Review Article - CiteSeer

European Heart Journal (1997) 18, 1056-1067

Review Article

Reconstruction and quantification with three-dimensionalintracoronary ultrasound

An update on techniques, challenges, and future directions

C. von Birgelen, G. S. Mintz*, P. J. de Feyter, N. Bruining, A. Nicosia, C. Di Mario,P. W. Serruys and J. R. T. C. Roelandt

Thoraxcenter, University Hospital Rotterdam-Dijkzigt, Erasmus University Rotterdam, The Netherlands, and the* Washington Hospital Center, Washington DC, U.S.A.

Introduction

Conventional two-dimensional intracoronary ultra-sound enables the extent, distribution, and therapy ofatherosclerotic plaques to be studied as it provides aunique tomographic visualization of both the vascularlumen and wall1'"71. The most important limitation oftwo-dimensional intracoronary ultrasound in pre/post-studies is the matching of the target sites for serialmeasurement. This problem is caused by the lack of athird dimension in conventional intracoronary ultra-sound examinations. Three-dimensional reconstructionof intracoronary ultrasound images'8"101 permits a moreadvanced assessment of vessel, lumen and plaquemorphology.

Measurement of the plaque area in an entirecoronary segment may provide more detail of the com-plex plaque architecture and avoids the difficult mentalconceptualization process1""15'. Moreover, during on-line three-dimensional reconstruction (Fig. 1) measure-ments of the target lesion and the reference segmentsmay be accessed immediately in a reconstructed longi-tudinal view. This may facilitate selecting the best typeand sized interventional device or the evaluation ofcomplications, as the relevant coronary segments can be

Key Words: Intravascular ultrasound, three-dimensional recon-struction, image processing, coronary artery disease.

Dr von Birgelen is the recipient of a Fellowship of the GermanResearch Society (Deutsche Forschungsgemeinschaft, Bonn,Germany.

Correspondence: Professor Jos R. T. C. Roelandt, MD, PhD,Director of the Division Cardiology, University HospitalRotterdam-Dijkzigt, Thoraxcenter, P.O. Box 1738, 3000 DRRotterdam, The Netherlands.

carefully examined before and/or after interventionalprocedures19'16"21'.

Basic processing steps

The basic processing steps to obtain a three-dimensionalreconstruction from the two-dimensional intracoronaryultrasound images are similar for all systems currentlyavailable.

Image acquisition

The better the two-dimensional intracoronary ultra-sound images the better the three-dimensional recon-structions. Thus machine settings must be optimizedbefore image acquisition is performed. Starting distal tothe stenosis, the imaging catheter is withdrawn throughthe arterial segment to be reconstructed. The imagingcore of sheath-based intracoronary ultrasound catheters,which are designed for repeated pull-backs and mostfrequently used with three-dimensional intracoronaryultrasound, does not have direct contact with the vesselwall. Such catheters are equipped with a 15 cm longtransparent distal sheath which houses the transducer.There are two catheter designs: a 3-2F short monorailand a 2-9F common distal lumen catheter. The 2-9F designhas a common distal lumen that houses the guide wire(during catheter introduction) or the transducer (duringimaging when the guide wire has been pulled back) alter-nately. These intracoronary ultrasound catheter designsstabilize the transducer pull-back trajectory and reduce therisk of a non-uniform speed in continuous pull-backs, but

0195-668X/97/071056+12 S18.00/0 1997 The European Society of Cardiology

Dow

nloaded from https://academ

ic.oup.com/eurheartj/article/18/7/1056/485801 by guest on 15 January 2022

Review 1057

Figure 1 Short eccentric plaque in a proximal coronary segment. Thelongitudinal reconstruction (lower panel) and the on-line measurement ofthe luminal cross-sectional area and minimal diameter (right upperpanel) were obtained from the acoustic quantification system (Echo-Quant).

during the first 5 to 10 s of a continuous pull-back theremay be straightening of the imaging core inside thecatheter before a constant withdrawal speed is attained.

There are different pull-back approaches whichcan be applied. A continuous-speed pull-back resultingin an equidistant spacing of adjacent images'221, is stillthe most common approach. Side-branches or spots ofcalcium are used as topographic landmarks to ensure areliable comparison of the same arterial segment in serialstudies. A modified concept of the continuous-speedpull-back is ECG-triggered video labelling during uni-form pull-back of the intracoronary ultrasound trans-ducer. Video frames coinciding with the R-wave of theECG are automatically labelled and images acquired atthe same phase of the cardiac cycle are used for off-linethree-dimensional reconstruction. This approach dis-plays the arterial segment and enables the vasculardimensions to be measured at any time during thecardiac cycle. It also minimizes the systolic-diastolicartifacts which are frequently observed in non-triggereduniform-speed pull-backs (Fig. 2).

