph density, then computer algorithms

identify the nodule region and colorize the other

dense image objects in the lung according to their

geometric form, hence using color to highlight the

region of prime interest to the radiologist (Fig. 6). A

magnified image of the nodule is shown in Fig. 7.

In Fig. 8A a CT image of a nonsolid nodule or

ground-glass opacity (GGO) is shown. For this im-

portant nodule type there is a distinctive difference

in density between the nonsolid material and other

solid tissue (eg, vessels and chest wall). To visualize

the nonsolid tissue the authors use a translucent

rendering method as shown in Fig. 8B. The interac-

tion between the vessels and the nodule can now be

rough the nodule shown in Fig. 1.

Fig. 3. 3D reconstructions of the nodule shown in Fig. 2 rendered from three orthogonal viewing directions: axial, sagittal,

and coronal.


seen. Further, the authors can remove the vessels

from consideration as shown in Fig. 8C to have an

unobstructed view of the whole nodule’s shape.

While visualization is important, especially for the

correct interpretation of image data by radiologists,

the real power of the computer is in the quantitative

analysis of the data to directly determine clinically

relevant information. The visualizations shown in

Fig. 9 show a rendering of a geometrical model of

a nodule derived from the nodule size measurements

extracted from the computer segmentation process.

This reference (see Fig. 3) is derived from the

quantitative analysis, whereas other renderings are

Fig. 4. Conventional focused study image of a small, solid

pulmonary nodule.

computed without explicit segmentation, and mea-

surements cannot be made from them.

Detection

A computer assistant for detection examines a

whole-lung CT scan for any evidence of pulmonary

nodules and reports the locations of suspected nodule

to the radiologist. The most common scenario (one

that is currently the most likely to be approved by the

U.S. Food and Drug Administration) is to use com-

puter-detected results as a second read. When the

Fig. 5. Ray-traced rendering of the nodule shown in Fig. 4.

Fig. 6. Color-coded rendering of the nodule shown in Fig. 4.

Fig. 7. Magnified view of the nodule shown in Fig. 6.


radiologist has performed a conventional reading of

the scan, the computer highlights possible nodule

locations that have not been documented by the

radiologist. The radiologist then examines these loca-

tions with a view to modifying the report.

The critical issue is that the detection system

must be sensitive enough to detect essentially all

nodules without indicating too many false alarms

(false-positives). To achieve this performance, the

detection algorithm must have a high sensitivity and

specificity. Furthermore, a sensitivity parameter needs

to be set to fix the sensitivity/specificity tradeoff to

the optimal value for a given clinical reading task.

Of importance to a detection system is the defini-

tion of a nodule or a reportable event. In general,

large nodules are easier to identify for the radiologist

and the machine. The task of nodule identification

becomes increasingly more difficult as the nodule

size approaches the voxel size of the scanner. Fur-

thermore, as nodules of a smaller size are detected

and characterized, there is a higher probability that

smaller, benign nodules will be detected. Clinically

relevant nodules are currently considered to be in the

3 mm to 3 cm size range. In addition, other lung

abnormalities might be of interest to the radiologist,

and the issue of reporting these abnormalities should

be addressed.

Experimental computer-aided systems for lung

nodule detection are currently being developed and

evaluated. Basic algorithms now exist for nodule

detection [1–3]; however, more work is needed to

optimize these methods and to establish appropriate

parameters for general clinical use. For example, the

specification of exactly what abnormalities are to

be reported and the appropriate size and sensitivity

settings commiserate with an acceptable rate of false-

positives for clinical use must be determined. Further-

more, as technology improves and more experience

is gained, it is anticipated that future methods will

achieve a significant improvement with respect to

sensitivity and specificity.

Characterization

The computer can aid the physician in a number

of ways by characterizing the detected nodule. Be-

yond a variety of special visualizations, the computer

can provide quantitative measurements on that nodule

for the physician to interpret or it can perform a

classification on these measurements (based on a

large number of previously diagnosed nodules) to

directly determine the probability of malignancy.

The basic procedure for nodule characterization is

1. Segmentation, determining which voxels be-

long to the nodule and which do not

2. Feature extraction, making quantitative mea-

surements on the nodule voxels

3. Diagnosis/classification, determining the prob-

ability of malignancy from a statistical analysis

of the extracted features

Two basic methods have been explored to deter-

mine the malignancy status of a nodule by computer

evaluation: shape features and size change. In the

Fig. 8. Visualization of a nonsolid nodule. (A) CT image. (B) Rendered as a translucent blue region with vessels marked in red.

(C) With vessels removed.

Fig. 9. Light-shaded views of the segmented nodule in Fig. 4.



shape feature method a number of measurements are

made on the nodule voxels from a CT scan and these

features are used to predict malignancy by way of a

classifier that has been trained on a database consist-

ing of documented malignant and benign nodules.

This method has been explored in research settings,

and while the initial results are quite promising [4],

more research is needed. The second method is to

measure growth rate from the nodule size change in

two time-separated CT scans. In preliminary studies

[5] a high growth rate has been found to be an ex-

cellent predictor of malignancy; however, this ap-

proach has the drawbacks of requiring a second CT

scan and the delay in diagnosis caused by the re-

quired time period between scans

CT manufacturers and other vendors are now pro-

viding 3D nodule growth estimation tools. Issues

with this approach are that the two scans must be

of a high quality and recorded with the same CT scan

parameters, and the time delay between scans must

be long enough to obtain a sufficiently accurate mea-

surement to predict malignancy, but this delay needs

to be minimized to reduce patient anxiety. For cur-

rent CT scanner technology, this optimal time period

Fig. 10. Nodule candidate locations displayed o

might be several months for small nodules, reducing

to perhaps less than 1 month for larger (1 cm) nodules.

Documentation and health evaluation

Beyond detecting andmeasuring nodules, the com-

puter system should also be capable of facilitating

other operations such as whole-lung health monitor-

ing and automated nodule cataloging. This is espe-

cially important for repeat scans in a screening or

treatment scenario. This operation in itself does not

require any new technological developments; rather, it

requires the development of a patient management

system that goes beyond conventional Radiological

Information Systems (RIS). For example, the authors

have built into their data management system a whole-

lung volume and emphysema analysis capability. The

computer automatically delineates the whole lung

parenchyma region from the CT images and computes

the lung volume. An analysis of the density distribu-

tion of the lung parenchyma is also computed to

produce a visualization of the spatial distribution of

emphysema. The outcome of this analysis presented

n a coronal projection image of the lungs.

Fig. 11. Emphysema visualization on a coronal projection of the lungs.


from a coronal viewpoint is shown in Figs. 10,11.

Whole-lung analysis of this type can be applied au-

tomatically to all CT scans before physician evalua-

tion on a routine basis.

Discussion

Recent computer methods have been used to

enhance the visualization of CT images for diagnostic

purposes and to provide quantitative measurements

on these images. It is in the latter respect that

computer methods are most powerful as a diagnostic

tool. The advantages to diagnostic radiology are self-

evident; however, these methods can also be applied

to aid treatment and intervention.

Many quantitative and qualitative improvements

can be anticipated in the anatomic analysis of the

chest CT. Algorithms have been developed for the

automatic segmentation of major anatomical regions

from chest CT scans; however, the authors anticipate

that future algorithm development will result in the

automatic segmentation of all major bone and tissue

regions in the thorax. The lung regions themselves

are extracted easily, as are the trachea and major

bronchi. Current efforts include the bone structures.

For the surgeon, these methods can be used for

surgical planning in the preoperative phase. The

boundaries of the lobes can be visualized and the

abnormality can be viewed from different directions

to present its spatial relationship to other structures

within the lung. The health of other structures can be

evaluated; the diameters of vessels, for example, can

be measured. The amount of healthy lung that is being

resected can be measured in emphysema patients so

that the amount of tissue removed can be modified.

Another area of development includes methods to

model how the lobes and the remaining lung will

remodel after surgery to anticipate any complications.

In addition, better scanner sensitivity coupled with

new algorithmic developments can achieve a more

precise evaluation of invasion, especially in critical

areas of the major vessels and the mediastinum.

This 3D visualization can be made available at

the operating table on a flat-panel screen, where it can

be referred to and manipulated in real time. Such

facilities can also be integrated with video-assisted

thoracic surgery (VATS) camera images. Additional

integrated visualizations can be provided based on

camera viewpoint, previous camera images, and avail-

able CT images.

In the postoperative domain, image analysis can

also provide a number of benefits. The effects of

medication can be measured accurately, and the

volume capacity of the lungs can be compared before

and after surgery and as healing progresses. Further,

the healing process itself can be monitored carefully;

the development of scar tissue and its growth pattern

can be ascertained. The effects of medication can be

evaluated and the optimal strategies for recovery can

be determined for an individual patient and for

refining the methods of standard practice.


Summary

The computer can be used in a number of ways

to aid the physician to interpret CT lung images.

Commercial tools are becoming available to assist

the radiologist in growth rate determination, hence

cancer diagnosis. Computer algorithms are in devel-

opment that will permit lung health evaluation, in-

cluding nodule detection. Finally, the results of such

efforts will probably produce more detailed visual-

izations of the lung region, including depictions of the

location and state of lung abnormalities. While com-

puter methods have found a first application with the

radiologist, these methods should also provide a

valuable aid to surgery and pathology.

References

[1] Lee Y, Hara T, Fujita H, Itoh S, Ishigaki T. Automated

detection of pulmonary nodules in helical CT images

based on an improved template-matching technique.

IEEE Trans Med Imaging 2001;20(7):595–604.

[2] Fan L, Qian J, Ordy B, Shen H, Naidich D, Kohl G,

et al. Automatic segmentation of pulmonary nodules by

using dynamic 3D cross-correlation for interactive CAD

systems. In: Sonka M, Fitzpatrick JM, editors. Proceed-

ings of the SPIE 2002;4648:1362–9.

[3] Taguchi H, Kawata Y, Niki N, Satoh H, Ohmatsu H,

Kakinuma R, et al. Lung cancer detection based on

helical CT images using curved surface morphology

analysis. SPIE Conference on Image Processing 1999;

3661:1307–14.

[4] Reeves AP, Kostis WJ, Yankelevitz DF, Henschke CI.

Three-dimensional shape characterization of solitary

pulmonary nodules from helical CT scans. In: Lemke

HU, Vannier MW, Inamura K, Farman AG, editors.

Proceedings of Computer Assisted Radiology and Sur-

gery (CARS ’99). New York: Elsevier Science; 1999.

p. 83–7.

[5] Yankelevitz DF, Reeves AP, Kostis WJ, Zhao B, Hens-

chke CI. CT evaluation of small pulmonary nodules

based on volumetrically determined growth rates. Radi-

ology 2000;217:251–6.


