RADIOTHERAPY PORTAL IMAGING QUALITY · films, or port films). They can be used in an interactive...

AAPM REPORT NO. 24

RADIOTHERAPYPORTAL

IMAGING QUALITY

Published for theAmerican Association of Physicists in Medicine

by the American Institute of Physics

AAPM REPORT NO. 24(with figures)

RADIOTHERAPYPORTAL

IMAGING QUALITY

REPORT OF AAPMTASK GROUP No. 28

Members

Lawrence E. Reinstein, SUNY at Stony Brook, ChairmanHoward I. Amols, Columbia UniversityPeter J. Biggs, Massachusetts General HospitalRonald T. Droege, Bethesda Oak HospitalAlexander B. Filimonov, Mary Hitchcock HospitalWendell R. Lutz, University of ArizonaShlomo Shalev, Manitoba Cancer Foundation

December 1987

Published for theAmerican Association of Physicists in Medicine

by the American Institute of Physics

DISCLAIMER: This publication is based on sources andinformation believed to be reliable, but the AAPM and theeditors disclaim any warranty or liability based on or relat-ing to the contents of this publication.The AAPM does not endorse any products, manufac-turers, or suppliers. Nothing in this publication should beinterpreted as implying such endorsement.

Further copies of this report may be obtained from

Executive OfficerAmerican Association of Physicists in Medicine

335 E. 45 StreetNew York, NY 10017

Library of Congress Catalog Card Number: 88-70048International Standard Book Number: 0-88318-557-1

International Standard Serial Number: 0271-7344

Copyright © 1988 by the American Associationof Physicists in Medicine

All rights reserved. No part of this publication may bereproduced, stored in a retrieval system, or transmittedin any form or by any means (electronic, mechanical,photocopying, recording, or otherwise) without the

prior written permission of the publisher.

Published by the American Institute of Physics, Inc.,335 East 45 Street, New York, New York 10017

Printed in the United States of America

Radiotherapy Portal Imaging Quality

I. INTRODUCTION

The major goal of radiation therapy is thedelivery of a prescribed radiation dose as accuratelyas possible to a tumor region while minimizing thedose distribution to the neighboring normal tissues.There are several geometric factors which tend tocompromise this goal such as patient movement,improper placement of shielding blocks, shifting ofskin marks relative to internal anatomy and incorrectbeam alignment. At present, the only method commonlyavailable for measuring and documenting the extent ofgeometric treatment accuracy is the radiotherapyportal film. These films are used by most radio-therapy institutions to evaluate the degree to whichthe actual delivered radiation therapy matches theplanned treatment.

Definitions

In the past the terms portal film, beam film andverification film have been used in an inconsistentmanner. The definitions given below will serve toclarify future discussions.

Portal Radiograph: A radiograph produced byexposing the image receptor to the radiationbeam which emanates from the portal of a therapyThree types of portal radiographs aredefined below.1. Localization Radiograph A portal radio-graph produced by an exposure which is shortcompared to the daily treatment time requiredfor' that treatment field. (Such images arefrequently called localization films, beamfilms, or port films).They can be used in an interactive manner toadjust the patient set up and field boundariesprior to the delivery of the major portion ofthe daily treatment.2. Verification Radiograph: A portal radio-graph produced when the image receptor isexposed to the entire treatment delivered withthat field. This requires the use of a rela-tively insensitive detector, e.g. a slow film.3. Double Exposure Radiograph: Localizationradiograph, produced by a sequence of twoexposures, first to a shaped treatment field,then to a larger rectangular field. Theresulting image serves to locate the treatmentfield borders with respect to the patient's

-3-


anatomy.Simulator Radiograph: A radiograph produced byexposing the image receptor (film) to the beamof a simulator unit. The simulator unit isusually a diagnostic quality x-ray unit which iscapable of mimicking the geometry and movementsof the radiation therapy unit.

The Need For Portal Radiographs

There is some evidence that accuracy in beamalignment is related to the use of portal filmverification. Several studies have been published(l-3) which show a link between decreased localiza-tion errors and frequency of verification films.Marks et al pursued a six year study of localizationerrors for patients treated with extended mantlefields. A significant decrease in percent localiza-tion error was demonstrated as the number of verifi-cation films per patient increased. Morespecifically, they showed that increasing thefrequency of verification films from an average ofnine per treatment course to twenty-four decreasedthe frequency of a localization and field designerror from 36% to 15%. The information provided bythe verification films enabled the physician tomodify patient positioning and the field blocking.The authors recommend that for complex fields with aknown high error rate. daily verification films betaken until a reproducible, accurate setup isestablished.

Regions of high setup error rate were describedby Byhardt et al (2). In a retrospective study, theymeasured the frequency of localization errors by com-paring localization and verification films to simula-tion films. The average error rate was 15% with awide variation, depending on the site being treated.Not surprisingly the highest error rate was prostateand bladder cancer (37% and 27% respectively) wherepatient anatomy is less conducive to precision setup; the lowest rate was for primary and secondarybrain tumors (6% and 2% respectively). For the siteswith high error rates they recommend detaileddescription of the setup tattoos and also frequentfilm checks.

Another more recent study describes the use ofportal films to analyze variations from the plannedtreatment of the anatomical volumes treated for 71patients (4). An average standard deviation ofapproximately 3 mm is reported independent of site.The authors show, however, far greater differencesbetween portal and simulator fields relative topatient anatomy, with the mean worst case discrepancy(averaged over all sites) of 7.7 mm.

-4-

Stephen Backmeyer


It is difficult to draw specific recommendationsfrom these studies. However, two general principlesare clear: portal films are essential to accurateradiation therapy and frequent filming may berequired for difficult treatments.

Pattern of Use

A questionnaire was sent by the AAPM Task Groupon Portal Film Quality (TG28) and a response wasreceived from 158 institutions (5). An analysis ofthe responses showed that 90% of the institutionstake portal films on the first day of treatment formore than 75% of their patients. In contrast, only40% of these institutions repeat the check for thesepatients on a weekly basis. Considering the data ofHarks and Byhardt, one must question whether suchconfidence in treatment reproducibility is justified.

Another survey question relates to the use ofverification films (V-film). The responses revealthat while 30% of the institutions use this techniqueoccasionally, less than 10% use it on a regularbasis. This may reflect the reputation for poorimage quality that verification images have acquired.We shall see later that this reputation is notnecessarily justified when proper technique is used.