By using an ECG-triggered pull-back device incombination with an ECG-gated image acquisition'231,the problem of cyclic motion artifacts can be over-come'24-251. A dynamic three-dimensional reconstructionsystem, initially designed for three-dimensional recon-struction of echocardiographic images (Echoscan,TomTec, Munich, Germany)'231, can dynamically dis-play the arterial segment, showing the motion of anentire cardiac cycle. Before image acquisition starts, theupper and lower limits of the RR interval are defined.

A maximum number of 25 intracoronary ultrasoundimages per cardiac cycle can be sampled if the length ofthe RR interval meets the pre-set range (Fig. 3).

Image digitization and segmentation

Digitization of the intracoronary ultrasound images canbe performed on- or off-line by sampling the videoframes with a framegrabber at a defined digitizationframe rate. The segmentation step identifying structuresof interest in the intracoronary ultrasound images canbe achieved by applying dedicated algorithms, whichdiscriminate between the blood-pool inside the lumenand structures of the vessel wall'261. The quality of thefinal three-dimensional reconstruction and the accuracyof the quantitative analysis are highly influenced by thequality of the segmentation algorithm. In addition to thethreshold approaches (Fig. 4)'101, segmentation can beachieved by using either an acoustic quantification algor-ithm'21-271 or a contour detection algorithm'1314'28'291.

The acoustic quantification method (Fig. 5) dis-tinguishes between the blood pool and vessel wall usingan algorithm for statistical pattern recognition (Echo-Quant, INDEC, CA, U.S.A.). Comparing the ultra-sound speckle pattern of flowing blood to the pattern ofthe vessel wall, there is much more variation in time inblood'301. The algorithm is able to distinguish betweenthese two patterns and to detect the interface betweenblood and vessel wall. Finally, pixels (picture elements)identified as part of the blood pool are removed.

Eur Heart J, Vol. 18, July 1997

Dow



1058 C. von Birgelen et al.

Figure 2 Cyclic artifacts in a venous bypass graft. The longitudinallyreconstructed image of a stented bypass graft shows enormous saw-shaped artifacts, resulting from cyclic vessel pulsation and the move-ment of the intracoronary ultrasound catheter inside ('catheterfluttering'). The artifacts are visible in both the longitudinal display andthe graph, showing the measurements of the luminal cross-sectional areaand the minimum diameter (right upper panel). (Reproduced withpermission1541.)

Another approach is a contour detection system(Fig. 6), which has been developed at the ThoraxcenterRotterdam. A minimum cost algorithm detects theintimal leading edge and the external vascular bound-ary (external elastic lamina)113'14'28'291. Another semi-automated contour detection system, described bySonka et al. detects the luminal contour and the con-tours of the internal and external elastic laminae. Initialin vitro studies show a good correlation with lumenand plaque area measurements obtained by manualtracing!31-32].

Image reconstruction and display

Different display formats can be used to present thethree-dimensional image data sets. The most commonlygenerated is the longitudinal format (Fig. 1). A cylindricalformat (Fig. 7). and a lumen cast format are also some-times used. General programs for three-dimensional pres-entation can display oblique and tangential cuts throughthe reconstructed structures, comparable to the displayoptions available in magnetic resonance imaging systems.A dynamic visualization of the artery after ECG-gatedimage acquisition is also possible'24'.

Current three-dimensional reconstructionsystems

Different systems are available. Each has a distincttechnical approach with specific advantages and dis-

advantages in applicability, imaging, and quantification.The following systems are used at the WashingtonHospital Center and/or the Thoraxcenter Rotterdam.

Acoustic quantification system

This system (EchoQuant), recently validated in rabbitaortas'271, can be used either on- or off-line, and itsamples intracoronary ultrasound images with a digi-tization frame rate of 8-5 frames . s ~' . The length of thereconstructed coronary segment is determined by thepull-back speed since the image acquisition and digitiz-ation rates are fixed. Using a pull-back speed of1 0 mm . s ~ ' images of 8 cm long segments can beacquired. Segmentation and reconstruction of a vascularsegment 3 cm long can be performed within 3 min. Thequality of the automated detection can be checked andmanually corrected in individual cross-sectional images.No geometric assumptions of the lumen shape arerequired; and the program may therefore provide ac-curate segmentation of an irregularly shaped lumen.However, application of the algorithm may be hamperedby the quality of the basic intracoronary ultrasoundimages'27'331. Some parameters determining the auto-mated identification process can be adjusted by the user;this is particularly important when the intracoronaryultrasound image quality is not optimal. Since thereconstruction is performed within a few minutes, it canbe used in the catheterization laboratory'2'1.