Future generation CT imaging

Deborah Walter, PhD*, Bruno De Man, PhD, Maria Iatrou, PhD,Peter M. Edic, PhD

Computed Tomography Systems and Applications Laboratory, GE Global Research Center, One Research Circle,

Niskayuna, NY 12309, USA

Since the 1970s CT has been used to generate porating multiple detector rows with improved reso-

cross-sectional images of human anatomy from x-ray

projection data acquired at many angular positions

around the body. The development of the first CT

scanner is credited to Hounsfield [1], using recon-

struction methods developed by Cormack [2]. Since

then several key technological advancements have

increased the usefulness of CT to the point where it

has become an essential tool in the clinical evaluation

of patients, with most hospitals having CT scanners

available 24 hours a day, 7 days a week. First-

generation CT systems, which used a single, thin

x-ray beam and a single detector element to acquire

the x-ray projection data necessary for image recon-

struction, quickly evolved into third-generation sys-

tems, which use a fan-shaped x-ray beam combined

with a rotating x-ray tube and detector to acquire the

x-ray projection data needed for image reconstruc-

tion. With the development of helical CT systems in

the late 1980s [3–6], large sections of anatomy could

be scanned quickly using linear, continuous table mo-

tion, enabling faster throughput and better registration

between adjacent slices of the body. Electron beam

CT (EBCT) technology, incorporating a stationary

detector and using an x-ray tube comprised of a

stationary target ring and a swept electron beam,

was developed in 1983 to scan the heart and coronary

arteries at subsecond scan times, thereby reducing

motion artifacts resulting from the beating of the heart

during data acquisition [7]. Recently, CT has further

evolved from single-slice scanners to systems incor-


doi:10.1016/S1547-4127(04)00040-4


E-mail address: [email protected] (D. Walter).

lution, with the introduction of multislice CT (MSCT)

in 1998, enabling high-resolution, thin-slice scanning

of large sections of human anatomy.

Over the years CT scanning has found widespread

use in imaging applications for trauma assessment,

thoracic and vascular studies, and stroke assessment, as

well as emerging applications in cardiac imaging, lung

cancer detection, colonoscopy, and brain and myo-

cardial perfusion. The goal of this article is to outline

some of the recent technological advances that will

drive future CT evolution and to describe the recently

enabled applications and trends in chest imaging.

Image acquisition aspects

Image acquisition with today’s multislice scanner

is more flexible when compared with a single-slice

scanner because many scanning modes are made

available to the clinician. For example, several com-

binations of slice thickness and helical pitch can now

be used to scan the full thorax within a single breath-

hold, giving clinicians many options to maximize the

clinical value of the acquired images. Because more

protocol options are available to the clinician, it is im-

portant to understand the relationship between scanner

parameters and resulting image quality so that the

acquisition can be tailored to meet the particular

imaging need. Some of the key geometric parameters

are identified in Fig. 1, a third-generation CT system.

Two important characteristics to consider when

thinking about CT image quality are spatial resolu-

tion and image noise. The key imaging parameters

s reserved.

Fig. 1. Third-generation multislice CT scanner. Front view (left) shows the x-ray tube and the x-ray detector. The patient lies

along the axis of rotation. The scan field of view is the circular region inscribed in the rotating fan beam. The side view (right)

shows the relationship between the cone angle, the longitudinal size of the detector, and the z-coverage, which is measured at the

isocenter. The z-dimensions are strongly exaggerated for clarity.

D. Walter et al / Thorac Surg Clin 14 (2004) 135–149136

that affect spatial resolution in reconstructed images

are the

� Size of the focal spot in the x-ray tube� Length, width, and configuration of the detector

cells� Location and number of angular positions at

which x-ray projection data are acquired as

the x-ray source and detector rotate around

the patient� Filter used to preprocess the x-ray projection

data before image reconstruction� Helical pitch ratio prescribing the table speed

The key imaging parameters that affect image

noise are the size and anatomy of the patient being

scanned, the operating voltage and current of the

x-ray tube, the rotational speed of the gantry com-

prising the x-ray tube and detector, in addition to all

of the parameters that affect spatial resolution.

The main source of image noise in reconstructed

CT images is the statistical nature of x-ray photons.

For each detector channel the deviation of the mea-

sured attenuation of x-rays along a particular direc-

tion from the true attenuation value decreases as the

number of detected photons increases. Consequently,

image noise decreases with increased x-ray tube

current (at the expense of increased patient dose),

increased detector size (at the expense of decreased

spatial resolution), and reduced attenuation (in a

small patient or low-density objects such as lungs).

Given the complex relationships between the imaging

parameters and their impact on image quality, future

imaging protocols will likely migrate from specifying

individual acquisition parameters to specifying image

quality and dose requirements, enabling the opti-

mal acquisition protocol to be implemented on the

scanner [8,9].

Technological advances and trends

Since its conception in the early 1970s CT has

undergone an enormous metamorphosis. The first CT

scanners [1] produced a 1 cm thick slice in about

4 minutes. Current scanners can produce 16 slices

at submillimeter resolution in less than 1 second

[10–13]. The following sections describe what the

authors believe are currently the major trends in

CT scanner technology development.

Spatial resolution

Today, modern medical CT scanners can resolve

objects that are less than 1 mm in size. Nevertheless,

many applications would benefit from the ability to

visualize even smaller structures. In particular, clini-

cians are interested in isotropic spatial resolution (ie,

the longitudinal resolution is equally good as the in-

plane resolution). Isotropic resolution allows the

clinician to visualize reconstructed data along various

planar reformats without loss of detail in the images.

The spatial resolution of a CT scanner is limited

by several factors: the effective size of the x-ray tube

focal spot, the size of the individual detector cells, the

data sampling pattern implemented by the imaging

geometry, and the stages in the reconstruction algo-

rithm. It is important to realize that spatial resolution

D. Walter et al / Thorac Surg Clin 14 (2004) 135–149 137

is not equal to the image pixel/voxel size but is an

intrinsic property of the scanner. Therefore, the op-

erator must adjust the pixel size by appropriately

selecting the field of view of the reconstruction so

that it is not a limiting factor of spatial resolution. To

develop CT scanners that have increased spatial

resolution, one has to reduce the focal spot size,

reduce the detector cell size, and increase the number

of angular positions at which x-ray projection data are

acquired. The increased number of detector channels,

the increased number of views, and the increased

number of reconstructed image voxels result in much

larger datasets, meaning larger computation times and

larger storage requirements. To maintain an accept-

able noise level at higher spatial resolution, one has to

increase the x-ray flux, resulting in an increased

patient dose (see section on dose reduction in this

article). Using a smaller focal spot size limits the

maximum output flux from the x-ray tube for thermal

reasons, resulting in higher image noise.

Researchers at the GE Global Research Center

developed a number of research CT scanners recently

based on a flat panel detector consisting of a grid of

1024 � 1024 detector cells, each 200 mm � 200 mmin size, to explore the clinical impact of higher spatial

resolution in combination with large volume cover-

age [14]. Fig. 2A shows the image of a plastinated

dog lung phantom scanned on a Lightspeed clinical

scanner (GE Medical Systems, Milwaukee, Wiscon-

sin). Fig. 2B shows GE’s volumetric CT research

system. The improvement in the image sharpness

Fig. 2. High-resolution axial image of plastinated dog lung phantom

1.8 cm2. (a) This image was obtained with a GE Lightspeed system

helical pitch of 1.5:1, and reconstructed with the bone kernel. (b) Th

volumetric CT prototype system capable of 250 mm isotropic re

comparing features of similar size, the high-resolution volumetric

achieved in future commercial CT scanners. Because the lung tissu

with air. (CT image courtesy of Rebecca Fahrig, PhD Stanford U

courtesy of GE Global Research Center, Niskayuna, NY; phantom

demonstrates the resolution improvement that is pos-

sible in future CT scanners.

Volume coverage

From CT’s conception until the late 1990s, CT

data have always been acquired one slice at a time.

One exception was CT Twin (Elscint, Haifa, Israel), a

dual-slice CT scanner made in 1992. It was not until

1998 that four manufacturers (GE, Siemens, Toshiba,

and Marconi) manufactured a four-slice CT scanner

simultaneously. The most recent commercial scanners

allow the acquisition of 16 slices simultaneously

[10–13]. The basic principle behind multislice CT in-

volves stacking a number of detector rows longitu-

dinally (Fig. 3) [15,16], which explains why some

people use the term multidetector instead of multi-

slice, although strictly speaking it is still one detector

consisting of a two-dimensional array of detector

cells. Typically, the detector consists of a larger num-

ber of detector rows that are binned together longi-

tudinally depending on the desired slice thickness.

Some manufacturers prefer to use thinner detector

rows in the center and thicker detector rows at the

ends; other manufacturers prefer to use detector rows

of equal size.

The main advantage of multislice scanners is the

ability to scan a given volume in a shorter time, or

vice versa, scan a larger volume within a given time.

This ability has an enormous impact on clinical

applications, as discussed in more detail in the

. Both images represent a region of interest of approximately

at 120 kVp, 100 mA, 4 � 1.25 mm detector collimation, a

e image is the same phantom imaged on a GE high-resolution

solution. The two images are not registered; however, by

CT image demonstrates the resolving power that could be

e was dried before plastination, all vessels appear to be filled

niversity, Stanford, CA; high-resolution volumetric image

developed by Dr. Robert Henry, University of Tennessee.)

Fig. 3. Comparison of single-slice and multislice CT scan-

ner. A single-slice CT scanner has only one detector row. In

multislice CT, the detector consists of multiple rows that are

stacked longitudinally and can be configured in one of sev-

eral topologies for various choices of slice thickness. The

x-ray source illuminates multiple rows simultaneously. A

16-slice scanner can read out 16 slices simultaneously.


clinical imaging applications section of this article.

Detectors that have wider coverage could eventually

allow organ imaging in a fraction of the time required

today. This advantage could remove the demand for

faster gantry rotational speeds, and it would enable

the imaging of a whole organ for perfusion studies in

a single rotation.

One issue that needs to be taken into account when

using larger scan coverage per rotation is the in-

creased cone angle (see Fig. 1), the longitudinal angle

at which the source covers the detector. The cone

angle has two implications. First, the relative amount

of scattered radiation increases almost linearly with

the size of the cone angle. Second, and perhaps more

important, as the cone angle increases, it becomes

more difficult to reconstruct high-quality images

based on the measured data. For smaller cone angles

such as those of 16-slice scanners, the algorithms for

single-slice CT are still sufficiently accurate, at least

with some minor modifications. The most famous

adaptation is the so-called Feldkamp reconstruction

approach [17], in which the classic fan beam algo-

rithm is adapted to take into account the true geome-

try of the cone beam acquisition; however, these

adaptations are still approximations, and for larger

cone angles they become too inaccurate, resulting in

cone beam image artifacts. For larger cone angles

with axial (nonhelical) CT scans it becomes mathe-

matically impossible to perform an exact reconstruc-

tion because of incomplete data.

