Int. J. Radiation Oncology Biol. Phys., Vol. 54, No. 2, pp. 584 –591, 2002 Copyright © 2002 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/02/$–see front matter
PII S0360-3016(02)02955-3
PHYSICS CONTRIBUTION
ELECTRONIC AND FILM PORTAL IMAGES: A COMPARISON OF LANDMARK VISIBILITY AND REVIEW ACCURACY JON J. KRUSE, PH.D., MICHAEL G. HERMAN, PH.D., CHRIS R. HAGNESS, BRIAN J. DAVIS, M.D., PH.D., YOLANDA I. GARCES, M.D., MICHAEL G. HADDOCK, M.D., KENNETH R. OLIVIER, M.D., SCOTT L. STAFFORD, M.D., AND THOMAS M. PISANSKY, M.D. Division of Radiation Oncology, Mayo Clinic, Rochester, MN Purpose: To quantitatively compare a scanning liquid ion chamber electronic portal imaging device (SLICEPID) and an amorphous silicon flat panel (aSi) EPID with portal film in clinical applications using measures of landmark visibility and treatment review accuracy. Methods and Materials: Six radiation oncologists viewed 39 electronic portal images (EPIs) from the SLIC-EPID, 36 EPIs from the aSi-EPID, and portal films of each of these treatment fields. The physicians rated the clarity of anatomic landmarks in the portal images, and the scores were compared between EPID and film. Nine hundred portal image reviews were performed. EPID and film portal images were acquired with known setup errors in either phantom or cadaver treatments. Physicians identified the errors visually in portal films and with computerized analysis for EPID. Results: There were no statistically significant (p < 0.05) differences between film and SLIC-EPID in ratings of landmark clarity. Eleven of 12 landmarks were more visible in aSi-EPID than in film. Translational setup errors were identified with an average accuracy of 2.5 mm in film, compared to 1.5 mm with SLIC-EPID and 1.3 mm with aSi-EPID. Conclusions: Both EPIDs are clinically viable replacements for film, but aSi-EPID represents a significant advancement in image quality over film. © 2002 Elsevier Science Inc. EPID, Portal film, Image quality.
enables the development of on-line correction protocols and daily targeting adjustments (19 –23). In addition to aiding acquisition, the digital nature of EPIs can be exploited to enhance the portal review process. Studies have examined the process of subjective portal image evaluation by clinicians and have found a wide variation among reviewers in reporting setup deviations in port films (24, 25). Many EPID systems offer computerassisted image review with anatomy-matching routines and quantitative alignment analysis. Conceivably, these systems should enable a more precise analysis of portal alignment than the relatively subjective process of portal film evaluation. Despite their potential for improving treatment accuracy and efficiency, EPIDs have not gained routine clinical acceptance. Among the barriers to routine use is the perception that the image quality from commercially available EPIDs is inferior to that from standard portal film, and therefore EPIDs are not suitable for clinical treatment setup verification. However, although the spatial resolution of
INTRODUCTION A fundamental tenet of radiation therapy is that successful outcomes require accurate alignment of the treatment beam to the target tissue (1–3). Weekly portal filming is the routine clinical standard for ensuring accurate targeting of external beam radiation therapy (4). However, the importance of geometric accuracy has driven the development of devices that have the potential to monitor treatment accuracy more effectively than weekly port filming (5–13), with minimal increase in work load (14). Electronic portal imaging devices (EPIDs) can acquire images automatically with near real-time display, store them digitally, and provide quantitative analysis tools. Electronic portal images (EPIs) can be acquired with higher frequency than film with only minimal additional costs. Studies have shown that increased portal imaging frequency can reveal daily variations in patient alignment that are not observed with weekly filming (15–18). Furthermore, EPIDs provide the user with immediate patient alignment information, without the delay involved in processing a film. Instant image availability Reprint requests to: Michael G. Herman, Ph.D., Mayo Clinic, 200 First St. SW, Rochester, MN 55905. Tel: (507) 284-7763; Fax: (507) 284-0079; E-mail:
[email protected] This work was supported in part by a grant from Varian Medical Systems.