Dow



Review 1059

ECG

Memory

Accepted Step Rejected Accepted Step

RR interval = 1OOO ms ± 100 msRespiration gating = OFF

Figure 3 The combined use of an ECG-gated pull-back device andimage acquisition by the dynamic three-dimensional reconstructionsystem (Echoscan, TomTec, Munich, Germany) allows the problemof cyclic artifacts to be overcome and a dynamic visualizationobtained. The range of the RR interval is defined (here:1000 ± 100 ms) before the image acquisition starts. During thepull-back procedure, the maximum number of 25 intracoronaryultrasound images per cardiac cycle for each scanning site isdigitized and sampled in the computer memory, unless the length ofthe RR interval fails to meet the pre-set range (here: third cardiaccycle). Each time a cycle is stored, the following heart beat isrequired to perform a pull-back step in order to reach the adjacentscanning site. (Reproduced with permission'24'.)

A selected cross-sectional image, a longitudinallyreconstructed image (Fig. 1), and a cylindrical three-dimensional view (presenting the segment opened longi-tudinally) (Fig. 5) are displayed on the monitor. Themeasurements of the automated cross-sectional luminalarea are displayed in a diagram. Although the algorithmis unable to detect the external contour of the totalvessel, the current version of this program provides anoption which allows manual tracing of the external con-tour of the vessel in selected two-dimensional images.

Thoraxcenter contour detection system

This analysis system digitizes a user-defined region ofinterest with a maximum of 200 tomographic images.Segments of approximately 2-5 cm (uniform pull-back)or 4 cm long (ECG-gated pull-back) can reliably beanalysed. The method depends less on the image qualitysince it operates interactively. Reliable segmentation andthree-dimensional reconstruction remain possible evenwhen the image quality is not optimal. However, userinteraction is required in the presence of irregular lumenshapes. On-line application of this system has recentlybeen started using ECG-gated image acquisition.

The contour detection procedure (Fig. 6) consistsof three steps'1314-281. First, the intracoronary ultrasound

images are modelled in a voxel space'341; and two per-pendicular cut planes running parallel to the longitudi-nal axis of the vessel are selected. Data located at theinterception of these cut planes and the voxel volumeare derived to reconstruct two longitudinal images ofthe vascular segment. The angle and location of thecut planes is defined by the user in order to optimizethe representation of the arterial segment on thelongitudinal sections.

In a second step, the contours of the luminal andexternal vascular boundaries (external elastic lamina) aredetected in the longitudinal images. This step is based onthe application of a minimum cost algorithm1351, whichhas previously been validated'361 and applied in two-dimensional intracoronary ultrasound images'37'. Theuser is free to add markers, forcing the contours to passthrough these sites. The optimal path of the longitudinalcontours is updated serially, using dynamic program-ming techniques. The contours of the longitudinalimages are then depicted as points in each cross-sectionalimage.

Finally, automated contour detection is per-formed in all the cross-sectional images, using the fouredge points, derived from the longitudinal contours, aslandmarks to guide the detection. The accuracy of thefinal contours can be checked and corrections may beperformed. The system permits the quantitative analysis


Dow




Figure 4 In vitro three-dimensional intracoronary ultrasound reconstruction (left panel)of a Palmaz-Schatz stent (Johnson & Johnson, Warren, U.S.A.) (right panel) based onsegmentation by thresholding. The articulation of the Palmaz-Schatz stent and thetypical strut pattern can be easily distinguished. (Reproduced with permission'20'.)

Figure 5 Complex coronary lesion in a mid right coronary arterybefore intervention. A large superficial calcification (arrowheads) isvisible in the transverse (left upper panel), the longitudinally recon-structed (right upper panel), and the cylindrical view (lower panel).Segmentation is performed by acoustic quantification (EchoQuant,Indec, Capitola, CA, U.S.A.) which detects and consecutively removesthe blood-pool (B) from the intracoronary ultrasound images. The lengthof the plaque calcification can easily be evaluated in the longitudinalreconstruction. The cursors (lines) in the transverse and longitudinalview permit the rotation of the longitudinal reconstruction and theselection of specific cross-sectional intracoronary ultrasound images.