Two solutions have been proposed to handle this

problem. The first solution is to acquire complete data

using circle-plus-line or circle-plus-arc trajectories, in

which a small linear scan accompanies each rotation,

similar to a scout scan in conventional CT [18]. The

second solution is to take large cone angle helical CT

acquisitions, in which exact reconstruction remains

theoretically possible. Tuy [19] and Grangeat [20]

showed how cone beam measurements could be used

for mathematically exact image reconstruction. Based

on this work several exact, wide cone beam algo-

rithms have been published recently [21–23]. For

ease of implementation, a number of approximate

algorithms have been proposed [24,25]. With the

prospect of CT geometries with larger and larger

cone angles, research in cone beam reconstruction

has become an active field in the past few years.

Temporal resolution

Another trend in CT is the ever-increasing acqui-

sition speed, which is important for two reasons: (1)

to avoid motion artifacts caused by, for instance,

cardiac motion, breathing, or patient movement, and

(2) to reduce the total scan time for one examination.

Reducing the scan time prevents slice misregistration,

facilitates studies with contrast agents, and decreases

patient examination time.

The straightforward way to improve temporal

resolution is to increase the gantry rotation speed;

however, rotating a 1 ton gantry at two rotations per

second or more results in extremely challenging me-

chanical constraints. Another problem is that large

amounts of data have to be transmitted in a shorter

time. Finally, and most important to the clinician,

faster scanning limits the total number of x-ray

photons that the x-ray tube can deliver per scanned

slice, resulting in images that have increased noise.

One way to overcome the speed limitation is to use a

large, stationary circular detector, replace the x-ray

tube with an electron gun, producing a beam of elec-

trons directed on a large stationary circular target, and

sweep this electron beam around the target to scan the

patient. This technique is the basis of the EBCT

scanner [7,26]. The absence of moving parts allows

scan speeds equivalent to a ‘‘rotation’’ time of about

50 ms, or 10 � faster than conventional CT scanners.

The most challenging application for CT and other

noninvasive imaging modalities is cardiac imaging

because of the dynamic nature of the heart. For car-

diac imaging, particularly CT angiography, there is a

demand for improved image quality (ie, better spatial

and temporal resolution, whole-heart scanning in one

breath-hold, improved signal-to-noise ratio, and low-

contrast detectability for determining functional in-

formation). These characteristics drive the technology

D. Walter et al / Thorac Surg

toward faster rotational speeds for the gantry, wider

coverage, more efficient detectors, and advanced

reconstruction techniques. The volume of the heart

changes drastically during the cardiac cycle. There-

fore, if the imaging system could acquire the neces-

sary projection data for reconstruction of the heart

during a phase of minimum motion, the resulting im-

age quality would contain reduced motion artifacts.

High temporal resolution is required to freeze the

heart motion, and high spatial resolution is needed to

allow clinicians to identify and characterize coronary

arteries, enabling visualization and quantification of

stenotic segments and discernment of constituent

components of atherosclerotic plaque that might be

present [27–30]. Fig. 4 shows an image of resteno-

sis in a stent of the proximal left anterior descend-

ing coronary artery from data acquired with a

multislice CT system at 0.625 mm slice thickness

(Lightspeed 16, GE Medical Systems, Milwaukee,

Wisconsin) and reconstructed with a multisector

approach. Sectors of x-ray projection from multiple

rotations of the gantry are combined to provide the

data necessary for image reconstruction at 0.3 mm

intervals with overlapping slices [28,29,31].

Fig. 4. Cardiac imaging using a multisector reconstruction

approach. The image shows restenosis in a stent placed in

the proximal left anterior descending coronary artery. Data

were acquired with a clinical CT scanner configured with

0.625 mm slice thickness (GE Lightspeed 16) and recon-

structed with a multisector approach at 0.3 mm intervals with

overlapping slices. Using state-of-the-art CT imaging

technology and multisector reconstruction algorithms,

today’s scanners have capabilities that approach the spatial

resolution and the temporal resolution needed to evaluate the

coronary arteries of the heart. The heart rate of this patient

during this examination ranged between 71 beats per minute

and 82 beats per minute. (Courtesy of J.L. Sablarolles, MD,

Centre Cardiologique Du Nord, Saint Denis, France.)

Dose reduction

No aspect of medical CT imaging has received

more attention in recent years than patient radiation

dose [32–34]. For diagnostic imaging in general, the

radiation dose should be kept to as low as reasonably

achievable while maintaining suitable diagnostic ca-

pability for features of interest. Radiation dose mea-

surements are reported in the CT–dose index (CTDI,

in Gy), dose– length–product (in mGy-cm), and

effective dose (in mSv). CTDI standards are useful

to objectively rate different scan protocols available

in a scanner and among different types of scanners,

but the dose index is not useful in predicting the

image quality (ie, noise level) produced in anatomical

imaging or the risk to the patient. The computation

of effective dose is used to rate the relative risk to

patients from radiologic procedures [35,36]. For the

calculation of this quantity, the absorbed dose is cal-

culated in specific organs or tissues and is weighted

by the relative sensitivity of these tissues to x-ray

radiation to predict risk to the patient (ie, tissue

weighting factors).

The noise and the quality of the diagnostic infor-

mation available in an image depend on the interplay

between the data acquisition parameters (eg, helical

pitch, detector collimation, tube current, tube voltage,

x-ray beam filtering), the reconstruction approach

(eg, helical weighting, reconstruction kernel, longitu-

dinal smoothing), and the postprocessing techniques

(eg, thresholding, segmentation, volumetric measure-

ments). Image noise also depends on the patient’s

anatomy. In modern CT scanners, the detector’s

efficiency (its ability to absorb an x-ray of a certain

energy) is near its theoretical limit; therefore, other

options must be explored to reduce patient doses.

Several techniques have been suggested to opti-

mize the information available in CT imaging or to

reduce the dose delivered to the patient. Two general

approaches are to: (1) reduce the dose by modulating

the intensity of the x-rays, and (2) use the measure-

ments in an optimal scheme to improve the diagnostic

information for a given dose.

One of the latest features on some multislice

scanners is the ability to modulate the tube current

for different sections of the body. The x-ray tube cur-

rent is increased along lateral directions in the patient,

where there is significant attenuation (eg, along the

shoulders), and reduced along anterior–posterior

directions, where less attenuation is encountered as

the x-ray tube and detector rotate around the body.

The x-ray tube current is also varied in a global

sense as the patient is translated horizontally with

the patient table in the helical scanning mode as the

Clin 14 (2004) 135–149 139


natural contour of the human body increases or de-

creases. In both cases the x-ray tube current is ad-

justed appropriately to meet a noise requirement that

is set by the radiologist (see [37] and references

therein). These techniques have been shown to reduce

the dose delivered to the patient significantly, but it

is critical to set an appropriate noise factor that does

not inhibit diagnostic information. In the future all

CT scanners will use such techniques.

One way to optimize dose usage is to make the

measured intensities more uniform within each x-ray

projection (denoted as a view), which is accomplished

by using specially shaped filters (eg, bowtie filters).

Central rays usually traverse larger sections of the

patient and are therefore filtered minimally by the

bowtie filter. Peripheral rays in the x-ray beam usually

traverse smaller sections of the patient and are there-

fore filtered more. The net result is that all rays

undergo roughly the same amount of attenuation. This

technique reduces the harmful absorbed dose at the

peripheral of the body [38]. As the radiological com-

munity continues to advocate lower doses, one can

conceive of special x-ray beam filtering techniques

that could be optimized for a specific organ of interest.

Current multislice helical reconstruction algo-

rithms based on the filtered backprojection method

use a variety of weighting and filtering schemes to

Fig. 5. Adaptive filtering to reduce noise artifacts for low-dose i

nonsolid pulmonary nodule (arrow). (a) Image reconstructed from x

120 kVp, 200 mAs, 0.5 s gantry speed (100 mAs), 4 � 2.5 mm de

GE Lightspeed Plus (GE Medical Systems, Milwaukee, Wisconsin)

image was reconstructed from x-ray projection data acquired usin

speed (5 mAs), 4 � 1.5 mm detector collimation, and a helical pit

generate the image shown in (b) was reconstructed using a reconstr

In all three images the nodule is visible; however, the image sho

techniques, contains less noise artifact when compared with the im

compared with the image in (a). (Courtesy of Akifumi Fujita, MD

trade off noise and artifacts in the reconstructed

image with the achievable spatial resolution. A par-

ticularly difficult area in the body for low-dose

imaging is the region near the shoulders. These slices

usually suffer from structured noise, which appears as

horizontal streaks across the image, because of the

significant attenuation of the x-ray signal through the

shoulders. Various schemes have been proposed to

improve the signal-to-noise ratio of measurements

acquired along projection lines with significant at-

tenuation by using information from neighboring

detector cells or views. In some cases special filtering

schemes also incorporate signals measured in the

longitudinal direction from data measured with the

multislice detector [24,39,40]. To demonstrate one

such scheme, Fig. 5 displays a scan of the right lung

of a human patient. On the left (Fig. 5a) the chest was

scanned using an x-ray beam current of 200 mA. In

the middle (Fig. 5b) the same anatomy is scanned with

much lower dose, using an x-ray current of 10 mA. On

the right (Fig. 5c) the data used to generate the image

shown in Fig. 5b are reconstructed using an adaptive

filtering scheme incorporated as part of the recon-

struction algorithm [39]. Although the adaptive filter-

ing algorithm reduces spatial resolution slightly, as

demonstrated in Fig. 5c, the benefits of reduced image

noise are shown clearly.

maging. The scan of the right lung of this patient shows a

-ray projection data acquired using a system configuration of

tector collimation, and a helical pitch ratio of 0.75:1 using a

. (b) The same anatomy was scanned at much lower dose; the

g a system configuration of 120 kVp, 10 mA, 0.5 s gantry

ch ratio of 1.5:1. (c) The same x-ray projection data used to

uction algorithm incorporating adaptive filtering techniques.

wn in (c), which was reconstructed using adaptive filtering

age in (b), at a slight reduction in the spatial resolution when

, Showa University, Fujigaoka Hospital.)


Another strategy to reduce noise in CT images or

to achieve the same noise level at reduced dose is the

use of statistical reconstruction algorithms, which

weigh x-ray projection data in an optimal sense for

CT image reconstruction. Although statistical re-

construction algorithms have been investigated in

nuclear medicine for many years, they have not yet

been commonly used for CT image reconstruction

mainly because of the computational complexity of

such approaches and the fact that the signal-to-noise

ratio in x-ray projection measurements is higher in

CT compared with nuclear imaging. Unlike filtered

backprojection reconstruction approaches, statistical

reconstruction has a framework that includes a statis-

tical model for the measurements. More recently, sev-

eral researchers have applied statistical reconstruction

to CT successfully [41–44]. Statistical reconstruction

also allows incorporation of prior information regard-

ing the scanned region of interest (ROI) to improve

image quality in reconstructed images in an efficient

manner. Apart from minimizing image noise, statisti-

cal reconstruction has also been shown to have

potential for artifact reduction [42,43] and for appli-

cations with missing x-ray projection data needed for

image reconstruction [45–47].