II. THE PROBLEM- -

The poor quality associated with high energyportal film imaging is, in general, caused by amixture of several factors:

1) Poor contrast due to the predominance ofCompton scattering which takes place at mega-voltage energies. For such images, there isno strong dependence on atomic number (Z) andtherefore very little of the differentialabsorption seen in diagnostic radiology.2) Image degradation due to scattered photons,which cannot easily be removed, and secondaryelectrons.3) Blurring of structures caused either bylarge source or focal spot size or patientmovement due to long exposure times. Thisunsharpness is enhanced as patient to filmdistance is increased.4) Beam edge "fuzziness" that makes itdifficult to determine the field edge inrelation to anatomy. This is a combination ofcollimator and phantom penumbra. The apparentpenumbra is derived from the collimator geometryas well as scattering in the phantom, althoughthe latter is usually the predominant factor forportal films. Also, in the case of an acceler-

-5-

Stephen Backmeyer


ator, the penumbra is generally greater in theradial (bending) plane. Galvin et al (6) haveobserved a large difference in the collimatorpenumbra of 6 MV accelerators from two differentsuppliers.5) Poor quality portal imaging can also becaused by bad technique. For example, asurprising number of low quality portal filmsare caused simply by improper exposure. As asecond example, if the front screen of thecassette is too thin, electrons exiting from thepatient have sufficient range to reach the film.

III. PRESENT UNDERSTANDING

Cassette Front Screen

The port film image is not formed directly fromthe incident primary photon beam, but rather fromCompton recoil electrons produced in the vicinity ofthe radiographic film. If no screen is used thenelectrons emanating from the exit surface of thepatient (Fig. 1A) and/or treatment couch are respon-sible for producing the radiographic image. Thespatial variations in electron fluence are clearlyproportional to the photon fluence transmittedthrough the patient, and thus contain image andcontrast information. These electrons, however, areobliquely scattered and non-uniformally attenuated,and thus produce an image with undesirable levels ofblur and contrast.

This image degradation can be greatly reduced byplacing a metal screen in close contact with thefilm, with the screen being thick enough to absorbthe shower of scattered electrons from the patient.The radiographic image information is then containedin the spatial variation of the x-ray fluence inci-dent on the metal screen. This in turn causes theemission of Compton electrons from the screen itselfwhich, being in good physical contact with the film,results in a better quality image (Fig. 1B).

A striking difference can be seen between portalradiographs using the same x-ray film but taken withno screen, insufficiently thick metal screens, andadequately thick metal screens. The increased screenthickness causes some loss of resolution since elec-trons originating within the screen now reach thefilm from more distant points and scatter laterallyin the process. But if the thickness is reduced toimprove resolution, electrons emanating from thepatient will reach the film and reduce contrast. Itfollows that for a given screen thickness in gm-cm2,resolution is expected to be best for screens ofrelatively high density (e.g., lead, copper etc).

Stephen Backmeyer

1A) Note that the images formed on the film (F) by electrons scattering at random anglesfrom different points within the patient. Note in both figures the dots are a cruderepresentation of the photon beam and the solid lines of the scattered electrons.

1B) In this configuration the cassette screen (C) absorbs the electrons which are scat-tered from within the patient. The image formed on film (F) is the result of electronswhich emanate from the screen (C) itself, thereby forming a sharper image. In thisparticular case the rear screen (P) is plastic and does not significantly contribute to theimage formation.

-7-

2) Plot of the scatter to primary ratio as a function of screen thickness for standard air gapgeometry. The details of the experimental method are described in Reference 7.

-8-


In two papers by Droege and Bjarngard (7, 8) theauthors point out that the metal screen cannotincrease the film gamma but can increase the overallcontrast by reducing the scatter to primary ratio,S/P. In this case S refers to the film dose due toboth scattered photons and electrons which originatein either the screen or the patient (Fig. 2). Theprimary dose P is due to unattenuated photons. Ifthe screen is too thin the S/P ratio will be high andan image of poor contrast will result. Droege'smeasurements at 4 MV and 8 MV showed that for eachenergy the S/P drops with increasing screen thick-ness, and that there is no significant differencebetween copper and lead screens once the screenthickness approaches the "build up" thickness.Beyond this thickness there is little decrease inS/P. For 8 MV it can be seen from Figure 2 that ascreen thickness of approximately 1 gm-cm-2 is quitereasonable for these high energies. This correspondsto a 0.9 millimeter thick lead or a 1.1 mm thickcopper front screen. It should be noted that for Co-60 beams there Is an enhanced response of thin leadscreens to the low energy scattered photons due tophoto-electric absorption in the screen. This leadsto an S/P which is 25% higher for a front screenthickness of 1 mm as compared to 2 mm of lead (7).Thus, at Cobalt-60, 1 mm is sufficient for copper,but 2 mm lead Is required for optimum contrast.

Results from the survey (5) mentioned earliershowed that of the 158 Institutions responding morethan 20% of the cassettes lacked any front metalscreen. In addition, of the 70% who used lead frontscreens, more than 10% were of thickness less than orequal to 0.1 millimeter, hardly adequate for mega-voltage radiography.

Metal Rear Screen

Ordinarily, there is little photon radiationscattered back to the film from structures beyond.For this reason a rear screen generally has littleeffect on image contrast. If electrons are scatteredback toward the cassette, a rear screen with athickness comparable to the maximum electron rangemay be used to stop such electrons. However, it ispreferable instead to minimize the source of thebackscattered electrons since the addition of such arear screen can significantly increase the weight ofthe cassette system.

A rear screen can affect speed and resolution.Speed is increased as much as 1.8 times when a high Z(e.g., lead) rear screen is used (9). That is, thefilm exposure is decreased by almost a factor of twodue to the backscatter of electrons from a high Z

-9-

Stephen Backmeyer


rear screen. A low Z rear screen will provide littlespeed or resolution change since few electrons arebackscattered from such materials. But such a screenwill reduce the artifacts caused by electrons back-scattered from structures beyond the screen.