Dow



Review 1061

it? tl

Figure 6 The principle of the tluee-tlunensional contourdetection system. The intracoronary ultrasound images,obtained during a motorized pull-back, are stored in thecomputer memory as a 'volumetric space'. The method isbased on the concept that edge points, derived fromlongitudinal contours which were previously detected ontwo longitudinally reconstructed images, guide and facili-tate the final contour detection on the transverse intracoro-nary ultrasound images. The position of an individualtransverse plane in the longitudinal sections is indicated bya horizontal cursor line which can be used to scroll throughthe whole series of transverse intracoronary ultrasoundimages. (Reproduced with permission'14'.)

of lumen and plaque, and even volumetric data (Fig. 8)can be obtained, as each cross-sectional image representsa slice of the reconstructed arterial segment'13"151. Area

and mean diameter measurements of the total vessel,lumen, and plaque are displayed in diagrams (Fig. 9).These diagrams also show diameter-stenosis, area-obstruction, and lumen symmetry functions1'314281.

Dynamic reconstruction system

The three-dimensional reconstruction tool installed ineach Echoscan system (TomTec) uses a segmentationwhich is based on the definition of thresholds in thescale of gray levels. The applicability of this algorithmdepends upon image quality. However, in instances withoptimal two-dimensional image quality, remarkabledynamic reconstructions can be obtained (Fig. 10). Asthe ECG-gated three-dimensional reconstruction of acoronary segment requires sampling and processing of alarge amount of data, the time of analysis is still slightlylonger than for a conventional analysis.

Using volume rendering techniques, the dynamicreconstruction system allows dynamic visualization ofthe reconstructed segment'241 with a maximum of 25frames per cardiac cycle. The reconstruction of varioustransverse and longitudinal sections is possible. Thelongitudinal reconstruction of a coronary artery seg-ment is readily available in the cardiac catheterizationlaboratory and similar to computer tomography ormagnetic resonance imaging, these longitudinal sectionscan pass through the centre of the vessel or cut the vesselwall tangentially.

Meanwhile, a version of the contour-detection-based analysis software of the Thoraxcenter,Rotterdam, has been customized for use with the Echo-scan system'25'291. The software package is available forusers of Echoscan systems.

Figure 7 Spatial view of a coronary segment (follow-up after previousdirectional coronary atherectomy), obtained from image data providedby the contour detection approach. This cylindrical display format is notrequired for the purpose of quantification, nevertheless additional insightinto plaque disruption may sometimes be obtained. (Reproduced withpermission114'.)


Dow




11

aCD

3

II

.2

II

500

400

300

200

100

0

500

400

300

200

100

0

500

y = O.96x + 7.60r = 0.99SEE = 4.84 mmJ

n = 20

_L _L _L100 200 300 400 500

g 20^ 15 -

c -5.E

c _5

g-10M -15

I S -20

y = l.OOx - 1.93r = 0.99

• SEE = 2.54 mm3

n = 20

g

s

olui

>m3o"aE

400

300

200

100

y =r =

= l.OOx -= 0.99

- SEE = 4.99n =

-

—

—

= 20

•

2.29

mm

0"

I I I100 200 300 400 500

Mean difference =- SD = ±2.66%

_ •

— •-*---•#-«

• •

1 1

-0.65%

• • • ,

1

2SD

-2SD

1

0 100 200 300 400

Mean lumen volume (mm )

500

II sS100 200 300 400 500 p

20

15

10

5

J l

-5

-10

-15

-20

Mean difference = 0.19%SD = ±0.67%

"2SD

-2SD

0 100 200 300 400 500

Mean total vessel volume (mm )

20

I

0

?olu

03a1

J3"3.a

nee

ffer

e

a

15

10

-5

0

- 5

-10

-15

-20

Mean difference = 0.95%- SD = ±2.81%

•

• « I— •

i i i

2 S D

-2SD

1

100 200 300 400 500Mean plaque volume (mm )

Figure 8 Inter-observer variability of volume measurements in-vivo by the contour detectionsystem. The lumen, total vessel, and plaque volume measurements in 20 coronary segments by twoindependent observers (I and II) showed a high reproducibility. In the right-hand panels relativeinter-observer diiferences are plotted against the average of the two measurements. Solid linesindicate the mean difference and the range of 2 standard deviations (SD), and dotted lines mark theline of identity in all panels. (Reproduced with permission"41.)

Challenges and future directions

Several factors, including problems related specificallyto intracoronary ultrasound139'401 as well as generallimitations of the three-dimensional reconstruction'411,influence the quality of the reconstruction. Bothlumen and plaque volume measurements showedminimal short-term biological variability upon repeatedpull-back of the same coronary artery segment'421.The quality of the basic intracoronary ultrasoundimage is crucial, as poor or incomplete visualizationof the lumen-plaque and plaque-adventitia boundariesin the presence of calcification is a problem which

hampers both reconstruction and quantification. Cur-rently available intracoronary ultrasound transducershave a limited lateral resolution'431 and image dis-tortion by non-uniform rotation or non-coaxial pos-itioning of the intracoronary ultrasound catheter inthe lumen may create complex artifacts in three-dimensional reconstructions'411. Moreover, motorizedpull-back devices or displacement sensors cannotalways assure an equal distance between adjacentimages, as bends of the ultrasound catheter may inducea difference between the movement of the distaltransducer and the proximal part of the intracoronaryultrasound catheter.