Clinical imaging applications

New scanners that have improved algorithms and

technology offer superior image resolution, greater

coverage, and shorter acquisition times. As a result of

advancing CT imaging technology, the amount of

information required for physicians to review has also

increased. To improve clinicians’ workflow and pro-

ductivity, three-dimensional image processing algo-

rithms, advanced workstations, and computer-assisted

tools have been developed. Advanced postprocessing

tools have many applications and have been shown to

improve accuracy and productivity [48,49]. Some

specific examples of advanced CT technology com-

bined with computer-assisted tools are discussed in

this section.

Lung cancer

The American Cancer Society has estimated that

171,900 new lung cancer cases would be diagnosed

and 157,200 lung cancer deaths would occur in the

United States in 2003 [50]. CT imaging currently

plays an important role in all aspects of lung can-

cer management including detection, diagnosis, stag-

ing, treatment planning, and patient follow-up care

[51–61]. In the past decade, spatial resolution achie-

vable with state-of-the-art CT scanners has increased

more than tenfold. CT now routinely provides detailed

images of early-stage lung cancer by imaging with a

spatial resolution of less than 1 mm. High-resolution

imaging has led to significant advances in clinical

understanding of early disease progression and has

redefined lung cancer management practices. This

section focuses on state-of-the-art and emerging CT

applications in the area of lung cancer and describes

the impact of future technological improvements.

Solitary pulmonary nodule detection and sizing

The superiority of CT over other radiologic

imaging techniques (eg, chest radiograph, MRI) for

the detection of solitary pulmonary nodules (SPNs)

is well documented [62–64]. Recent technological

advancements in multislice CT, namely the ability

to scan the entire human thorax at submillimeter

resolution within a breath-hold, enable the detection

of much smaller lesions [63,65]. This realization has

spurred several studies to reinvestigate the hypothesis

that the early detection of lung cancer could justify

screening for lung cancer in asymptomatic patients

(see [51] for a review of current lung cancer screen-

ing literature).

In addition to enabling the detection of small

SPNs, high-resolution CT imaging has also been

shown to be a highly sensitive for the measurement

of small changes in volume [61,66,67]. Radiologic

phantom studies, used to characterize the sensitivity of

high-resolution CT imaging for growth estimation,

indicate that volume accuracy using CT is within 3%

of the true volume in nodules between 3 and 6 mm in

diameter and within 1% in nodules greater than 6 mm

in diameter [61]. These results were obtained under

ideal imaging conditions: small field-of-view imag-

ing, application of resolution-enhancing reconstruc-

tion kernels, thin-slice imaging mode, low helical

pitch, and the absence of confounding anatomical

structure. These results should therefore be considered

an upper limit of current CT capability. The clinical

evidence for in situ volumetric accuracy of nodule

sizing is not conclusive for several reasons: (1) nodule

sizing methods have not been standardized, (2) the

highest available resolution capabilities of the scanner

are not routinely used for examinations, and (3) ad-

vanced segmentation and computation tools have only

become recently available.

In general, three-dimensional computer-assisted

segmentation tools generate the most accurate and

repeatable nodule size measurements [53]. An exam-

ple case showing results from the use of advanced

analysis tools is presented in Figs. 6 and 7. Fig. 6a

shows a CT image of an SPN attached to a vascu-

Fig. 6. Segmentation of pulmonary nodules with minimal user interaction. An SPN is detected by the radiologist, and using

advanced analysis tools (Advanced Lung Analysis, GE Medical Systems, Milwaukee, Wisconsin) the nodule is segmented

automatically and the volume is determined. (a) Solid pulmonary nodule detected by the radiologist in an axial CT scan.

(b) The Nodule and vessel structure are segmented from the parenchyma and displayed as a three-dimensional surface.

(c) Nodule is separated automatically from the vascular structures and its volume is computed. (Courtesy of Lawrence Good-

man, MD, Froedtert Hospital, Milwaukee, WI.)


lar structure. Fig. 6b shows the nodule and vessel

structure, which have been segmented from the lung

parenchyma and displayed as a three-dimensional

surface. The nodule was separated automatically

from the vascular structure and the volume was com-

puted. The three-dimensional surface model of the

Fig. 7. Determining nodule growth rate with volumetric measurem

shown. The volume at each time instance was measured using

(a) Using the CT image of the nodule measured at time one, the di

nodule measured one month later is 12.8 mm in diameter. (c) Thre

one. (d) Three-dimensional surface model of the same nodule mea

volume, corresponding to a 103-day doubling time assuming expo

Program, New York Presbyterian Hospital, Weill Cornell Medical

nodule separated from the vascular structure is shown

in Fig. 6c.

The images in Fig. 7 show a nodule that was mea-

sured in two CTscans one month apart; the nodule size

exhibits a 24% volume increase, corresponding to a

103-day doubling time, assuming exponential growth.

ents. Multiple images of a nodule, measured 1 mo apart, are

GE Medical System’s Advanced Lung Analysis program.

ameter of the nodule is shown to be 11.7 mm. (b) The same

e-dimensional surface model of the nodule measured at time

sured at time two. This nodule exhibited a 24% increase in

nential growth. (Courtesy of The Early Lung Cancer Action

Center, New York, NY.)

Fig. 8. CAD used to detect lung nodules automatically. In

this image a radiologist detected two potential lung cancers

in the right lung (arrows). One lung nodule is a solid type

(left arrow); another nodule is a nonsolid type (right arrow).

Both of these nodules were also detected using a CAD

algorithm developed at the GE Global Research Center.

The nodule on the left is difficult for most CAD algorithms

to detect because it has structures attached to the chest

wall. The nodule on the right is challenging because the

reconstructed intensity is lower than most solid lung

nodules. The patient was scanned using a GE Lightspeed

system at 120 kVp, 60 mA, and 2.5 mm slice thickness.

(Courtesy of The Early Lung Cancer Action Program, New

York Presbyterian Hospital, Weill Cornell Medical Center,


Another revelation in the understanding of

the characteristics of SPNs relates to the increased

frequency of detection of subsolid lesions (or ground-

glass opacities) [55,68,69]. The importance of sub-

solid lesions has recently become apparent because

nonsolid and part-solid lesions appear to be highly

correlated to malignancy. In one study, nonsolid

nodules comprised 52% of all confirmed lung can-

cers, and their probability of malignancy was between

18% (for nonsolid nodules) and 63% (for part-solid

nodules), compared with a 7% malignancy rate

among strictly solid SPNs [69]; however, it should

be noted that the prevalence of nonsolid and part-

solid lesions is much lower than the prevalence of

solid lesion types. High-resolution CT imaging will

play an important role in the further investigation of

these lesion types because many of the most subtle

occurrences of these lesion types are not detectable

when thick-slice CT scanning is used because of

partial volume effects.

Computer-automated detection of pulmonary nodules

The evaluation of today’s high-resolution MSCT

data presents a significant challenge to radiologists

because they must review up to 500 CT images for a

single examination and discriminate normal lung

anatomy from small lesions that potentially represent

the early onset of lung cancer. The promise of CT for

high-resolution imaging of early lung cancer coupled

with the challenges in reviewing these large data sets

has led to an increased focus on computer-aided de-

tection (CAD) of lung lesions [70–74]. In Fig. 8,

two nodules identified using an automated detection

algorithm are shown. The nodule identified by the left

arrow is a solid type; the nodule identified by the right

arrow is a nonsolid type. Both nodules present a

challenge for automated detection algorithms—the

solid nodule because it has structures connecting it

to the chest wall; the nonsolid nodule because it has a

lower reconstructed attenuation value when compared

with solid nodules. It is initially expected that lung

cancer CAD techniques will improve radiologic diag-

nosis by highlighting suspicious regions that the

radiologist might otherwise have overlooked. A recent

multireader, multi-institution study for lung cancer

detection demonstrated an average reduction in the

false-negative rate from 10% to 3% when comparing

the performance of a radiologist alone to that of a

radiologist using a prototype CAD algorithm [75].

As the spatial resolution and amount of data per

patient examination continues to increase, CAD tech-

nology will become a necessary tool in the radiolo-

gist’s evaluation of chest CT examinations.

CT for noninvasive diagnosis, staging, and treatment

planning

Because of the reduced scanning time for a CT

examination using advanced MSCT, there has been a

renewed interest in the use of contrast-enhanced CT

imaging for the diagnosis and staging of lung nodules

[76–78]. The use of iodine contrast agent uptake as a

measure of malignancy is predicated on the fact that

there are distinct differences in the vascular charac-

teristics of benign and malignant nodules. Although

positron emission tomography (PET) is more accurate

than CT imaging for the detection of mediastinal

metastases [79], the role of CT imaging is still recom-

mended as a diagnostic tool because it is valuable for

further workup, including biopsy or surgery planning

[80]. Furthermore, the role of CT and PET image

fusion has shown considerable promise as an optimal

method for noninvasive diagnosis and staging of lung

cancer [81,82]. CT imaging also plays a primary role

in radiation treatment planning [53].

New York, NY.)

Fig. 9. High-resolution volumetric CT for improved air-

way wall measurements. A high-resolution volumetric CT

research scanner (GE Global Research Center, Niskayuna,

NY) capable of 250 mm isotropic resolution was used to

image an excised pig lung. The airway lumen (insert) was

measured with minimal user interaction using prototype

software (GE Global Research Center, Niskayuna, NY) and

was measured to be about 0.5 mm in diameter.

Surg Clin 14 (2004) 135–149

Chronic obstructive pulmonary disease

Chronic obstructive pulmonary disease (COPD)

affects roughly 16 million adults in the United States

and is the fourth leading cause of chronic morbidity

and mortality in the United States [83]. High-resolu-

tion CT imaging has been useful in the evaluation of

the presence and extent of emphysema and as a tool

to quantify morphologic changes caused by chronic

bronchitis [84]. Many of these recent advances have

been spurred by the technological improvements in

MSCT, namely the ability to acquire a large volume

of contiguous high-resolution, thin-slice data in less

than 10 seconds.

The superiority of CT imaging over chest radiog-

raphy in the detection and evaluation of the preva-

lence of bullae in the lung and the diagnosis of

emphysema has been well documented [84,85].