The electron backscatterinq between high Z frontand rear screens produces a "cross-over" similar tothat which occurs in diagnostic radiology whenluminescent screens are used. Thus, a loss in reso-lution is expected when rear screens are used. Sucha loss has been documented through a dramatic changein the MTF for single emulsion films when a rearscreen is added (8). The degradation is expected tobe less severe for double emulsion films. An obser-ver study by Reinstein et al (10) did not findsignificant degradation in image detectability (or"film quality score") due to the presence of a rearmetal screen provided that good front screen/filmcontact was maintained. It appears that the reso-lution decrease caused by the presence of a rearscreen is overshadowed by the image degradation dueto the use of double emulsion film and the unsharp-ness caused by the finite source (target) size in the"air gap" geometry. If a rear screen is not used,the rear of the cassette comes into contact with thefilm and in effect becomes the rear "screen". Assuch, it should be of a low Z material (e.g., alumi-num or plastic) to minimize the backscattering ofelectrons.

Luninescent Screens

Luminescent screens are not expected to beuseful in portal imaging. In spite of their poten-tial to increase film contrast for a given film (ll),a reduction in subject contrast is expected due totheir sensitivity to secondary electrons scatteredfrom the patient. To exclude such electrons, theluminescent screen must be fronted by a metal screen.However, this combination is expected to have reso-lution inferior to a metal screen (8).

Cassette Design

The principal of good screen film contact is asimportant in therapy as in diagnostic radiology. Thecassette provides the obvious functions of protectingthe film from light and the screens from mechanicaldamage. However, it also serves the important func-tion-of providing intimate contact between the screenand the film. Many cassettes fail this latter re-quirement. For example. thin plastic or cardboardcassettes provide inadequate film-screen contact.Even rigid aluminum cassettes with rear panel pres-

-1O-


sure bars may warp and exhibit non-uniform film-screen contact if damaged or poorly constructed.Certain commercially available cassettes are con-structed with bowed pressure plates which aredesigned to maintain a uniform film/screen contact;this feature makes them particularly suitable forportal radiography. A wire mesh imaged in contactwith the cassette/screen can be used to evaluate theeffectiveness of film screen contact over the entiresurface of the image receptor.

If the rear of the cassette or its supportstructure contain moderately high atomic number (Z)materials, significant electron scattering backtoward the film can result. This reduces contrastand/or creates image artifacts. Therefore, high Zmaterials should be avoided in the construction ofthe cassette backing or its support structures.Otherwise, a rear screen may be required.

Image Quality and Beam Energy

Observation suggests a degradation of portalfilm quality as beam energy is increased from the lowmegavoltage range (4 MV and 6 MV) to the higherenergy range (10 MV and up). This appears to beattributable to changes in both contrast and resolu-tion, although the relative importance of thesefactors is unclear.

Subject contrast undoubtedly decreases as beamenergy is increased from the diagnostic kV range tothe therapeutic MV range. This is due to the reducedprobability of photo-electric interactions. Withinthe megavoltage range, however, Compton interactionsdominate. This statement may not be true for veryhigh energy beams (i.e., >20 MeV), where pair produc-tion is also of importance, and this is discussedbelow. If the effect of multiple scattered photonsis ignored, contrast is expected to decrease as thephoton energy increases, due to the decreasing proba-bility of Compton interactions. However, the magni-tude and direction of the scattered photon fluencealso change with energy, with scatter being moreforward peaked as energy increases. This has beentheoretically analyzed by Amols et al (12) usingdifferential Compton cross sections. The theory isconsistent with results [i.e., image contrast (asmeasured by a parameter termed "visual contrast")decreases significantly as beam energy increases from4 MV to 15 MV]. These results were derived in arelatively low -scatter geometry [i.e., with a thinphantom (8 to 9 cm) and a 10 x 10 cm field size].

In high scatter geometry, however, contrast isnot so severely affected at the higher megavoltageenergies. Droege's (7) measurements with a thick

-11-

Stephen Backmeyer


phantom (20 cm) and a 30 x 30 cm field size indicatedonly a slight contrast reduction from 4 MV to 8 MV.This is also consistent with the theory of Amols,since increased field size increases the scatteredphoton fluence at the center of the image (thusreducing contrast). The contrast reduction is slightfor high megavoltage beams (10 and 15 MV) since scat-ter tends to be forward directed. At 4 MV, scattergenerated near the periphery of large fields is lessforward directed and more likely to degrade contrastat the center of the image. Accordingly, the signi-ficant contrast advantage observed at low megavoltageenergies under low scatter geometry tends to be lostif large field sizes are used. Increased patientthickness has a similar effect. Both Amols andDroege measured significant increases in contrast asthe film-screen detector is separated from the phan-tom by an air gap.

Image resolution also decreases with increasingphoton energy. Droege (8) measured the modulationtransfer function (MTF) of film-screen combinationsand demonstrated significant reductions in detectorresolution from 4 MV to 8 MV. This is explained bythe increased range of the Compton electrons gene-rated in the screen.

At very high energies, however, (i.e., 20 MeV),photon interactions via pair production become acompeting process to Compton scatter. At 10 MeV forexample, 23% of all photon interactions in wateroccur via pair production. At 20 MeV, the percentagerises to 44%. In addition, unlike the Comptoneffect, pair production is Z-dependent. Thus thepossibility exists that portal film contrast mightactually improve at very high energies. Thisphenomenon, however, has not been explored experi-mentally. Further, it should be noted that even veryhigh MV photon beams contain relatively small frac-tions of high MeV photons.

To date no comprehensive study of differentfilms in combination with a standard metal screencassette and megavoltage beam has been done.Certainly a desirable localization film should have ahigh film gamma. Some rather scanty evidence hasbeen published (10, 13) which suggests differences inquality in the megavoltage x-ray range for severalavailable films. A more complete study of thisquestion is to be encouraged.

Noise in portal images has not been seriouslystudied by previous investigators. This is unfor-tunate since noise is known to affect the perceptionof low contrast objects and film-grain noise is known

-12-

Stephen Backmeyer


to be visually evident in radiographic images.Members of this task group have found fine grain film(e.g., Kodak Verification Film) to perform surpris-ingly-well in visual detection tests 'when compared tofilms having similar film gamma. The low film-grainnoise is thought to be partially responsible. Lownoise film may be especially advantageous if postprocessing of the original radiography is performed.Investigations concerning the role of image noise areto be encouraged.

Other factors to be considered when choosing themost suitable film for portal radiography are: speed,storage, handling convenience, cost. These maynecessitate compromise with optimum image quality andeach other.