Dow



Review 1063

30.0

25.0

20.0

15.0

10.0

5.0

0.0

-i

-

Area measurement (All frames)

ft

i

2

Area (mm )

1 1

-&

Frame

Obstruction, G:Area, R:Diameter (All frames)

50 100 150 200 100 150 200

Diameter measurement (All frames)1.0

0.8

0.6

0.4

0.2

0.0

Wax/Min

" k

IT

G:BEM, R:Lumen,

i i

B:Plaque (All frames)

I 1

I I. Frame

0 50 100 150 200 0 50 100 150 200

Figure 9 Standard display of the results, by the contour detection method, of theThoraxcenter, Rotterdam. The clinical example shows the results of an intracoronaryultrasound analysis performed 6 months after directional coronary atherectomy in aproximal left anterior descending coronary artery. The left mid and lower panels showarea and mean diameter measurements of lumen, total vessel, and plaque. The grayareas represent the coronary plaque and the upper and lower boundaries of the grayzones correspond with the dimensions of the coronary lumen and the total vessel. Theabsolute plaque measurements are shown as a single line function for both area anddiameter measurements (left mid and lower panels). Functions of the diameter-stenosis(%) and area-obstruction (%) are displayed in the mid right panel. The right lower panelshows the symmetry of the lumen and total vessel and the eccentricity of the plaque.(Reproduced with permission'14'.)

The cyclic movement of the intracoronary ul-trasound catheter and systolic-diastolic changes in ves-sel dimensions can originate typical saw tooth-shapedimage artifacts (Fig. 2)'41'441. ECG-gated image acqui-sition and pull-back have the potential to minimizethese cyclic artifacts and to optimize image acquisition,allowing reliable volumetric measurement125'29'45'. How-ever, compared to continuous pull-back, image acqui-sition by the ECG-gated approach requires a longer

acquisition time which may cause problems in patientswith very severe coronary stenoses.

Vessel curvatures with a radius of less than5 cm may cause a significant distortion of the three-dimensional reconstructed image1461. Over-estimationand under-estimation of certain portions of the plaquemay be caused by vessel curvatures'47' as a result ofthe curved pull-back trajectory of the intracoronaryultrasound transducer. The contour detection system


Dow




Figure IV "EC'G-gated three-dimensional reconstruction of a proximalcoronary artery with an eccentric plaque formation (on the right-handside). A custom-designed pull-back device with a stepper motor, devel-oped at the Thoraxcenter, Rotterdam, and the Echoscan (TomTecGmbH, Munich, Germany) were used to obtain this reconstruction.The plaque is visible on both a longitudinal section through the artery(right panel) and in a three-dimensionally reconstructed view (leftpanel).

demonstrates artificial curvatures caused by a localizedeccentric plaque burden. Other three-dimensional recon-struction systems, as for instance the acoustic quantifi-cation system, straighten the display of the coronarysegment artificially. The combined use of data obtained

Figure 11 Combined use of biplane angiography andthree-dimensional intracoronary ultrasound by ANGUS.This novel method allows the true geometry of the coronarylumen and plaque to be investigated, taking arterial curva-tures and catheter bends into account. A reconstruction ofan atherosclerotic right coronary artery is displayed in afrontal projection. Intracoronary ultrasound data providedby the contour detection method were spatially arrangedand interpolated, using biplane data on both the pull-backtrajectory and the angiogram. (Reproduced with permis-sion'191.)

from biplane angiography and intracoronary ultrasound(Fig. 11) may help overcome many of these limitationsto the provision of information on the real vessel curva-tures and the orientation of the intracoronary ultra-sound catheter148'491. Using ANGUS'491 —a technicalapproach which has been developed at the ThoraxcenterRotterdam — in a geometric vessel phantom of knowndimensions, a high accuracy was observed; and firstapplications in humans yielded good results. In themeantime, our findings have been confirmed by anothergroup using a similar approach1501. The use of newforward looking transducers'511 may also help to over-come some of the current limitations but the value ofthis device is still limited by the low image resolution andthe large dimensions of the ultrasound transducers.

Miniaturization of the imaging catheters andimprovements in the computer technology will also helpto increase future applications of three-dimensionalintracoronary ultrasound, which has the potential tolargely replace quantitative coronary angiography inthe future (Fig. 12) and permits volumetric quanti-fication'13"15'291 without need for laborious manualtracing'52'531.