Received Nov 5, 2001, and in revised form Apr 15, 2002. Accepted for publication May 15, 2002.
584
Image quality of film and EPID
some EPID systems is inferior to that of portal film, a technical investigation by Munro et al. (26) showed that the contrast resolution of a fluoroscopic EPID is superior to that of portal film. Additionally, the improved spatial and contrast resolutions of next-generation active matrix EPIDs are expected to represent significant advances in image quality obtained with an EPID. Several observer studies subjectively compared EPID image quality to that of standard portal film, assessing the viability of replacing film with EPID for routine treatment review (27–29). These studies demonstrated that for most treatment sites, image quality was comparable between the two modalities, but that for several sites, such as head and neck or lateral pelvis, EPID image quality was inferior and did not represent a viable replacement for film. In contrast to the present investigation, all three of these studies compared film to hard copy printouts of EPIs, which do not allow the user to take advantage of the digital nature and large dynamic range of an electronic portal image (EPI). Most EPID systems offer a variety of image enhancement tools to tailor the display to the needs of the individual observer, as well as window and level adjustments to optimize the visibility of any anatomic feature. With the aid of image enhancement and quantitative analysis, EPIDs should provide an equivalent, if not superior, means of treatment verification. This study objectively and quantitatively addressed the issues of EPID image quality and clinical efficacy for two different commercial EPID systems, each separately compared to film. Image quality with these systems was measured as physicians rated the clarity of anatomic landmarks in both EPID and film images. For the purpose of this work, clinical efficacy was related to physicians’ accuracy in detecting and reporting quantitative setup errors in both film and EPID portals. METHODS AND MATERIALS Two different EPID systems were evaluated in this work. The first, a scanning liquid ion chamber (SLIC) (Varian PortalVision Mark II, Varian Medical Systems, Palo Alto, CA), was developed in the late 1980s (9) and has been commercially available for many years. The second system, an amorphous silicon (aSi) (Varian PortalVision aS500, Varian Medical Systems, Palo Alto, CA) active matrix flat panel imager (12), has only recently come to the market. The SLIC detector consists of an array of 256 ⫻ 256 square pixels, 1.27 mm on a side, whereas the aSi’s 384 ⫻ 512 square pixels are 0.78 mm in size. Both detectors are controlled by software (Varian PortalVision 6.0, Varian Medical Systems, Palo Alto, CA) that acquires, stores, and displays portal images. The detectors are mounted to the gantry by identical robotic arms that allow the imagers to deploy automatically to a range of source-to-detector distances, as well as to off-center displacements in either lateral or longitudinal directions. All images used for this study were acquired with the imager centered on the X-ray
● J. J. KRUSE et al.
585
beam’s central axis, at a source-to-detector distance of 140 cm. At this position, the SLIC detector covers an active area of 25 cm by 25 cm at isocenter, and the aSi imager covers an area 22 cm long by 29 cm wide at isocenter. All portal images used for this work were double exposure images that indicate the boundaries of the treatment portal and show additional anatomy beyond the treatment aperture. The first blocked field exposure was acquired during the actual treatment beam with the EPIDs. The treatment beam energy was either 6 or 18 MV and was delivered at 400 monitor units per minute (MU/min). After the treatment fraction, the field-shaping blocks were removed, secondary collimators were retracted, and another 4 MU of 6-MV beam were used to image anatomy inside and surrounding the field aperture. The EPID software detected the field boundary in the first exposure and displayed this contour on the second exposure. Both EPID systems require a fixed beam-on time to produce a portal image, so extrafractional open-field imaging was done at the lowest dose rate, 80 MU/min. The beam-on time for 4 MU at a low-dose rate enabled