Researchers have commented on qualitative methods

of assessing the severity of emphysema with CT

imaging, correlating radiologic indications with lung

function [86–88]. Recent research in the radiologic

assessment of emphysema has focused on the devel-

opment of computational tools that can objectively

evaluate the extent of emphysema through the mea-

surement of lung density [89,90]. Because lung

density measurements can be affected by scanner

calibration, reconstruction parameters [91,92], patient

size, and depth of inspiration, the key to reducing

inter- and intravariability of these examinations is to

model and correct for these effects.

The ability to use CT imaging to detect bronchial

wall thickening caused by chronic bronchitis (which

is associated with reduced lung function in the case

of COPD and asthma) has been well documented in

the recent literature [84,93]. The ability to measure

small changes in the airway lumen area and in the air-

way wall area is at the limit of the resolving power of

state-of-the-art CT scanners. Because of the scanner’s

inherent resolution capabilities, quantitative measure-

ment accuracy is limited by the partial volume effect

[94]. Several reports have shown that computational

methods are more accurate than manual methods

[93,95,96], and methods that correct for the scanner

point spread function, or resolving power of the CT

system, have a higher rate of repeatability [97].

The availability of thinner slices on state-of-

the-art scanners (eg, 0.625 mm) and high-resolution

scanning protocols could further improve the accu-

racy of airway measurements. As resolution in CT

scanners continues to improve, the ability to measure

airways less than 1 mm accurately will become

possible, and the sensitivity of the detection of mor-

phologic changes will increase. For example, a high-

D. Walter et al / Thorac144

resolution CT image of an excised pig lung recon-

structed from data acquired with a prototype high-

resolution volumetric CT research system is shown in

Fig. 9. The lumen of an airway that is approximately

0.5 mm in diameter is clearly visible.

Pulmonary embolism

It is estimated that the incidence of pulmo-

nary embolism (PE) in the United States is roughly

630,000 cases per year [98]. Since its introduction,

[99] CT imaging has been shown to be highly sen-

sitive and specific in the diagnosis of PE when com-

pared with angiography, which is the current standard

[100,101]. Recent technological advances in reso-

lution (thinner slices) and speed (faster scans) of

multislice CT scanners have improved the detection

of PE [102].

CT offers several advantages over other imaging

techniques for PE detection. The primary advantage

of CT imaging is that scanning can be performed

quickly in critically ill patients. Clinicians also have

the ability to evaluate patients using CT for other

lung-related conditions that could indicate an in-

creased risk of PE such as cardiac disease, chest

trauma, pneumonia, and lung cancer.

One impediment to detecting PE in CT images is

that the current resolution capability limits the detec-

tion of subsegmental PEs [103], although the clinical

importance of such cases is a subject of debate. As

Fig. 10. Multislice CT for the detection of PE. (a) On the left, an axial CT image shows an occurrence of PE (arrow).

(b) Maximum-intensity projection image of the volume is shown on the right in sagittal reformat. The PE (arrow) is readily

detectable. The patient was imaged using a GE Lightspeed 16. (Courtesy of Lawrence Goodman, MD, Froedtert Hospital,

Milwaukee, WI.)


resolution continues to improve and advanced MSCT

technology becomes more available, these limitations

could possibly be surmounted, as suggested by some

researchers [102,104]. Some misdiagnoses are caused

by the difficulty in quickly finding an abnormality in

a complicated vessel structure [105]. To review the

entire chest in high resolution, the radiologist must

review between 100 and 300 axial images. This task

might be facilitated with advanced segmentation

techniques. For example, an occurrence of a PE is

shown in an axial image (Fig. 10a), but in a maxi-

mum-intensity projection image of a sagittal cross-

section (Fig. 10b), the PE is readily detectable and

there are considerably less images to review. Some

researchers have suggested using CAD methods to

detect PE in CT images rapidly [106,107].

Interventional procedures

CT imaging has been a standard in surgery plan-

ning, radiation therapy planning, virtual endoscopy,

and guidance for percutaneous needle biopsy. Image-

guided surgery, a procedure in which the physician

uses a two- or three-dimensional image along with

the registration of the surgical instruments in the re-

constructed volume to guide the procedure, is at the

forefront of medical technology. Two-dimensional

imaging techniques are limited because the surgeon

must virtually conceptualize the three-dimensional

and correlate the image with the anatomy. The use

of preoperative CT images and three-dimensional

visualization offers a more natural view of the anat-

omy, creating new possibilities for more minimally

invasive techniques. Procedures for use in the sinus,

spine, and head are becoming more widespread

[108–110].

Of particular interest in surgical treatment for

oncology is the use of percutaneous, image-guided,

in situ tumor ablation techniques [111]. The most

advanced of these techniques uses a radiofrequency

(RF) thermal source positioned by way of an elec-

trode placed with image guidance into the tumor to be

ablated. Although the standard of care for a detected

lung cancer is resection, some researchers have pro-

posed RF ablation for pulmonary nodules in patients

who are not candidates for surgery [112–114].

One barrier to extending these techniques to other

areas in the thorax is the fact that the chest undergoes

significant movement during the breathing cycle,

which makes it difficult to register the current location

of the electrode with a preoperative scan or requires

surgical procedures to be performed in the vicinity of

the CT scanner. The development of deformable

models [115,116], which can predict the organ move-

ment, advanced visualization techniques [117], and

multimodality data integration, will extend these

techniques in the future.

Summary

X-ray CT technology has been available for more

than 30 years, yet continued technological advances

have kept CT imaging at the forefront of medical


imaging innovation. Consequently, the number of

clinical CT applications has increased steadily. Other

imaging modalities might be superior to CT imaging

for some specific applications, but no other single

modality is more often used in chest imaging today.

Future technological developments in the area of

high-resolution detectors, high-capacity x-ray tubes,

advanced reconstruction algorithms, and improved vi-

sualization techniques will continue to expand the im-

aging capability. Future CT imaging technology will

combine improved imaging capability with advanced

and specific computer-assisted tools, which will ex-

pand the usefulness of CT imaging in many areas.

Acknowledgments

The authors thank Rick Avila, Bob Senzig, Samit

Basu, and Rajiv Gupta for their useful comments

and insights.

References

[1] Hounsfield G. Computerized transverse axial scan-

ning (tomography): part 1. Description of system.

Br J Radiol 1973;46:1016–22.

[2] Cormack A. Representation of a function by its line

integrals, with some radiological applications. J Appl

Phys 1963;34(9):2722–7.

[3] Mori I. 1986. US Patent No. 4630202.

[4] Nishimura H. 1988. US Patent No. 4789929.

[5] Kalender WA, Seissler W, Klotz E, Vock P. Spiral

volumetric CT with single-breath-hold technique,

continuous transport, and continuous scanner rota-

tion. Radiology 1990;176(1):181–3.

[6] Crawford CR, King KF. Computed tomography scan-

ning with simultaneous patient translation. Med Phys

1990;17(6):967–82.

[7] Boyd D, Lipton M. Cardiac computed tomography.

Proc IEEE 1983;71:298–307.

[8] McCollough CH. Optimization of multi-detector ar-

ray CT acquisition parameters for CT colonography.

Abdom Imaging 2002;27(3):253–9.

[9] Honda JT, Yamamoto S, Tomiyama N, Koyama M,

Kozuka T, Mihara N, et al. Evaluation of image qual-

ity and spatial resolution of low-dose high-pitch multi-

detector-row helical high-resolution CT in 11 autopsy

lungs and a wire phantom. Radiat Med 2001;19(6):

279–84.

[10] Medical Systems GE. Available at: http://www.

gemedicalsystems.com/rad/ct/products/light_series/

ls16. Accessed January 6, 2004.

[11] Siemens. Available at: http://www.siemens.com.

Accessed January 6, 2004.

[12] ToshibaMedical Systems. Available at: http://medical.

toshiba.com/clinical/cardiology/aquilionmulti16.htm.


[13] Philips Medical Systems. Available at: http://www.

medical.philips.com. Accessed January 6, 2004.

[14] Ross WR, Basu S, Edic PM, Johnson M, Pfoh A, Rao

R, et al. Design and image quality results from volu-

metric CT with a flat-panel imager. In: Antonuk LE,

Yaffe MJ, editors. Proceedings of SPIE Medical Im-

aging. San Diego: SPIE International Society for Op-

tical Engineering; 2001. p. 783–91.

[15] Hu H. Multi-slice helical scan and reconstruction.

Med Phys 1999;26(1):5–18.

[16] Kalender WA, Fuchs T. Principles and performance

of single- and multislice spiral CT. In: RSNA cate-

gorical course in diagnostic radiology physics: CT

and US cross-sectional imaging. 2000. p. 127–42.

[17] Feldkamp LA, Davis LC, Kress JW. Practical cone

beam algorithm. J Opt Soc Am 1984;A(1):612–9.

[18] Hu H. A new cone beam reconstruction algorithm for

the circle-and-line orbit. In: Nuclear Science Sympo-

sium and Medical Imaging Conference, Volume 3.

IEEE; 1994. p. 1261–5.

[19] Tuy HK. An inversion formula for cone-beam recon-

struction. SIAM J Appl Math 1983;43:546–52.

[20] Grangeat P. Analyse d’un systeme d’imagerie 3D par

reconstruction a partir de radiographies X en geome-

trie conique. PhD thesis, Ecole Nationale Superieure

des Telecommunications, France, 1987. [in French].

[21] Schaller S, Noo F, Sauer F, Tam K, Lauritsch G, Flohr

T. Exact radon rebinning algorithm for the long object

problem in helical cone-beam CT. IEEE Trans Med

Imaging 2000;19(5):361–75.

[22] Defrise MC. A cone-beam reconstruction algorithm

using shift-variant filtering and cone-beam back-

projection. IEEE Trans Med Imaging 1994;13(1):

186–95.

[23] Katsevitch A. Analysis of an exact inversion algorithm

for spiral cone-beam CT. Phys Med Biol 2002;47:

2583–97.

[24] Kachelriess M, Schaller S, Kalender WA. Advanced

single-slice rebinning in cone-beam spiral CT. Med

Phys 2000;27(4):754–72.

[25] Turbell H, Danielsson P. Helical cone-beam tomogra-

phy. Int J Img Sys Tech 2000;11:91–100.

[26] Medical Systems GE. GE Medical Systems Car-

diology EBT series. Available at: http://www.

gemedicalsystems.com/cardiology/ebt/index.html.


[27] Sablayrolles JL, Besse F, Giat P. Technical develop-

ments in cardiac CT: 2000 update. Rays 2001;26(1):

3–13.

[28] Kachelriess M, Ulzheimer S, Kalender WA. ECG-cor-

related image reconstruction from subsecond multi-

slice spiral CT scans of the heart. Med Phys 2000;27:

1881–902.

[29] Pan T, Shen Y. Multi-sector reconstruction for multi-

slice cardiac (CT). Radiology 2000;217:161.