ConclusionUntil recently, three-dimensional intracoronary ultra-sound appeared to be restricted to pure research appli-cations, but we feel that the method will gain furtherimportance and become a routine technique, if theinterest, effort and technical developments in the fieldare sustained.


Dow



Review 1065

Figure 12 Primary lesion in a proximal left anterior descendingcoronary artery (LAD), assessed by the three-dimensional intracoronaryultrasound contour detection system of the Thoraxcenter, Rotterdam.An arrowhead indicates the target stenosis in the cylindrical reconstruc-tion. The measurements are shown in the lower panel: The white zonerepresents the coronary plaque and the upper and lower boundaries ofthis zone correspond with the values of the coronary lumen and the totalvessel areas (mm2). The absolute values of the plaque area (mm2) arealso shown as a single line. The area stenosis at the site of the midsegment of the LAD was approximately 20%. According to the mech-anism described by Glagov, an enlargement of the total vessel area,partly compensating the plaque burden, may be expected at the site ofrelatively focal plaque formation. In this case, however, a paradoxicalreduction of the total vessel area, from the distal reference in the midLAD (MID) to the target stenosis in the proximal segment (PROX) wasobserved. (Reproduced with permission1 '.)

References

[1] Hodgson J McB, Graham SP, Sarakus AD et al. Clinicalpercutaneous imaging of coronary anatomy using an over-the-wire ultrasound catheter system. Int J Cardiac Imaging 1989;4: 186-93.

[2] Yock PG, Linker DT. Intravascular ultrasound. Lookingbelow the surface of vascular disease. Circulation 1990; 81:1715-8.

[3] Nissen SE, Gurley JC, Grines CL et al. Intravascular ultra-sound assessment of lumen size and wall morphology innormal subjects and patients with coronary artery disease.Circulation 1991; 84: 1087-99.

[4] Fitzgerald PJ, St. Goar FG, Connolly AJ, et al. Intravascularultrasound imaging of coronary arteries: is three layers thenorm? Circulation 1992; 86: 154-8.

[5] Ge J, Erbel R, Rupprecht HJ el al. Comparison of intra-vascular ultrasound and angiography in the assessment ofmyocardial bridging. Circulation 1994; 89: 1725-32.

[6] Mintz GS, Popma JJ, Pichard AD et al. Arterial remodelingafter coronary angioplasty: a serial intravascular ultrasoundstudy. Circulation 1996: 94; 35-43.

[7] Erbel R, Ge J, Bockisch A et al. Value of intracoronaryultrasound and Doppler in the differentiation of angiographi-cally normal coronary arteries: a prospective study in patientswith angina pectoris. Eur Heart J 1996; 17: 880-9.

[8] Matar FA, Mintz GS, Douek P et al. Coronary artery lumenvolume measurement using three-dimensional intravascular

ultrasound: validation of a new technique. Cathet CardiovascDiagn 1994; 33: 214-20.

[9] Rosenfield K, Kaufman J, Pieczek A, Langevin RE, Razvi S,Ilsner JM. Real-time three-dimensional reconstruction ofintravascular ultrasound images of iliac arteries. Am J Cardiol1992; 70: 412-5.

[10] Rosenfield K, Losordo DW, Ramaswamy K., Isner JM.Three-dimensional reconstruction of human coronary andperipheral arteries from images recorded during two-dimensional intravascular ultrasound examination. Circu-lation 1991; 84: 1938-56.

[11] Losordo DW, Rosenfield K, Peiczek A, Baker K, Harding M,Isner JM. How does angioplasty work? Serial analysis ofhuman iliac arteries using intravascular ultrasound. Circu-lation 1992; 86: 1845-58.

[12] von Birgelen C, Di Mario C, Serruys PW. Structural andfunctional characterization of an intermediate stenosiswith intracoronary ultrasound: a case of "reverse Glagovianmodeling". Am Heart J 1996; 132: 694-6.

[13] von Birgelen C, van der Lugt A, Nicosia A et al. Computer-ized assessment of coronary lumen and atherosclerotic plaquedimensions in three-dimensional intravascular ultrasoundcorrelated with histomorphometry. Am J Cardiol 1996; 78:1202-9.

[14] von Birgelen C, Di Mario C, Li W et al. Morphometricanalysis in three-dimensional intracoronary ultrasound: anin-vitro and in-vivo study using a novel system for the contourdetection of lumen and plaque. Am Heart J 1996; 132: 516-27.


Dow




[15] von Birgelen C, Slager CJ, Di Mario C, de Feyter PJ, SerruysPW. Volumetric intracoronary ultrasound: a new maximumconfidence approach for the quantitative assessment ofprogression/regression of atherosclerosis? Atherosclerosis1995; 118 (Suppl): S1O3-13.