the SLIC system to raster across the detector once in “fast” acquisition mode and produce an image. In the same quantity of beam-on time, the aSi scanned eight individual frames and averaged them to produce a single portal image. All digital portal images in this study were viewed with the commercial acquisition and display software, which displayed them in an 8-bit grayscale on a 21-inch monitor. The software offers a variety of image enhancement and measurement tools, which users were encouraged to use in viewing EPIs. Each EPI used in this study was accompanied by a standard film portal of the same treatment field on the same patient. Film portals of treatment fields were acquired either the week before or the week after the electronic portal imaging session. The imaging sessions were separated by 1 week, because all images used were weekly verification images, as prescribed by clinical treatment protocols. No portal images were acquired, and no extrafractional dose was given to patients during this study. Film and EPI portals of phantoms and cadavers used in studies of review accuracy were acquired in immediate succession. The portal films used here were acquired with a metal screen and ready-pack film system (Kodak PP-L film, Eastman Kodak Co., Rochester, NY), which is the current standard for port films at our institution. Portal film double exposures were acquired either before or after the treatment fraction. Generally, the film was exposed by 2 MU of 6-MV X-rays with the field-shaping block in place. The block was removed, secondary collimators were opened, and an additional 2 MU were delivered to image anatomy surrounding the field aperture. An exception to this protocol was in imaging lateral pelvis fields, which required a total of 8 MUs to produce acceptable films. Image quality measurement The first phase of this study produced objective comparisons of image quality between film and the two EPID
586
I. J. Radiation Oncology
● Biology ● Physics
Table 1. List of critical anatomic landmarks for each treatment site studied Treatment site AP pelvis Pubic arch Pelvic brim Ischial tuberosity Symphysis Obturator fossae
Lateral pelvis Sacrum Coccyx Symphysis Femoral head
Abdomen
Volume 54, Number 2, 2002
clarity was calculated between film and its corresponding EPI for every review of the two portal images for each landmark. The differences between film and EPI were averaged over all reviews for each landmark.
Extremity
Vertebral column Long bone Lung Joint Tissue/air Ribs interface
systems. SLIC-EPID was compared to film with images of four treatment sites—anterior-posterior pelvis, lateral pelvis, abdomen, and extremity—whereas aSi-EPID was separately compared to film for anterior-posterior pelvis, lateral pelvis, and abdomen. A panel of six physicians indicated the prominent anatomic landmarks used in reviewing portals of each of the treatment sites. The landmarks for each treatment site are given in Table 1. In the SLIC-EPID film comparison, each physician reviewed 10 clinical EPIs for each of the four sites, with the exception of extremity, in which case only 9 images were reviewed. In the comparison between aSi-EPID and film, 12 aSi portals each for anteriorposterior pelvis, lateral pelvis, and abdomen were observed by the same reviewers. Using a method analogous to that in the study by Yin et al. (27), observers rated on a scale of 1 to 5, with 1 being the clearest, the visibility of each landmark in each image. Each observer independently rated the clarity of the landmarks in standard portal films of each of the same treatment fields. To foster uncorrelated reviews of EPID and film portals of the same fields, the EPID folder and film folder for each treatment site were viewed at different times. The portal films were viewed on a standard light box, whereas digital portal images were displayed using the EPID software package. The reviewers routinely used the window/ level controls and image enhancement filters when viewing EPIs, and some reviewers occasionally used the zoom and pan features as well. Seventy-five EPIs and 75 films were reviewed by six physicians, for a total of 900 portal image reviews. A potential limitation of EPIDs is their limited field of view compared to port films. These limitations were reflected in this comparison, because