Is there measurable degradation in quality whenusing the convenient (but more costly) "Ready Pack"film in its light tight wrapper enclosed in a cas-sette? The study by Reinstein et al (10) tested a"high quality" (copper screen) cassette using XTLfilm with and without its paper packaging on a 10 MVlinac beam. Three situations' were compared:

(1) XTL alone,(2) XTL in Ready Pack with paper insert removed

and,(3) XTL in Ready PackAlthough the results show all 3 situations to be

at least 'acceptable", the data does confirm theexpected degradation in quality due to the insertionof the wrapping materials between the film and thescreen. For the above 3 situations, the 50% detec-tion thicknesses (i.e., the thickness of a PVC testobject which could be correctly identified 50% of thetime) were found to be 11.4 mm. 12.9 ma, and 13.7 mmrespectively. Thus, in using- Ready Pack wrappers,one suffers a decrease in PVC (and presumably bone)detectability of more than 2 mm, which may be clini-cally significant.

Proper Exposure

What is the best optical density range forviewing portal radiographs using conventional hospi-tal view boxes? A recent observer study (10) using aportal film phantom (13) has shown that the lowcontrast detectability was "excellent" in the opticaldensity range from 1.6 to 2.0, and "acceptable" downto 1.2 and as high as 2.3. The films in this studywere viewed under good conditions with essentially notime limit imposed.

A technique chart which consists of tabulatedvalues of exposure parameters, is useful in producingsuitable optical densities in radiographic images.Technique charts for portal films are quite easy to

-13-


determine and use and a simple methodology is de-scribed in the literature (14).

Observer Study: Results

The observer study, previously referred to,evaluated a selection of 23 film/screen/cassettecombinations using a 10 MV linear accelerator. Theresults suggested-that portal film cassette systemsfell into three categories.

1) Excellent: These systems all had metalfront screens of either lead or copper. Thelead screen systems in this category all hadthicknesses of at least 0.8 mm and those withcopper front screens had thicknesses of at least1.0 mm. (Copper screens of less than 1.0 mmthick were not considered in this study.)

The cassettes were conventional rigiddiagnostic cassettes made of aluminum or rigidplastic, all assured a close contact between thefilm and the front screen. No significantdifference was seen between the lead or copperscreens of the same thickness, in this group.There was no paper separating the film from thefront screen as in the "Ready Pack" format.(There were 13 systems which fell into thisgroup.)2) Acceptable: The system in this categoryalso used rigid cassettes as above but eitherhad thinner-front screen (0.5 millimeterstainless steel or 0.3 millimeter lead) or hadless reliable film screen contact, e.g., using"Ready Pack" film or interleaf paper separatingthe screen from the film. (There were 5 systemsin this category.)3) P o o r : T h e s e systems were noticeablyinferior and included cassettes with poor filmscreen contact, (warped soft cardboard or steelsandwiches) and front screens of .2 millimeterlead or less. (The remaining 5 were in thisgroup.)

Admittedly, the cutoff points for these group-ings were arbitrary, but meaningful conclusions canstill be drawn. The data are consistent with theprevious discussions regarding desired screen thick-ness, cassette construction and film screen contact.It should also be pointed out that these results wereobtained using a 10 MV linear accelerator exclu-sively, and may not be easily extended to very highenergy. It is hoped that further exploration ofoptimum film screen cassette combinations will becarried out at higher energies.

-14-


Viewing Conditions

A. Image BrightnessThe resolution of the eye is strongly dependent onimage brightness, so it is desirable to assure anappropriate level of view box luminance. Measure-ments performed by members of this task groupindicate average luminance levels of 1300 to 1900 cd.m-2. It should be recalled that luminance is a meas-ure of " brightness" at the surface of the radiator orview box, while the illuminance of an area, a sensor,or a viewers eye is the flux density incident on thatarea (measured in lux). Assuming a value of 1300 cd.m-2 , the maximum resolution of the eye is about12 lp/mm (line pairs per millimeter) at a viewingdistance of 25 cm (15). although reduced resolutionresults if the eye accommodates to a darker sur-rounding environment (16). Radiographs with anaverage density of 1.6 reduce the illuminance by afactor of approximately 40, at which the resolutionof the eye is reduced to about 5 lp/mm. This reso-lution loss is of little consequence since minimallymagnified portal images convey almost no objectinformation beyond 5 lp/mm even in the absence ofpatient motion (17). However, if a film image is toodark (e.g., a film density over 2.5) and/or the viewbox too dim, resolution can drop below 4 lp/mm andsignificant object information might not be appre-ciated by the eye under these limited conditions. Inaddition, if the film optical density Is high enoughto be on the shoulder of the H&D curve there will bea loss of contrast and therefore Image information.

B. Ambient LightThe contrast detection of the eye is

approximately 2% of the illuminance to which the eyeis adapted, provided the difference in illuminance isgreater than about 0.3 cd. m-2 (18). Assuming theeye suitably accommodates to the relatively lowillumiation level of the radiograph, this impliesthat radiographic density differences as small as0.01 are detectable. When a radiograph is viewed ina situation of relatively bright ambient light, theability of the viewer to detect small changes incontrast is degraded. This is because the contrastdetection limit of the eye is now 2% of the combinedilluminance from the radiographs and ambient lightsources.C. Practical Implementation

Proper film viewing requires uniform and asufficiently bright level of view box luminance(about 1600 cd.m-2). An additional high intensity"hot light" should be available to provide sufficientluminance for slightly overexposed small portions of

-15-


the image. Such a light should provide at least a 2Xincrease in luminance (preferably variable up to 4X)compared to the conventional view box. When a hotlight is available, slight overexposure can betolerated. By comparison, an underexposed film hasreduced contrast (i.e., lower gamma) which cannot beCorrected by altered viewing conditions. The preference for overexposed (compared to underexposed) filmsshould be considered in the preparation of techniquecharts (i.e., the selection of a target density).

Viewing room light levels should be reduced sothat the illuminance at the viewers' eye from theambient sources is less than that from the radiographitself. That is, room lights should be dimmed, andunused view boxes should be turned off or covered.Even the view box being used should be appropriatelymasked if unexposed or lightly exposed areas of thefilm transmit significant extraneous light to the eyeof the observer.