[16] Coy KM, Park JC, Fishbein MC et al. In vitro validation ofthree-dimensional intravascular ultrasound for the evaluationof arterial injury after balloon angioplasty. J Am Coll Cardiol1992; 20: 692-700.

[17] Rosenfield K, Kaufman J, Pieczek AM et al. Human coronaryand peripheral arteries: on-line three-dimensional reconstruc-tion from two-dimensional intravascular US scans. Radiology1992; 184: 823-32.

[18] Mintz GS, Leon MB, Satler LF et al. Clinical experience usinga new three-dimensional intravascular ultrasound systembefore and after transcatheter coronary therapies. J Am CollCardiol 1992; 19: 292A.

[19] Schryver TE, Popma JJ, Kent KM, Leon MB, Eldredge S,Mintz GS. Use of intracoronary ultrasound to identify thetrue coronary lumen in chronic coronary dissection treatedwith intracoronary stenting. Am J Cardiol 1992; 69: 1107-8.

[20] Mintz GS, Pichard AD, Satler LF, Popma JJ, Kent KM,Leon MB. Three-dimensional intravascular ultrasonography:reconstruction of endovascular stents in vitro and in vivo.J Clin Ultrasound 1993; 21: 609-15.

[21] von Birgelen C, Gil R, Ruygrok P et al. Optimized expressionof the Wallstent compared to the Palmaz-Schatz stent: onlineobservations with two and three-dimensional intracoronaryultrasound after angiographic guidance. Am Heart J 1996;131: 1067-75.

[22] Fuessel RT, Mintz GS, Pichard AD et al. In vivo validationof intravascular ultrasound length measurements using amotorized transducer pullback system. Am J Cardiol 1996; 77:1115-8.

[23] Roelandt JRTC, ten Cate FJ, Vletter WB, Taams MA.Ultrasonic dynamic three-dimensional visualization of theheart with a multiplane transesophageal imaging transducer.J Am Soc Echocardiogr 1994; 7: 217-29.

[24] Bruining N, von Birgelen C, Di Mario C et al. Dynamicthree-dimensional reconstruction of ICUS images based on anECG gated pull-back device. In: Computers in Cardiology1995, Los Alamitos: IEEE Computer Society Press, 1995:633-6.

[25] von Birgelen C, Mintz GS, Nicosia A et al. ECG-gatedintravascular ultrasound image acquisition after coronarystent deployment facilitates on-line three-dimensional recon-struction and automated lumen quantification. J Am CollCardiol 1997 (in press).

[26] Chandrasekaran K, D'Adamo AJ, Sehgal CM. Three-dimensional reconstruction of intravascular ultrasound im-ages. In: Yock PG, Tobis JM, eds. Intravascular UltrasoundImaging. New York: Churchill-Livingstone, 1992: 141-7.

[27] Hausmann D, Friedrich G, Sudhir K et al. 3D intravascularultrasound imaging with automated border detection using2.9F Catheters. J Am Coll Cardiol 1994; 23: 174A.

[28] Li W, von Birgelen C, Di Mario C et al. Semi- automaticcontour detection for volumetric quantification of intra-coronary ultrasound. In: Computers in Cardiology 1994,Los Alamitos: IEEE Computer Society Press, 1994: 277-80.

[29] von Birgelen C, de Vrey E, Mintz GS et al. ECG-gatedthree-dimensional intravascular ultrasound: feasibility andreproducibility of the automated analysis of coronary lumenand atherosclerotic plaque dimensions in humans. Circulation(in press).

[30] Li W, Gussenhoven EJ, Zhong Y et al. Temporal averagingfor quantification of lumen dimensions in intravascularultrasound images. Ultrasound Med Biol 1994; 20: 117-22.

[31] Sonka M, Zhang X, Siebes M, DeJong S, McKay CR, CollinsSM. Automated segmentation of coronary wall and plaquefrom intravascular ultrasound image sequences. In: Com-puters in Cardiology 1994, Los Alamitos: IEEE ComputerSociety Press, 1994: 281—4.

[32] Sonka M, Zhang X, Siebes M et al. Semi-automated detectionof coronary arterial wall and plaque borders in intravascularultrasound images. Circulation 1994; 90: 1-550.

[33] von Birgelen C, Kutryk MJB, Gil R et al. Quantification ofthe minimal luminal cross-sectional area after coronary stent-ing: two-dimensional and three-dimensional intravascularultrasound versus edge detection and videodensitometry. AmJ Cardiol 1996; 78: 520-5.

[34] Kitney R, Moura L, Straughan K. 3-D visualization ofarterial structures using ultrasound and voxel modelling. Int JCardiac Imag 1989; 4: 134-43.