no attempt was made to correct the clarity comparisons for differing fields of view of the film and electronic images. If a given landmark was outside the field of view of a particular portal image, its clarity was rated as 5, or invisible, in that image. Therefore, landmarks that were routinely seen by one imaging modality but not its counterpart were rated as clearer, on average, in the modality with the larger field of view. The raw clarity scores for each landmark were averaged over all film reviews, and EPI reviews were averaged separately. To minimize physician-to-physician variability, i.e., differences in each physician’s rating scale, the difference in
Accuracy-of-review measurement Clinical efficacy of film and EPID systems was determined via physicians’ accuracy in detecting and reporting known setup errors in both film and EPI. In the SLIC-EPID– film comparison, portals used were film and EPIs of an anthropomorphic head-and-neck phantom. In the aSi-EPID, head-and-neck images of a human cadaver were used. The subject was placed on the simulator, and a lateral film was obtained. A field aperture was drawn on the simulation film, and a corresponding multileaf collimator file was created to achieve the prescribed blocking. The simulation image was also captured digitally by the simulator fluoroscopy display and stored in the EPID system to serve as a reference image for EPID reviews. The subject was then set up as prescribed on the treatment machine and imaged with both film and EPID. Next, a series of five additional portal image pairs of film and EPID were acquired with known setup errors. All five of the flawed portals contained translational errors ranging in magnitude from 2.5 mm to 13 mm. These errors were achieved by moving the indexed treatment couch, and several images contained compound translations, that is, errors in both the longitudinal and lateral directions. In addition to translational errors, one portal contained a rotational error that was achieved by rotating the phantom in the plane of the image by 3°. In another, the inferior jaw was moved by 1 cm, changing the shape of the field aperture. Figure 1 shows the simulation image of the head-and-neck phantom and an SLIC-EPID portal in which the phantom setup has been translated in the superior-inferior direction. These images were randomized, and the physicians reviewed each portal and reported, quantitatively, the type and magnitude of setup error. The film portals were reviewed alongside the simulation film on a light box. Only one port film was shown at a time, so the reviewers could refer only to the simulation film and not compare multiple port films. The EPIs were reviewed using the analysis tools of the EPID software. With this package, each EPI was displayed alongside the digitally captured simulation image. Before treatment review, a physicist contoured anatomic landmarks and the prescribed field aperture on the reference image. Total preparation time for the reference image in the EPID software was approximately 10 minutes. Landmarks contoured for portal analysis included the bones of the skull, frontal sinus, sella turcica, and cervical vertebrae. To review portal alignment in an EPI, physicians invoked an “anatomy match,” which overlaid the field aperture and anatomy contours on the portal image. The reviewers aligned the anatomy contours to features in the EPI, and the software reported setup deviations between prescribed and treated portals. Rotations were reported in terms of degrees, whereas translational deviations were given as centimeters at isocenter. Before EPI review was begun, physicians were
Image quality of film and EPID
● J. J. KRUSE et al.
587
Fig. 1. Simulation and portal image of the head-and-neck phantom. The image on the left is a digital fluoroscopy capture from the treatment simulator. Contoured on the image are prominent anatomic features and the prescribed treatment field aperture. The image on the right is an SLIC-EPID portal of the same phantom. The detected field aperture is displayed on the portal, as are the prescribed aperture and anatomy contours from the simulation image. The misalignment of detected and prescribed collimator leaf patterns indicates that in this case, the phantom is shifted in the superior-inferior direction.