[30] Achenbach S, Giesler T, Ropers D. Detection of

coronary artery stenoses by contrast-enhanced, retro-

http:\\www.gemedicalsystems.com\rad\ct\products\light_series\ls16

http:\\www.siemens.com

http:\\medical.toshiba.com\clinical\cardiology\aquilionmulti16.htm

http:\\www.medical.philips.com

http:\\www.gemedicalsystems.com\cardiology\ebt\index.html


spectively electrocardiographically gated, multislice

spiral computed tomography. Circulation 2001;103:

2535–8.

[31] Cesmeli E, Edic PM, Iatrou M, Pfoh A. A novel

reconstruction algorithm for multiphasic cardiac im-

aging using multislice CT. In: Antonuk LE, Yaffe

MJ, editors. Proceedings of the SPIE International

Symposium on Medical Imaging. San Diego: SPIE,

International Society for Optical Engineering; 2001.

p. 645–54.

[32] Huda W, Chamberlain C, Rosenbaum A, Garrisi W.

Radiation doses to infants and adults undergoing head

CT examinations. Med Phys 2001;28(3):393–9.

[33] Morin R, Gerber T, McCollough C. Radiation dose

in computed tomography of the heart. Circulation

2003;107:917–22.

[34] McCollough C, Zink F. Performance evaluation of

a multi-slice CT system. Med Phys 1999;26(11):

2223–30.

[35] (ICRP) ICoRP. 1990 Recommendations of the In-

ternational Commission on Radiological Protec-

tion. ICRP Publication 60. Oxford (UK): Pergmon

Press; 1991.

[36] McCollough CH, Schueler BA. Calculation of effec-

tive dose. Med Phys 2000;27(5):828–37.

[37] Toth T, Limkeman MK. 1997. US Patent No.

5625662.

[38] Harpen M. A simple theorem relating noise and pa-

tient dose in computed tomography. Med Phys 1999;

26(11):2231–4.

[39] Hu H, Shen Y. Helical CT reconstruction with longi-

tudinal filtration. Med Phys 1998;25(11):2130–8.

[40] Kachelriess M, Watzke O, Kalender W. Generalized

multi-dimensional adaptive filtering for conventional

and spiral single-slice, multi-slice, and cone-beam

CT. Med Phys 2001;28(4):475–89.

[41] Nuyts J, Dupont P, Stroobants S, Maes A, Mortel-

mans L, Suetens P. Evaluation of maximum-like-

lihood based attenuation correction in positron

emission tomography. In: Nuclear Science Sympo-

sium. Toronto: IEEE; 1998. p. 1836–41.

[42] De Man B, Nuyts J, Dupont P, Marchal G, Suetens P.

An iterative maximum-likelihood polychromatic al-

gorithm for CT. IEEE Trans Med Imaging 2001;

20(10):999–1008.

[43] Elbakri IA, Fessler JA. Statistical image reconstruc-

tion for polyenergetic x-ray computed tomography.

IEEE Trans Med Imaging 2002;21(2):89–99.

[44] Sukovic P, Clinthorne NH. Data weighted vs. non-

data weighted dual energy reconstructions for x-ray

tomography. In: Nuclear Science Symposium. To-

ronto: IEEE; 1998. p. 1581–3.

[45] De Man B. 2001. Iterative reconstruction for reduc-

tion of metal artifacts in computed tomography. PhD

thesis, University of Leuven.

[46] Wang G, Frei T, Vannier MW. A fast iterative algo-

rithm for metal artifact reduction in x-ray CT. Acad

Radiol 2000;7(8):607–14.

[47] De Man B. Reduction of metal streak artifacts in

x-ray computed tomography using a transmission

maximum a posteriori algorithm. IEEE Trans Nuc

Sci 2000;47(3):977–81.

[48] Lawler LP, Fishman EK. Multi-detector row CT of

thoracic disease with emphasis on 3D volume render-

ing and CT angiography. Radiographics 2001;21(5):

1257–73.

[49] Salvolini L, Bichi Secchi E, Costarelli L, De Nicola

M. Clinical applications of 2D and 3D CT imaging

of the airways—a review. Eur J Radiol 2000;34(1):

9–25.

[50] American Cancer Society. Cancer facts and figures

2002. Available at: http://www.cancer.org.docroot/

STT/stt_0.asp. Accessed January 6, 2004.

[51] Bach PB, Kelley MJ, Tate RC, McCrory DC. Screen-

ing for lung cancer: a review of the current literature.

Chest 2003;123(90010):72S–82S.

[52] Beckles MA, Spiro SG, Colice GL, Rudd RM. Initial

evaluation of the patient with lung cancer: symptoms,

signs, laboratory tests, and paraneoplastic syndromes.

Chest 2003;123(90010):97S–104S.

[53] Hopper KD, Singapuri K, Finkel A. Body CT and

oncologic imaging. Radiology 2000;215:27–40.

[54] Mathur PN, Edell E, Sutedja T, Vergnon J-M. Treat-

ment of early stage non-small cell lung cancer. Chest

2003;123(90010):176S–80S.

[55] Nakata M, Saeki H, Takata I, Segawa Y, Mogami H,

Mandai K, et al. Focal ground-glass opacity detected

by low-dose helical CT. Chest 2002;121(5):1464–7.

[56] Patz EF. Imaging bronchogenic carcinoma. Chest

2000;117:90S–5S.

[57] Porter JC, Spiro SG. Detection of early lung cancer.

Thorax 2000;55(Suppl 1):556–62.

[58] Swensen SJ. Functional CT: lung nodule evaluation.

Radiographics 2000;20(4):1178–81.

[59] Tan BB, Flaherty KR, Kazerooni EA, Iannettoni

MD. The solitary pulmonary nodule. Chest 2003;

123(90010):89S–96S.

[60] Toloza EM, Harpole L, Detterbeck F, McCrory DC.

Invasive staging of non-small cell lung cancer: a re-

view of the current evidence. Chest 2003;123(90010):

157S–66S.

[61] Yankelevitz DF, Reeves AP, Kotis WJ, Zhao B,

Henschke CI. Small pulmonary nodules: volumetri-

cally determined growth rates based on CT evalua-

tion. Radiology 2000;217:251–6.



Cancer Action Project: overall design and findings

from baseline screening. Lancet 1999;354(9173):

99–105.

[63] Henschke CI, Naidich DP, Yankelevitz DF. Early

Lung Cancer Action Project: initial findings on repeat

screenings. Cancer 2001;92(1):153–9.

[64] Sone S, Takashima S, Li F, Yang Z, Honda T, Maru-

yama Y, et al. Mass screening for lung cancer with

mobile spiral computed tomography scanner. Lancet

1998;351(9111):1242–5.

[65] Mulshine JL, Smith RA. Lung cancer 2: screening and

http:\\www.cancer.org.docroot\STT\stt_0.asp


early diagnosis of lung cancer. Thorax 2002;57(12):

1071–8.

[66] Yankelevitz DF, Gupta R, Zhao B, Henschke CI. Small

pulmonary nodules: evaluation with repeat CT—

preliminary experience. Radiology 1999;212(2):

561–6.

[67] Winer-Muram H, Jennings S, Tarver R, Aisen A,

Tann M, Conces D, et al. Volumetric growth rate of

stage I lung cancer prior to treatment: serial CT scan-

ning. Radiology 2002;223:789–805.

[68] Kin E, Johkoh T, Lee K, Han J, Fujimoto K, Adohara

J, et al. Quantification of ground-glass opacity on

high-resolution CT of small peripheral adenocarci-

noma of the lung: pathologic and prognostic implica-

tions. AJR Am J Roentgenol 2001;177(6):1417–22.

[69] Henschke C, Yankelevitz DF, Mirtcheva R, McGuin-

ness G, McCauley DI, Miettinen OS. CT screening

for lung cancer: frequency and significance of part-

solid and nonsolid nodules. AJR Am J Roentgenol

2002;178(5):1053–7.

[70] Armato III S, Giger M, MacMahon H. Automated

detection of lung nodules in CT scans: preliminary

results. Med Phys 2001;28(8):1552–61.

[71] Niki N, Kawata Y, Kubo H, Ohmatsu M, Kakinuma

R, Kaneko M, et al. A CAD system for lung cancer

based on 3D CT images. In: Lemke HU, Vannier

MW, Inamura K, Farman AG, Doi K, Reiber JHC,

editors. Proceeding of International Congress of

Computer Assisted Radiology and Surgery. Paris:

Springer; 2002. p. 701–5.

[72] McNitt-Gray M, Hart E, Wyckoff N, Sayre J, Goldin

J, Aberle D. A pattern classification approach to char-

acterizing solitary pulmonary nodules imaged on high

resolution CT: preliminary results. Med Phys 1999;

26(6):880–8.

[73] Takizawa H, Yamamoto S, Matsumoto T, Tateno Y,

Linuma T, Matsumoto M. Recognition of lung nod-

ules from x-ray CT images using 3D MRF models.

In: Tezak EJ, Inamura K, Lemke HU, et al, editors.

Proceeding of International Congress of Computer

Assisted Radiology and Surgery. Berlin: Elsevier Sci-

ence; 2001. p. 605–14.

[74] Armato III SG, Li F, Giger ML, MacMahon H, Sone

S, Doi K. Lung cancer: performance of automated

lung nodule detection applied to cancers missed in

a CT screening program. Radiology 2002;225(3):

685–92.

[75] McCulloch C, Yankelevitz D, Henschke C, Patel S,

Kazerooni E, Sirohey S. Reader variability and com-

puter aided detection of suspicious lesions in low-

dose CT lung screening exams. Radiology 2003;

226(2):37A.

[76] Zhang M, Kono M. Solitary pulmonary nodules:

evaluation of blood flow patterns with dynamic CT.

Radiology 1997;205(2):471–8.

[77] Miles KA, Griffiths MR, Fuentes MA. Standardized

perfusion value: universal CT contrast enhancement

scale that correlates with FDG PET in lung nodules.

Radiology 2001;220(2):548–53.

[78] Swensen SJ, Viggiano RW, Midthun DE, Muller

NL, Sherrick A, Yamashita K, et al. Lung nodule en-

hancement at CT. Multicenter study. Radiology 2000;

214(1):73–80.

[79] Toloza EM, Harpole L, McCrory DC. Noninva-

sive staging of non-small cell lung cancer: a review

of the current evidence. Chest 2003;123(90010):

137S–46S.

[80] Silvestri GA, Tanoue LT, Margolis ML, Barker J, Det-

terbeck F. The noninvasive staging of non-small cell

lung cancer: the guidelines. Chest 2003;123(90010):

147S–56S.

[81] Vansteenkiste JF, Stroobants SG, Dupont P, De Leyn

PR, DeWeverWF, Verbeken EK, et al. FDG-PETscan

in potentially operable non-small cell lung cancer: do

anatometabolic PET-CT fusion images improve the

localisation of regional lymph node metastases? Eur

J Nucl Med 1998;25(11):1495–501.