[35] Li W, Bosch JG, Zhong Y et al. Image segmentation and 3Dreconstruction of intravascular ultrasound images. In: Wei Y,Gu B, eds. Acoustical Imaging, Vol. 20. New York: PlenumPress, 1993: 489-96.

[36] Li W, Gussenhoven EJ, Zhong Y et al. Validation of quanti-tative analysis of intravascular ultrasound images. Int JCardiac Imag 1991; 6: 247-54.

[37] Di Mario C, The SHK, Madretsma S et al. Detection andcharacterization of vascular lesions by intravascular ultra-sound: an in-vitro study correlated with histology. J Am SocEchocardiogr 1992; 5: 135-46.

[38] Masotti L, Pini R. Three-dimensional imaging. In: Wells,PNT, ed. Advances in Ultrasound Techniques and Instrumen-tation. New York: Churchill Livingstone, 1993: 69-77.

[39] Ge J, Erbel R, Seidel I et al. Experimented Uberpriifung derGenauigkeit und Sicherheit des intraluminalen Ultraschalls.Z Kardiol 1991; 80: 595-601.

[40] Di Mario C, Madretsma S, Linker D et al. The angle ofincidence of the ultrasonic beam: a critical factor for the imagequality in intravascular ultrasonography. Am Heart J 1993;125: 442-8.

[41] Roelandt JRTC, Di Mario C, Pandian NG et al. Three-dimensional reconstruction of intracoronary ultrasoundimages. Rationale, approaches, problems and directions.Circulation 1994; 90: 1044-55.

[42] Dhawale P, Rasheed Q, Berry J, Hodgson J McB. Quantifi-cation of lumen and plaque volume with ultrasound: accuracyand short term variability in patients. Circulation 1994; 90:1-164.

[43] Benkeser PJ, Churchwell AL, Lee C, Abouelnasr DM. Resol-ution limitations in intravascular ultrasound imaging J AmSoc Echocardiogr 1993; 6: 158-65.

[44] Di Mario C, von Birgelen C, Prati F et al. Three-dimensionalreconstruction of two-dimensional intracoronary ultrasound:clinical or research tool? Br Heart J 1995; 73 (Suppl 2):26-32.

[45] Dhawale PJ, Wilson DL, Hodgson J McB. Optimal dataacquisition for volumetric intracoronary ultrasound. CathetCardiovasc Diagn 1994; 32: 288-99.

[46] Waligora MJ, Vonesh MJ, Wiet SP, McPherson DD. Effect ofvascular curvature on three-dimensional reconstructionof intravascular ultrasound images. Circulation 1994; 90:1-227.

[47] Klein HM, Gunther RW, Verlande M et al. • 3D-surfacereconstruction of intravascular ultrasound images using per-sonal computer hardware and a motorized catheter control.Cardiovasc Intervent Radiol 1992; 15: 97-101.

[48] Koch L, Kearney P, Erbel R et al. Three-dimensional recon-struction of intracoronary ultrasound images: roadmappingwith simultaneously digitised coronary angiograms. In: Com-puters in Cardiology 1993, Los Alamitos: IEEE ComputerSociety Press, 1993: 89-91.

[49] Slager CJ, Laban M, von Birgelen C et al. ANGUS: A newapproach to three-dimensional reconstruction of geometryand orientation of coronary lumen and plaque by combineduse of coronary angiography and ICUS. J Am Coll Cardiol1995; 25: 144A.

[50] Evans JL, Ng KH, Wiet SG et al. Accurate three-dimensionalreconstruction of intravascular ultrasound data: spatially cor-rect three-dimensional reconstructions. Circulation 1996; 93:567-76.


Dow



Review 1067

[51] Ng K-H, Evans JL, Vonesh MJ et al. Arterial imaging with [54] von Birgelen C, Di Mario C, Reimers B et al. Three-a new forward-viewing mtravascular ultrasound catheter, II: dimensional intracoronary ultrasound imaging: methodologythree-dimensional reconstruction and display of data. and clinical relevance for the assessment of coronary arteriesCirculation 1994; 89: 718-23. and bypass grafts. J Cardiovasc Surg 1996; 37: 129-39.

[52] Matar FA, Mintz GS, Farb A et al. The contribution of tissueremoval to lumen improvement after directional coronaryatherectomy. Am J Cardiol 1994; 74: 647-50.

[53] Dussaillant GR, Mintz GS, Pichard AD et al. Small stent sizeand intimal hyperplasia contribute to restenosis: a volumetricintravascular ultrasound analysis. J Am Coll Cardiol 1995; 26:720-4.


Dow



Review Article - CiteSeer

Documents

Transcript of Review Article - CiteSeer