Film Processor Quality Assurance

In diagnostic radiology, it has been documentedthat unsuitably darkened films are often due toimproper film processing. In one study, 30% of allretakes, due to improper film density, were attrib-uted to processor variation (19). In another study,both the number and type of film retakes were foundto be highly correlated with processor "speed" varia-tions (20). Similar retake problems may be expectedto occur In radiation therapy departments if filmprocessor quality is not assured. It is thereforerecommended that all port film processors be evalu-ated daily. A test, requiring only a few minutes,should be performed in the morning so that correctiveaction (if necessary) can be completed before clini-cal films are developed. The reader is directed toother references for details concerning the estab-lishment and maintenance of a film processor testingprogram (21, 22, 23). Here, a protocol is brieflyoutlined, primarily to indicate the ease with whichsuch testing can be performed.

Procedure: A sensitometer is used to exposeadjacent portions of a test film toincreasing illumination levels.

"steps" ofTypically, the test

film is selected to be the same type as that usedclinically, but is taken from a supply reservedexclusively for processor testing. Once theprocessor has been given sufficient time for fluidtemperatures to stabilize, and temperatures arewithin acceptable limits, the exposed test film isfed into the processor. After development a densi-tometer is used to measure the film density ofselected steps. The measured values are compared to

-16-


the range of acceptable values to determine if theprocessor is functioning properly.

The steps to be measured are selected on thebasis of the information desired. One step should beselected to indicate processor "speed". This stepshould have a density on the steep portion of theFilm response curve and have a density of at least1.0 when the processor is functioning properly.Measured values for this step should be within about± 0.1 optical density units of the expected value forwell controlled processors. Variations exceeding± 0.2 should not be tolerated. It is often recom-mended that steps of greater and lower density aremeasured so that "fog" and "contrast" can bemonitored. However, if the speed measurement iswithin acceptable limits fog and contrast willgenerally be acceptable. Thus, a single speeddetermination generally provides adequate processormonitoring. Nevertheless, baseline fog and contrastvalues should be established, since these parametersare often helpful in diagnosing a processor problemif the speed value is found to be unacceptable.

The most critical element in processor testingis reproducibility. Each day the test film should bedrawn from the same supply (same box), and the sameemulsion should be exposed (the two emulsions of adouble emulsion film are not always identical). Thefilm should be fed into the processor identicallyeach day (e.g., low density step first). Thedensitometer accuracy should be checked for day-to-day reproducibility by means of a calibrated filmstrip. To increase precision, the test film can beexposed twice (at different locations) so that anaverage speed value can be determined. This requireslittle additional time or effort.

Other: In addition to the daily processor test-ing, clinical films should be examined for processorartifacts which may result from inadequate processormaintenance (24). An occasional test is also recom-mended to evaluate darkroom safelights and possiblelight leaks. The reader is directed to the appro-priate references (25, 21) for details of safelighttest methods.

IV. Recommendations and Practical Considerations- - -

Recommended Indications for Frequent Portal Filming

It is recommended that frequent portal films betaken in the following situations:

1) An uncooperative patient.2) Treatment of a critical site where accuracy

on the order of 3-4 millimeters is needed.

-17-

Stephen Backmeyer


3) A difficult set-up such as an obese patientor one with moveable, unstable skin marks.

4) Treatments where matching of field edgesis important (e.g., breast, mantle para-aortic, Total CNS).

5) Pediatric treatment.

Considerations for Obtaining Good QualityPortal Radiographs

In general, one should be wary of using visualimpressions to identify the cause of image qualitydifferences. For example, resolution loss may bevisually indistinguishable from contrast loss (17).Whenever possible, objective measures of noise,contrast and resolution, should be obtained forcomparison. Unfortunately, It is difficult tocombine such measures of image quality into a singleparameter which is indicative of observer performancefor a particular task. Therefore, visual tests arepreferred when comparing the clinical utility ofimaging systems. If the visual test is based onphantom images, phantom design should attempt tosimulate the tasks required In clinical portal filmevaluation, e.g., visualization of bony landmarks,field edges, etc.

The following are important for obtaining goodquality high energy radiographs:

1) Excellent film screen contact -a) the use of high quality rigid commercialfilm cassettes, especially those which arespecifically designed to provide good filmscreen contact.b) Flat (unwarped) screens.

2) Adequate screen thickness -a) approximately l-2 mm of lead or copperwill be suitable over the energy range of 4MV to 15 MV. (For Cobalt-60, a copper frontscreen approximately 1 mm thick is prefer-able.)b) Optional rear screen for "intensifi-cation" or backscatter artifact reduction.

3) Long term stability -a) the cassette/screen system chosen toavoid degradation through bowing, warping,screen damage (scratching), loose hinges,etc.Towards this end, copper screens are clearly

superior to lead.

-18-

Stephen Backmeyer


Practical Considerations

There are some practical considerations whichsometimes preclude using optimal filming conditions.These include:

1) Cassette Weight -Although the thick front screen may improvethe image quality and the rear screen willreduce exposure time (and, therefore, thelikelihood of blurring), they also tend tomake these cassettes extremely heavy. Acompromise may be required.

2) Cassette Placement and Mounting -In general localization films provide bettervisualization of anatomic structures when thepatient to film distance is kept small.There are, however. some trade-offs. Atsmall patient-to-film distances unsharpnessis minimized. On the other hand, for small

3)

distances there is more loss of contrast dueto patient scattered radiation (7) than atlarger distances.Often it is considered desirable that simu-lator and portal/localization films be takenat the same, standard magnification. Thiscriterion may result in a larger than optimalpatient-to-film distance. On most machinescassettes can be supported under the treat-ment couch on special rails. For othergantry angles, several cassette holderdesigns exist. Some attach to the couch, butseveral free standing cassette holders arecommercially available which are more or lessconvenient to use, depending on design (26).It should be noted that several of these donot assure that the film is perpendicular tothe radiation beam axis. Care must be takenin the use of these.One particular type of holder, now in use atseveral centers (27), allows the cassetteholder to be mounted on the gantry counter-weight so that it is always aligned with thebeam central axis during any isocentricgantry rotation. Its advantages are ease ofuse and standardization of magnification. Adisadvantage is that the isocenter to filmdistance must be 40-50 cm, which can produceexcessive magnification for large fields(e.g., "Mantles") and increased unsharpness.Localization vs. Verification -The use of "V" film is not very popular,probably reflecting the poor quality imagesassociated with radiographs taken using the"Ready Pack" alone without cassette or

-19-


blurred by patient motion. Recent experi-ments (10), however, show that acceptablequality portal verification films can betaken using the "V" film in a well designedportal film cassette with adequate metalfront screen. This study does not, however,take into account the potential blurring dueto the increased probability of patientmotion during the long exposure time. On theother hand, the advantages of the "verifica-tion" film technique Is that it involves lesstechnician time, uses a finer grain emulsion,(which thereby reduces noise), and can beused to document patient motion during theentire treatment fraction. Its major disad-vantage is that the "double exposure" tech-nique can not be used and, therefore, thetreatment field is not viewed in the contextof its anatomic surroundings. The quality ofverification films can be improved if the "V"film is taken out of its "Ready Pack" envel-ope and used In a high quality portal filmcassette.