[82] Magnani P, Carretta A, Rizzo G, Fazio F, Vanzulli A,

Lucignani G, et al. FDG/PET and spiral CT image

fusion for mediastinal lymph node assessment of

non-small cell lung cancer patients. J Cardiovasc Surg

[Torino] 1999;40(5):741–8.

[83] National Heart, Lung, and Blood Institute, National

Institutes of Health. Data fact sheet: chronic ob-

structive pulmonary disease (COPD). Available at:

http://www.nhlbi.nih.gov/health/public/lung/other/

copd_fact.htm. Accessed January 6, 2004.

[84] Muller NL, Coxson H. Chronic obstructive pulmonary

disease 4: imaging the lungs in patients with chronic

obstructive pulmonary disease. Thorax 2002;57(11):

982–5.

[85] Thurlbeck WM, Muller NL. Emphysema: definition,

imaging, and quantification. AJR Am J Roentgenol

1994;163:1017–25.

[86] Nakano Y, Muro S, Sakai H, Hirai T, Chin K, Tsukino

M, et al. Computed tomographic measurements of

airway dimensions and emphysema in smokers. Cor-

relation with lung function. Am J Respir Crit Care

Med 2000;162(3):1102–8.

[87] Miller RR, Muller NI, Vedal S, Morrison NJ, Staples

CA. Limitations of computed tomography in the as-

sessment of emphysema. Am Rev Respir Dis 1989;

139:980–3.

[88] Kuwano K, Matsuba K, Ikeda T, Murakami J, Araki

A, Nishitani H, et al. The diagnosis of mild em-

physema: correlation of computed tomography and

pathology scores. Am Rev Respir Dis 1990;141:

169–78.

[89] Muller N, Staples C, Miller R, Abboud R. Density

mask. An objective method to quantitate emphysema

using computed tomography. Chest 1988;94(4):

782–7.

[90] Brown MS, McNitt-Gray MF, Goldin JG, Greaser

LE, Hayward UM, Sayre JW, et al. Automated mea-

surement of single and total lung volume from CT.

J Comp Ass Tomo 1999;23:632–40.

[91] Gierada DS, Yusen RD, Pilgram TK, Crouch L, Slone

RM, Bae KT, et al. Repeatability of quantitative CT

http:\\www.nhlbi.nih.gov\health\public\lung\other\copd_fact.htm


indexes of emphysema in patients evaluated for lung

volume reduction surgery. Radiology 2001;220(2):

448–54.

[92] Bankier AA, De Maertelaer V, Keyzer C, Gevenois

PA. Pulmonary emphysema: subjective visual grading

versus objective quantification with macroscopic

morphometry and thin-section CT densitometry.

Radiology 1999;211(3):851–8.

[93] Nakano Y, Muller NL, King GG, Niimi A, Kalloger

SE, Mishima M, et al. Quantitative assessment of

airway remodeling using high-resolution CT. Chest

2002;122(90060):271S–5S.

[94] King GG, Muller NL, Pare PD. Evaluation of airways

in obstructive pulmonary disease using high-resolu-

tion computed tomography. Am J Respir Crit Care

Med 1999;159(3):992–1004.

[95] Wood SA, Zerhouni EA, Hoford JD, Hoffman EA,

Mitzner W. Measurement of three-dimensional lung

tree structures by using computed tomography. J Appl

Physiol 1995;79(5):1687–97.

[96] McNamara AE, Muller NL, Okazawa M, Arntorp J,

Wiggs BR, Pare PD. Airway narrowing in excised

canine lungs measured by high-resolution computed

tomography. J Appl Physiol 1992;73(1):307–16.

[97] Reinhardt JM, D’Souza ND, Hoffman D. Accurate

measurement of intrathoracic airways. IEEE Trans

Med Imaging 1997;16(6):820–7.

[98] Dalen J. Pulmonary embolism: what have we learned

since Virchow?: natural history, pathophysiology, and

diagnosis. Chest 2002;122:1440–56.

[99] Sinner WN. Computed tomographic patterns of pul-

monary thromboembolism and infraction. J Comp

Ass Tomo 1978;2:395–9.

[100] Rathbun SW, Raskob GE, Whitsett TL. Sensitivity

and specificity of helical computed tomography in

the diagnosis of pulmonary embolism: a systematic

review. Ann Intern Med 2000;132:227–32.

[101] Mullins MD, Becker DM, Hagspiel KD, Philbrick JT.

The role of spiral volumetric computed tomography

in the diagnosis of pulmonary emboli. Arch Intern

Med 2000;160(3):293–8.

[102] Ghaye B, Szapiro D, Mastora I, Delannoy V, Duha-

mel A, Remy J, et al. Peripheral pulmonary arteries:

how far in the lung does multi-detector row spiral CT

allow analysis? Radiology 2001;219(3):629–36.

[103] Goodman LR, Curtin JJ, Mewissen MW, Foley WD,

Lipchik RJ, Crain MR, et al. Detection of pulmonary

embolism in patients with unresolved clinical and

scintigraphic diagnosis: helical CT vs. angiography.

AJR Am J Roentgenol 1995;164:1369–74.

[104] Schoepf UJ, Holzknecht N, Helmberger TK, Crispin

A, Hong C, Becker CR, et al. Subsegmental pulmo-

nary emboli: Improved detection with thin-collima-

tion multi-detector row spiral CT. Radiology 2002;

222(2):483–90.

[105] Bates SM, Ginsberg JS. Helical computed tomogra-

phy and the diagnosis of pulmonary embolism. Ann

Intern Med 2000;132:240–1.

[106] Remy-Jardin M, Remy J, Cauvain O, Petyt L, Wan-

nebroucq J, Beregi JP. Diagnosis of central pulmonary

embolism with helical CT: role of two-dimensional

multiplanar reformations. AJR Am J Roentgenol

1995;165(5):1131–8.

[107] Schoepf UJ, Das M, Schneider A, Anderson M,

Wood S, Costello P. Computer aided detection

(CAD) of segmental and subsegmental pulmonary

emboli on 1-mm multidetector-row CT (MDCT)

studies. Paper 579 in Session G02. Chicago: Radio-

logical Society of North America; 2002.

[108] Peters TM. Image-guided surgery: from x-rays to vir-

tual reality. Comput Methods Biomech Biomed Engin

2000;4(1):27–57.

[109] Hassfeld S, Muhling J. Computer assisted oral and

maxillofacial surgery—a review and an assessment

of technology. Int J Oral Maxillofac Surg 2001;1:

2–13.

[110] Sama AA, Khan SN, Girardi FP, Cammisa Jr CF.

Computerized frameless stereotactic image guidance

in spinal surgery. Orthop Clin N Am 2002;2:375–80.

[111] Gazelle GS, Goldberg SN, Solbiati L, Livraghi T.

Tumor ablation with radio-frequency energy. Radiol-

ogy 2000;217(3):633–46.

[112] Dupuy DE, Mayo-Smith WW, Abbott GF, DiPetrillo

T. Clinical applications of radio-frequency tumor ab-

lation in the thorax. Radiographics 2002;22(90001):

259S–69S.

[113] Zagoria RJ, Chen MY, Kavanagh PV, Torti FM. Ra-

dio frequency ablation of lung metastases from renal

cell carcinoma. J Urol 2001;166(5):1827–8.

[114] Goldberg SN, Gazell GS. Radio-frequency in the rab-

bit lung: efficacy and complications: in vivo study in

rabbit lung. Acad Radiol 1995;2:776–84.

[115] Suzuki S, Suzuki N, Hattori A, Takatsu A, Uchiyama

A. Deformable organ model using the sphere-filled

method for virtual surgery. In: Enderle JD, editor.

Proceedings of the 22nd Annual International Confer-

ence of the IEEE. Toronto: IEEE; 2000. p. 2416–8.

[116] Tseng D-C, Lin J-U. A hybrid physical deformation

modeling for laparoscopic surgery simulation. In:

Enderle JD, editor. Engineering in Medicine and Bi-

ology Society. Proceedings of the 22nd Annual Inter-

national Conference of the IEEE. Toronto: IEEE;

2000. p. 3032–4.

[117] Satava RM, Jones SB. Current and future applications

of virtual reality for medicine. Proc IEEE 1998;86(3):

484–9.

Index

Note: Page numbers of article titles are in boldface type.

A

Adenomas, lung cancer and, 9

Adrenal glands, metastasis to, from lung cancer, 9,

103–104

Airway disease, virtual bronchoscopy of, 79–86

benign disease, 81–82

technique for, 77–79

thoracic malignancies, 79–81

Angiography, computed tomography, of pulmonary

embolism, 114–117, 119

magnetic resonance, of pulmonary embolism,

117–118, 119

of pulmonary embolism, 113

B

Barrett’s esophagus, chromoendoscopy of, 88

diagnosis of, 85

high-resolution/high-magnification endoscopy of,

89–90

Bones, metastasis to, from lung cancer, 9–10,

104–105

Brain, metastasis to, from lung cancer, 9

Bronchial carcinoid, imaging of, 97

Bronchogenic carcinoma, imaging of, 18–19

Bronchogenic cysts, imaging of, 29, 31

Bronchoscopy, fluorescent. See Fluorescent

bronchoscopy.

virtual, of airway disease. See Airway disease.

C

Cartilaginous lesions, of chest wall, imaging of, 19

Central nervous system, metastasis to, from lung

cancer, 9

Chemoprevention, in fluorescent bronchoscopy, of

lung cancer, 71

Chest wall tumors, imaging of, 17–20

lung cancer with chest wall invasion, 17–18

metastatic lesions, 17

Pancoast tumors, 18–19

primary osseous and cartilaginous lesions, 19

soft tissue tumors, 19–20

Chondrosarcoma, imaging of, 19

Chromoendoscopy, of Barrett’s esophagus, 88

of esophageal cancer. See Esophageal cancer.

of intestinal metaplasia, 88–89

Chronic obstructive pulmonary disease, computed

tomography of, future directions in, 142

Chylothorax, mediastinal surgery and, 36

Computed tomography, future directions in, 135–149

dose reduction, 137–139

for chronic pulmonary obstructive disease, 142

for interventional procedures, 143

for lung cancer, 139–141

for pulmonary embolism, 142–143

image acquisition, 133–134

spatial resolution, 134–135

temporal resolution, 136–137

volume coverage, 135–136

in screening, for lung cancer. See Lung cancer.

in staging, of lung cancer, 1–2, 5–6, 9

with positron emission tomography, 7–8

of esophageal cancer, 60–64

of hemorrhage, after mediastinal surgery, 35

of lung cancer, with chest wall invasion, 17–18

of malignant mesothelioma, 13–16

of mediastinal lymph nodes, 35

of mediastinal masses, 24–25

of mediastinitis, 36

of Pancoast tumors, 18–19

of thymomas, 26–27

Computed tomography angiography, of pulmonary

embolism, 14–117, 119

Computed tomography bronchoscopy, of airway

disease. See Airway disease.