presented a sample EPI and performed a match under the guidance of a physicist. Subsequent analysis of the EPIs for the study was performed by all physicians independently, using the matching software. The reviewers were not told what type of setup errors to expect in the images, but were told that some images could contain multiple errors, whereas others contained none. The score sheet for each image asked the reviewer to list the types and magnitudes of errors present. For the SLIC-EPID, four physicians reviewed six films and six EPID portals in this portion of the study, whereas six physicians reviewed six films and six EPID portals for the aSi-EPID. A total of 144 portal reviews were performed. The review-accuracy comparison between film and aSiEPID was performed in exactly the same manner as in the SLIC-EPID study, except that instead of a head-and-neck phantom, a human cadaver was used as the imaging subject. The field portal was a lateral head-and-neck treatment, and six physicians reviewed six films and six aSi-EPID portals. RESULTS Image quality measurement Figure 2a plots the clarity of anatomic landmarks in SLIC-EPID and film portals, as rated by our panel of physicians. The score for each landmark is averaged over all images of each modality, and over all reviewers. The error
bars on the points represent ⫾1 standard deviation of the data. The largest difference in image quality is seen in lateral pelvis images, where EPID was somewhat inferior to film for all landmarks, whereas anterior pelvis had the most favorable rating of EPID relative to film. Figure 2b plots the clarity of landmarks in aSi-EPID and film portals, measured and averaged as the SLIC-EPID data in Fig. 2a. All landmarks except lung in abdominal images were rated clearer in aSi-EPI than in film images. Of note in Fig. 2 are the sizable error bars on all data points. This broad range in landmark clarity scores has two dominant sources. The first is the wide variety in patient body habitus that makes some subjects easier to image than others. Second, there were average offsets between the physicians’ rating scales. To minimize physician-to-physician variability from the data, the difference in clarity for each landmark was calculated between film and EPID for every review of the two portal images. These differences were then averaged for each landmark, and the resulting scores no longer reflected the various offsets in the reviewers’ scoring scales. The differences were calculated such that a negative result demonstrated a preference for film, whereas a positive difference represented superior EPID performance. The results of these calculations for the SLIC-film comparison are completely summarized in Table 2, whereas the aSi-film comparison is summarized in Table 3. Each table gives mean
588
I. J. Radiation Oncology
● Biology ● Physics
Volume 54, Number 2, 2002
Fig. 2. (a) Average rated clarity of anatomic landmarks, averaged over all patients and reviewers. The data points give the average values for both film and EPID, and the error bars represent ⫾1 standard deviation of the data. (b) Average rated clarity of anatomic landmarks, averaged over all patients and reviewers. The data points give the average values for both film and aSi-EPID, and the error bars represent ⫾1 standard deviation of the data.
differences between film and EPID. Also listed are the standard deviations of the data for each landmark. Student’s paired t test was performed on the mean clarity difference data for all anatomic landmarks. In the third column of Tables 2 and 3 are the p values for the hypotheses that the mean difference between the film and EPID is zero. In cases for which statistically significant differences (p ⬍ 0.05) between the imaging media were observed, the fourth columns of Tables 2 and 3 indicate the preferred imaging medium for each landmark. Eleven of the 15 landmarks compared for film and SLICEPID were rated more visible in film, whereas 4 were clearer in EPID. However, none of the differences (⌬) between the film and SLIC-EPID were observed to be statistically significant (p ⬍ 0.05). Of the 12 landmarks compared in film and aSi-EPIs, only lung in abdominal images was rated more clearly visible in film, although the difference was not statistically significant (⌬ ⫽ ⫺0.65 p ⫽ 0.1). All 11 remaining landmarks were rated significantly clearer in aSi-EPIs than in film. The average difference in clarity between film and aSi-EPID for
these 11 landmarks is ⌬ ⫽ 1.37, and the p values for all of these comparisons are less than 0.05. Accuracy of review The reviewers’ average error in detecting and reporting translational shifts was 2.0 mm with film review and 1.5 mm with SLIC-EPID. In one portal, the phantom was rotated in the plane of the image by 3°. Each physician detected the rotational error with EPID and reported its magnitude to within a degree, whereas none reported it in the film. Finally, the field aperture was altered in one of the portals by moving the inferior jaw of the linac into the field 1 cm. One reviewer detected the altered field shape with film, and two noticed the error with the EPID system. In the comparison between film and aSi-EPID, the reviewers’ average error in detecting and reporting translational shifts was 3.0 mm with film review and 1.3 mm with aSi-EPID. In two of the portals, the head of the cadaver was rotated in the plane of the image by 4°. Five of six reviewers detected the rotational changes and reported their magnitudes to an average accuracy of 1.3° with aSi-EPID. Two of
Image quality of film and EPID
● J. J. KRUSE et al.