4) Film Choice & Exposure Time -As stated earlier, a desirable localizationfilm should have a high gamma, fine resolu-tion, and a speed slow enough to permitoptimization of optical density (particularlynecessary in double exposure techniques) butfast enough to reduce patient dose and motionblurring.The time needed for optimal exposure of aportal film can vary by a factor of 10according to the selection of differentradiographic film sensitivities as well aswhether or not a rear metal screen is used.Short exposure times reduce potential motionerror, as well as unnecessary exposure touninvolved regions using the "double expos-ure" technique. Long exposure times,however, allow greater adjustment precisionin selecting the optimal technique to producea good density film. Exposure times can bereduced significantly, through the use ofrear screens. While it has been shown thatthe use of rear screen reduces resolutionthere will not necessarily be a noticeablereduction in image quality due to severalother effects.

5) Daylight vs Darkroom Cassette Loading:Ready Pack -

This issue involves questions of convenience,quality, and philosophy. It is certainlymore convenient for technologists to be able

-20-


to load and reload therapy cassettes withoutcarrying the cassette to the darkroom. Inthis practice Ready Pack film can be usedeither with specially designed cassettes orin a currently available smaller format whichenables it to fit a standard therapy cas-sette. As discussed earlier, there is amodest decrease in quality with Ready Packand an increase in cost.Another convenience suggested for the use ofReady Pack film is the ability to delayprocessing of all films until the end of thetreatment day, when they can be processed inbatch for review. Herein lies the philo-sophical question: What is the ideal use ofthe portal localization film? It is clearlymore in keeping with a strict interpretationof quality assurance review for the portal/localization film to be processed and evalu-ated immediately, with the patient still onthe treatment table. In this case, thefeedback from the radiotherapist is used toadjust the field prior to treatment delivery.It is understood, however, that the compro-omise of daylight loading and deferredevaluation may be a necessary expedient.

A final word about the use of portal films forevaluating the accuracy of radiation therapytreatment. Often the difficulty in determiningwhether the actual delivered treatment is identicalto the planned or simulated treatment is not due tothe quality of the megavoltage image but rather tothe lack of a common reference frame on which to basethe evaluation. Besides poor image quality othergeometric conditions which render this task moredifficult are differences in magnification and non-orthogonal film positioning. A device which wasdesigned to minimize these latter two problems and toenhance the therapists' ability to evaluate thedegree of difference between the simulator and portalfilms is called a "graticule" (28) which projects aprecise scale on the image to be used as a commonreference frame.

V. IMAGE PROCESSING

Photographic Method

Several methods for enhancing the quality ofportal films have been reported. The simple andinexpensive photographic technique described byReinstein and Orton (29) can be performed usingequipment generally available in the radiotherapy

-21-


department. A contact copy of the original portalfilm is produced within the department's darkroomusing a ceiling mounted low intensity light bulbconnected to a timed switch (Fig. 3). A variety ofdifferent films are suitable for use as the copymedium, although the wide latitude XL film has beenfound Lo be very practical, since this film largelyprevents the loss of the field edge caused by theoptical density dropoff in the penumbra. The preciselocalization of field edge is critical for the properinterpretation of the portal film. The contact copyis processed using an X-Omat processor, and theresulting film is a reversed ("black bone") imagewhose effective gamma (contrast) is the product ofthe gammas of the original and the copy film.

Sometimes the single enhancement is adequate butoften it is necessary to repeat the contact COPY pro-cess a second time. This yields a final high-gammaimage with the original ("white bone") polarity.Using this technique- extremely high contrast imageshave been achieved which reveal good bony detail,adequate for portal evaluation. With certain filmseffective gammas of 30 and greater have been achievedbut better results with less noise are obtained inthe gamma range of 15-20. Several drawbacks of thistechnique are the slight loss in resolution, themagnification of film processor noise, the sharpdecrease in latitude, and increased processing time.

A recent study (30) has shown that under goodviewing conditions with "unlimited" viewing time theprobability of small object, low contrast detectionwas not statistically different between the photo-graphically enhanced and the unenhanced images.However, when viewing conditions worsen and viewingtime was limited, the average quality scores weresignificantly better for the enhanced films. Thus,the decision to incorporate photographic contrastenhancement into the portal film quality assuranceprogram should depend on the viewing circumstances ofeach particular radiotherapy department.

Digital Techniques

Alternative methods using digital imagingtechnology have been reported (31, 32, 33, 34) toachieve -similar results. In addition, severalproducers of commercial "tele-radiography" systemshave been applying video enhancement techniques tothe improvement of radiotherapy portals (35). Mostof these systems digitize the film via a high qualitylow light video camera or laser scanning techniques,and process the data with a specialized graphicprocessor or digital imaging computer. The typical.commercial tele-radiography system produces a 512 x

3) Diagram illustrating the photographic contrast enhancement technique, for details seeReference 28.


512 electronic matrix with 256 gray levels, with"real time" digitizing capabilities, electronic zoom,and disk storage facilities. While the use of suchcommercial systems for the radiotherapy department ismerely a spinoff of the much larger diagnosticimaging market, several of these manufacturers aremaking a serious effort to develop this new applica-tion.