1547-4127/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved.

doi:10.1016/S1547-4127(04)00066-0


Computed tomography venography, of pulmonary

embolism, 113–114

Computer-aided diagnosis, of pulmonary system,

125–133

benefits of, 124

characterization in, 127, 129

detection in, 126–127

documentation and health evaluation in,

129–130

for lung cancer, 41, 46–48, 141

visualization in, 124–126

Cysts, mediastinal, imaging of, 29, 31–32

D

Digital radiography, in screening, for lung cancer,

42–45

Dual energy radiography, in screening, for lung

cancer, 45–46

Duplex ultrasonography, of pulmonary

embolism, 113

E

Embolism, pulmonary. See Pulmonary embolism.

Endoscopic ultrasonography, in staging, of lung

cancer, 8


Endoscopy, high-resolution/high-magnification, of

Barrett’s esophagus, 89

of intestinal metaplasia and dysplasia, 89–90

Energy subtraction radiography, in screening, for

lung cancer, 45–48

Esophageal cancer, imaging of, 61–69

chromoendoscopy in, 86–89

Lugol’s solution in, 86–87

methylene blue in, 87–89

squamous cell carcinoma, 86–87

computed tomography in, 60–64

distant metastases, 59–61

for response to therapy, 65

high-resolution/high-magnification endoscopy

in, 89–90

lymphatic metastases, 63–65

magnetic resonance imaging in, 60–61

nuclear medicine studies in, 61

positron emission tomography in, 61, 64

primary tumors, 61–63

ultrasonography in, 60, 62–65

F

Fibrous tumors, pleural, imaging of, 16–17

Fluorescent bronchoscopy, of lung cancer, 71–77

chemoprevention for, 71

clinical experience with, 73

clinical trials of, 71–72

for staging, 72–73

future directions in, 73–74

instrumentation for, 70

risk stratification for, 70–71

technique for, 70

Fluorodeoxyglucose, in positron emission

tomography, 94, 95, 99, 101–108

G

Ganglioneuroblastomas, mediastinal, imaging of, 29

Ganglioneuromas, mediastinal, imaging of, 29

Gastroenteric cysts, imaging of, 31

Germ cell tumors, mediastinal, imaging of,

28–29, 33

Goiter, imaging of, 27–28

H

Hematomas, retrosternal, mediastinal surgery and, 36

Hemorrhage, mediastinal surgery and, 35

Hodgkin’s disease, imaging of, 27

I

Infections, mediastinal surgery and, 35–36

Intestinal dysplasia, high-resolution/high-magnifica-

tion endoscopy of, 89–90

Intestinal metaplasia, chromoendoscopy of, 88–89

high-resolution/high-magnification endoscopy of,

89–90

Iodine, in nuclear medicine studies, 95

L

Lipomas, of chest wall, imaging of, 19

Liposarcoma, pleural, imaging of, 17

Liver, metastasis to, from lung cancer, 105–106

Lugol’s solution, in chromoendoscopy, of esophageal

cancer, 86–87

Index / Thorac Surg Clin 14 (2004) 151–155152

Lung cancer, fluorescent bronchoscopy of. See

Fluorescent bronchoscopy.

missed, causes of, 41–42

positron emission tomography of, 101–107

and management, 106–107

and prognosis, 107for adrenal masses, 103–104

for diagnosis, 101–102for mediastinal lymph node involvement,

102–103for metastasis to bone, 104–105

for metastasis to liver, 105–106for staging, 7, 10, 102–106

screening for, computed tomography in, 53–59

and curability, 55–56computer-aided diagnosis in, 41,

46–48, 141diagnostic distribution in, 54

false-positive diagnoses in, 54–55future directions in, 139–141

regimen for, 52–53plain films in, 43–52, 55

computer-aided diagnosis in, 46–48digital radiography in, 42–45

energy subtraction radiography in, 45–48missed cancers in, 41–42

temporal subtraction in, 48staging of, 1–13, 102–106

extrathoracic disease in, 8–10computed tomography in, 9

in adrenal glands, 9, 103–104in bones, 9–10, 104–105

in brain, 9in liver, 105–106

magnetic resonance imaging in, 9positron emission tomography in, 9, 10

fluorescent bronchoscopy in, 72–73for primary tumor, 1–4

computed tomography in, 1–2magnetic resonance imaging in, 2–3

thoracoscopy in, 3–4mediastinal lymph nodes in, 4–8, 102–103

computed tomography in, 5–6computed tomography/positron emission

tomography fusion in, 7–8

endoscopic ultrasonography in, 8magnetic resonance imaging in, 6–7

positron emission tomography in, 7

with chest wall invasion, imaging of, 17–18

Lung imaging fluorescence endoscope. See

Fluorescent bronchoscopy.

Lymph nodes, in staging, of lung cancer. See Lung

cancer, staging of.

mediastinal, sampling of, 33–35

metastasis to, from esophageal cancer, 63–65

Lymphography, magnetic resonance, in staging, of

lung cancer, 6–7

Lymphomas, imaging of, 27

M

Magnetic resonance angiography, of pulmonary

embolism, 117–118, 119

Magnetic resonance imaging, in staging, of lung

cancer, 2–3, 6–7, 9


of lipomas, of chest wall, 19–20

of lung cancer, with chest wall invasion, 18


of mediastinal masses, 25

of Pancoast tumors, 18–19

Magnetic resonance lymphography, in staging, of

lung cancer, 6–7

Malignant mesothelioma. See Pleural tumors.

Mediastinal lymph nodes, in staging, of lung cancer.

See Lung cancer, staging of.

sampling of, 33–35

Mediastinitis, imaging of, 35–36

Mediastinum, imaging of, 25–42


for cysts, 29, 31–32

for germ cell tumors, 28–29, 33

for invasive tumors, 32–33

for lymphomas, 27

for mediastinal lymph nodes, 33–35

for meningoceles, 32

for neurogenic tumors, 29

for peripheral nerve tumors, 29

for postoperative complications, 35–36

for sympathetic ganglia tumors, 29

for thymic masses, 26–27, 33

for thyroid masses, 27–28

magnetic resonance imaging in, 25

nuclear medicine studies in, 25–26

plain films in, 23–24

ultrasonography in, 25

Meningoceles, imaging of, 32

Mesothelioma. See Pleural tumors.

Methylene blue, in chromoendoscopy, of esophageal

cancer, 87–89

N

Nerve sheath tumors, invasive, imaging of, 33

Neural tumors, invasive, imaging of, 33

Index / Thorac Surg Clin 14 (2004) 151–155 153

Neuroblastomas, mediastinal, imaging of, 29

Neuroenteric cysts, imaging of, 31

Neurogenic tumors, mediastinal, imaging of, 29

Non-Hodgkin’s lymphoma, imaging of, 27

Nuclear medicine studies, of esophageal cancer, 61


of pulmonary embolism, 112–113, 119

of thoracic malignancies. See

Thoracic malignancies.

O

Osseous lesions, of chest wall, imaging of, 19

P

Pancoast tumors, imaging of, 18–19

Peptides, radiolabeled, in nuclear medicine studies,

95, 97, 99

Pericardial cysts, imaging of, 31–32

Peripheral nerve tumors, mediastinal, imaging of, 29

Plain films, in screening, for lung cancer. See

Lung cancer.

of hemorrhage, after mediastinal surgery, 35


of mediastinitis, 35–36

of pulmonary embolism, 111–112, 119

of solitary fibrous pleural tumors, 16–17

Plaques, pleural, imaging of, 16

Pleural tumors, imaging of, 13–17

liposarcoma, 17

malignant mesothelioma, 13–16


magnetic resonance imaging in, 13–14

positron emission tomography in, 15–16

pleural metastasis, 17

pleural plaques, 16

solitary fibrous tumor, 16–17

Positron emission tomography, instrumentation for,

94–95

of esophageal cancer, 61, 64

of lung cancer. See Lung cancer.


Pulmonary embolism, imaging of, 113–124

algorithm for, 118–120

angiography in, 113

computed tomography angiography in,

114–117, 119

computed tomography in, future directions in,

142–143

magnetic resonance angiography in,

117–118, 119

plain films in, 111–112, 119

venous imaging in, 113–114, 119

ventilation-perfusion scintigraphy in,

112–113, 119

R

Radiolabeled peptides, in nuclear medicine studies,

95, 97, 99

Receiver operating characteristic, in statistical

analysis, of lung cancer, 41–42

Retrosternal hematomas, mediastinal surgery

and, 36

S

Single-photon emission computed tomography,

instrumentation for, 94

Soft tissue tumors, of chest wall, imaging of, 19–20

Somatostatin, in nuclear medicine studies, 95, 97, 99

Squamous cell carcinoma, esophageal, 85

chromoendoscopy of, 86–87

Squamous cell dysplasia, esophageal, 86

Stenoses, airway, virtual bronchoscopy of, 81–82

Sympathetic ganglia tumors, mediastinal, imaging

of, 29

T

Technetium, in nuclear medicine studies, 95, 99

Temporal subtraction, in screening, for lung

cancer, 48

Thallium, in nuclear medicine studies, 95

Thoracic duct injury, mediastinal surgery and, 36

Thoracic malignancies, lung cancer. See Lung cancer.

nuclear medicine studies of, 95–112

fluorodeoxyglucose in, 99, 101, 107–108

instrumentation for, 94–95

radiolabeled peptides in, 95, 97, 99

radionuclides in, 95

virtual bronchoscopy of, 79–81

Thoracoscopy, in staging, of lung cancer, 3–4

Index / Thorac Surg Clin 14 (2004) 151–155154

Thymic masses, imaging of, 26–27

Thymomas, imaging of, 26–27, 33

Thyroid masses, imaging of, 27–28

U

Ultrasonography, duplex, of pulmonary

embolism, 113

endoscopic, in staging, of lung cancer, 8


of esophageal cancer, 60

of mediastinal masses, 25

V

V/Q scans, of pulmonary embolism, 112–113, 119

Venography, computed tomography, of pulmonary

embolism, 113–114

Venous imaging, of pulmonary embolism,

113–114, 119

Ventilation-perfusion scintigraphy, of pulmonary

embolism, 112–113, 119

Video-assisted thoracoscopic surgery, for mediastinal

lymph nodes, 34–35

Virtual bronchoscopy, of airway disease. See

Airway disease.

Index / Thorac Surg Clin 14 (2004) 151–155 155

Imaging modalities in general thoracic surgery

Health & Medicine

Transcript of Imaging modalities in general thoracic surgery