A variation on this concept being pursuedcommercially is the use of a reusable imaging medium(RIM) to replace conventional radiographic film forthe production of portal imaging and diagnosticradiographs (36). Ordinary cassettes are used duringthe exposure of the RIM (a photo stimulable lumines-cent material) and the image information is captured.Afterwards, the RIM is read by a laser scanner toproduce the digital image. The RIM can be erased andreused many times. Such films can be loaded indaylight and scanned in less than a minute to producea 2048 x 2048 point sample matrix with 4096 shades ofgrey. Research is currently underway to develop theideal RIM material for high energy therapy imaging.

Photographic vs. Digital Enhancement

The major advantage of the contact copy gamma-multiplication technique is that it can be done withavailable equipment at small expense and produceshigh quality enhancements. Although the initialinvestment for a digital enhancement system is high,it has the following benefits:

1) Enhancement algorithms can be chosen to suitindividual situations. Software for edgeenhancement, histogram equalization, gammacorrection, and low frequency filtering areavailable. For large fields with severevariation in patient thickness, these algorithmscan be used to optimize the display of availableinformation.2) Image storage and fast retrieval is easilyincorporated into the system.3) Software can be developed for superimposinganatomical landmarks as well as field outlinesof films taken on different simulation andtreatment days. These can be used to aid in thecomparison of planned versus executed treatmentas well as repeatability.

VI. ALTERNATIVES TO CONVENTIONAL PORT FILMS

One method to improve the quality of portalfilms is the creation of a special low energy "portfilm mode" on the linear accelerator. Severalmanufacturers have provided this option in order to

-24-

Stephen Backmeyer

Radiotherapy Portal Imag ing Quality

counteract the degradation in quality seen at thehigher energies. It is, of course, unnecessary inthe newer dual energy machines with beam energies aslow as 6 MV.

Another alternative is the gantry mounted diag-nostic x-ray tube, an idea which dates back to 1958on some cobalt units (37). In a recent publication,Biggs et al (38) describe a system which has thecapacity for checking fields with fabricated blocks.The x-ray tube is mounted on the gantry at a fixedoffset angle from the therapy beam target. The"portal film" is taken of the patient in the setupposition by a precise offset of the gantry. Earlyversions of this unit could only be used to checkalignment of the rectangular field edges while thisnew unit makes possible beam alignment of shapedfields using diagnostic quality films. A majordrawback is that this attachment prevents collimatorrotation so that all fields need special blocking anda specially designed rotating wedge tray was re-quired. It implies a rather time consuming setupand, in general, is not recommended as a "workhorse"unit in a busy clinic. A commercial version of thegantry mounted diagnostic tube is currently available(39).

An innovative application of the gantry mountedx-ray tube technique was developed by Shiu et al(40). This new approach is to superimpose the stand-ard megavoltage portal image on the diagnostic x-rayimage using a single film. With careful alignmentprocedures this technique can provide the radiothera-pist with "diagnostic quality" portal images.

The use of an on-line radiotherapy fluoroscopysystem was described several years ago by Bailey(41). In this setup, an E2 fluoro screen wascemented to a l/16" thick steel front screen. Theimage was intensified using a low light TV camera. A90° bend necessary for the side mounting of thissystem was achieved through the use of a planarmirror. The images were plagued by electronicinterference resulting from the linac. With thissystem. the entire treatment could be easilyvideotaped and used for patient motion studies andother teaching purposes.

Another effort at real time on line imaging wasreported by Partowmah and Lam (42) who use a scanninglinear array of silicon diodes on the exit side ofthe therapy beam. The array is mounted 150 cm fromthe target with a detector separation of 2 mm yield-ing an estimated 1.5 mm per pixel resolution. Thelinear detector array is mechanically scanned and animage reconstructed using digital processing andsignal averaging techniques.

-25-


Digital megavoltage imaging is being developedat several other centers using digital image proces-sing techniques (43, 44). With these systems theimage produced on a fluorescent screen is captured bya video camera and digitized in a 512 x 512 matrix.Using sophisticated image processing techniques theauthors have demonstrated the ability to produceclinically useful on line portal images in a matterof seconds.

Efforts are being made to use computer technol-ogy to help expedite and improve the evaluation ofpatient treatment accuracy. Ideally a computerassisted verification system can be used for anautomated "go/no go" treatment decision. A firststep towards this end would be the superposition ofthe shaped treatment field (as drawn on the simula-tor/localization film) on a digitized portal image.Even more exciting is the notion of accomplishingthis task in "real time" in an on-line imaging mode.Some very promising preliminary results in thisdirection have already been discussed.

-26-


1)

2)

3)

4)

5)

6)

7)

8)

9)

10)

11)

12)

13)

14)

15)

References

Marks, J.E., Haus, A.G., Sutton, H.G., Griem,M.L.: The value of frequent verification filmsin reducing localization error in the irradia-tion of complex fields, Cancer 37: 2755-2761,1976.

Byhardt. R.W., Cox, J.D., Horngurg, A.,Liermann, G.: Weekly localization films anddetection of field placement errors, Int. J.of Rad. Onc. Biol. Phys. Med. 4: 881-887, 1978.Marks, J.E., Davis, M.D., Haus, A.G.: Anatomicand geometric precision in radiotherapy,Radiology and Clinical Biology, 43: l-20, 1974.Rabinowitz, I., Broomberg, J., Goitein, M.,McCarthy, K., Leong, J.: Accuracy of radiationfield alignment in clinical practice, Int. J.Rad. Onc. Biol. Phys., Vol. 22, 1857-1867.Task Group 28 of Radiotherapy Committee AAPM,Reinstein, L.E., Chairman, PersonalCommunication.Galvin, J.M., Kumar, P.: Design considerationsfor radiation linear accelerators. Med. Phys.12: 675, (ABS), 1985.Droege, R.T., Bjarngard, B.: Influence of metalscreens on contrast in megavoltage x-rayimaging. Med. Phys. 6: 487-493, 1979.Droege, R.T., Bjarngard, B.: Metal screen-filmdetector MTF at megavoltage x-ray energies.Med. Phys., 6: 515-518, 1979.Haus, A., Rochester, New York: Personal

Communication.Reinstein, L.E., Lagueux, B.J., Alquist, L.,

Amols, H.I.: Evaluation of film screen systemsfor 10 MeV radiotherapy portal films. Mod.Phys. 11, 3: 395 (ABS), 1984.Springer, E.B., Pape, L., Elsner, F., Jacobs,M.L.: High energy radiography (Cobalt-60 andCesium-137) for tumor localization and treatmentplanning, Radiology, 78: 260-262, 1962.Amols, H.I., Reinstein, L.E., Lagueux, B.:A quantitative assessment of portal filmcontrast as a function of beam energy. Med.Phys., Vol. 13, No. 5, 1986.Lutz, W.R., Bjarngard, B.E.: A test object forevaluation of portal films. Int. J. Rad. Onc.Biol. Phys., 11, 3: 631-634, 1985.Droege, R.T., Stefanakos, T. K.: Portal filmtechnique charts. Int. J. Rad. Onc. Biol.Phys., 11: 2027-2032, 1985.Gregg, E.C.: Image manipulation in radiology,in physics of diagnostic radiology-proceedingsof a summer school held at Trinity University,San Antonio, Texas, 1971, ed. Wright, D.J.;

-27-

Stephen Backmeyer

16)

17)

18)

19)

20)

21)

22)

23)

24)

25)

26)

27)

28)

29)


Food and Drug Administration publication FDA 74-8006, 1973, p. 292.Davson, H.: The physiology of the human eye,Academic Press, 1972, New York, p. 237.Droege, R.T., Cytacki, E.P.: The significanceof screen resolution in treatment verification.Int. J. Rad. Onc. Biol. Phys., 8: 873-877, 1982.Davson, H.: The physiology of the human eye,Academic Press, 1972, New York, p. 137.Burkhart, R.L.: A basic quality assurance

program for small diagnostic radiologyfacilities, Food and Drug Administrationpublication FDA 83-8218, 1983,p. 24.Goldman, L.W.: Effects of film processing

variability on patient dose and image quality,in second image receptor conference: radio-graphic film processing, Proceedings of a Con-ference held in Washington, D.C., 1977, 61-63.Gray, J.E.: Photographic quality assurance indiagnostic radiology, nuclear medicine, andradiation therapy. Volume 1: The basic prin-ciples of daily photographic quality assurance.Food and Drug Administration publication FDA 76-8043, 1976.Gray, J.E.: Photographic quality assurance indiagnostic radiology, nuclear medicine, andradiation therapy. Volume 2: Photographicprocessing, quality assurance, and the evalu-ation of photographic materials. Food and DrugAdministration publication FDA 77-8018, 1977.Lawrence, D.J.: A simple method of processorcontrol. Med. Radio. and Photo., 49: 2-6, 28,1973.Mitchell, J.R., Lee, C.E.: The role of the

processor control. Med. Radiog. and Photog.,49: 2-6, 1973.Hurtgen, T.P.: Safelighting in the automatedradiographic darkroom. Med. Radiog. andPhotog., 54 32-38, 1978.Portal Film Holders available from Huestis

Machine Corp. Bristol, R.I.; Mick NuclearInstruments, Bronx, N.Y.; Varian Associates,Palo Alto, CA.; Engineering Prototype Services(EPS), Portland, OR.Personal Communication, described in Biggs et al(36), Joint Center, Massachusetts General Hospi-tal, Univ. of Pennsylvania, Univ. of Arizona.Van de Geijn, J, Harrington, FS, Fraass, BA: Agraticule for evaluation of megavolt X-Ray portfilms. Int. J. Rad. Onc. Biol. Phys., 8, 11:1999-2000, 1982.Reinstein, L.E., Orton, C.G.: Contrast enhance-ment of high energy radiotherapy films. Brit.J. Rad., 52: 880-887, 1979.

-28-

Stephen Backmeyer

Stephen Backmeyer


30)

31)

32)

33)

34)

35)

36)37)

38)

Reinstein, L.E., Alquist, M., Durham, M.:Evaluation of port film contrast enhancement.Med. Phys., 11, 3: 395 (ABS), 1984.Shalev, S., Arenson, J., et al: Digital

enhancement of treatment verification films.Radiology, 153: 154, 1984.Leong, J.: A digital image processing systemfor high energy x-ray portal images. Phys. inMed. and Biol., 29: 1527-1535, 1984.Meertens, H.: Digital processing of high energyphoton beam images. Med. Phys. 12: 111-113,1985.Sherouse, G.W., Rosenman, J., McMurry, H.L.,

Pizer, S.M., Chaney, E.L.: Automatic digitalcontrast enhancement of radiotherapy films.Accepted for publication, Int. J. Rad. Onc.Biol. Phys.Some of the commercial organizations attemptingto apply tele-radiography and image processingtechniques to the improvement of radiotherapyportal films are DataSpan, Orchard Park, N.Y.;Matrix Instruments, Inc., Roxbury, MA.; DigiRad, Palo Alto, CA; Kodak, Rochester, N.Y.Digi Rad, Inc., Palo Alto, California.Holloway, A.F.: A localizing device for a

rotating-cobalt therapy unit. Brit. J. of Rad.,31: 227, 1958.Biggs, P.J., Goitein, M., Russell, M.D.: A

diagnostic x-ray field verification device for a10 MeV linear accelerator. Int. J. Rad. Onc.Biol. Phys., 11: 635-643, 1985.

39) Haynes, Radiation Limited. Alameda. CA.40) Shiu, A.S., Hogstrom, K.R., Janjan, N.A.,

Fields, R.S., Peters, L.J.: A technique forachieving megavoltage portal images withdiagnostic quality, M.D. Anderson Hospital,Houston, Texas. Submitted for publication.Personal Communication.

41) Bailey, N.A., Horn, R.A., Kampt, T.D.:Fluoroscopic visualization of megavoltagetherapeutic x-ray beams. Int. J. Rad. Onc.Biol. Phys., 6, 7: 935-940, 1980.

42) Partowmah, M., Lam, W.C., Lam, K.C.: An on-lineelectronic portal imaging system for externalbeam radiotherapy. Brit. J. Radiology, 59:1007-1013, 1986.

43) Shalev, S. and Lee, T.: Personal Communication.44) Leong, J.: Use of digital fluoroscopy as an on-

line verification device in radiation therapy.Phys. Med. Biol., Vol. 31: 985-992, 1986.

-29-

Stephen Backmeyer

RADIOTHERAPY PORTAL IMAGING QUALITY · films, or port films). They can be used in an interactive...

Documents

Transcript of RADIOTHERAPY PORTAL IMAGING QUALITY · films, or port films). They can be used in an interactive...