589
Table 2. Average difference in clarity of landmarks between film and SLIC-EPID Ave. Diff. Std. Dev. p value Preference AP pelvis Pubic arch Pelvic brim Symphysis Ischial tuberosity Obturator fossae Lateral pelvis Sacrum Coccyx Symphysis Femoral head Abdomen Vertebrae Lung Ribs Extremity Long bone Joint Tissue/air interface
⫺0.27 0.25 ⫺0.40 0.15 0.10
1.46 0.91 1.09 1.24 1.03
0.578 0.409 0.277 0.712 0.762
None None None None None
⫺0.28 ⫺0.70 ⫺0.48 ⫺0.62
1.04 1.28 1.27 1.06
0.413 0.119 0.259 0.099
None None None None
⫺0.27 0.23 ⫺0.08
0.99 1.02 0.98
0.416 0.494 0.805
None None None
⫺0.07 ⫺0.28 ⫺0.33
0.87 1.18 1.26
0.804 0.493 0.450
None None None
Notes: The clarity of each of these landmarks was rated from 1 to 5 in film and EPID images. A negative difference indicates that the film was clearer, whereas a positive difference indicates a preference for EPID. p values are given for the hypothesis that the average landmark clarity ratings in film and EPID are equal. In the case of statistically significant differences, the preferred medium is indicated in column four.
⫺0.27, p ⫽ 0.416). The improved clarity of vertebrae in this study may be because reviewers in this work observed EPIs on-line, using image enhancement tools, Table 3. Average difference in clarity of landmarks between film and aSi-EPID Ave. Diff. Std. Dev. p value Preference
Fig. 2. (Cont’d)
the six reviewers detected a rotational error in the correct film portals, and another two reviewers reported rotational errors in films that contained none. Finally, one of the film-EPID portal pairs was treated with a modified multileaf collimator aperture, simulating incorrect leaf positions. One physician noted the error in the film review, whereas three of six detected the problem in EPID review. DISCUSSION The observer comparison between film and EPID performed by Yin et al. found that reviewers had a much harder time spotting vertebrae in EPIs than in films of abdominal treatments (27), whereas this study finds no significant difference between film and SLIC-EPID (⌬ ⫽
AP pelvis Pubic arch Pelvic brim Symphysis Ischial tuberosity Obturator fossae Lateral pelvis Sacrum Coccyx Symphysis Femoral head Abdomen Vertebrae Lung Ribs
1.53 1.06 1.36 1.83 2.01
1.16 0.87 1.17 1.26 1.13
⬍0.001 0.001 0.002 ⬍0.001 ⬍0.001
EPID EPID EPID EPID EPID
0.78 1.19 1.24 1.32
1.15 1.41 1.09 0.93
0.039 0.014 0.002 ⬍0.001
EPID EPID EPID EPID
1.88 ⫺0.65 0.83
0.99 1.26 1.23
⬍0.001 0.100 0.039
EPID None EPID
Notes: The clarity of each of these landmarks was rated from 1 to 5 in film and EPID images. A negative difference indicates that the film was clearer, whereas a positive difference indicates a preference for EPID. p values are given for the hypothesis that the average landmark clarity ratings in film and EPID are equal. In the case of statistically significant differences, the preferred medium is indicated in column four.
590
I. J. Radiation Oncology
● Biology ● Physics
instead of on hard copy printouts of EPIs, as in the work by Yin et al. (27). Window/level settings that optimize visibility of vertebrae in an abdominal port often saturate the rest of the image. If users are forced to choose a single window/level setting before printing an image, they will probably not choose a display in which most of the port is saturated, even if it does show the vertebral bodies. Presenting a hard copy of an EPID portal image to a physician for treatment review is certainly the easiest way to initiate an electronic portal imaging program. However, this study suggests that the time involved in training physicians to review images on-line and optimize display for various anatomic landmarks will translate into enhanced effectiveness of the system. The single landmark that was rated as more visible in film than in aSi-EPI was the lung in abdominal images. As discussed at the outset, these comparisons did not correct for larger field of view of film relative to EPI. Certainly the higher average score of lung clarity in film portals is because of a number of treatments in which the diaphragm is on the periphery of a portal film, but beyond the field of view of the aSi-EPID. Although the SLIC-EPID subtends a slightly smaller total field of view than aSi (625 cm2 vs. 638 cm2), its longitudinal extent is greater than that of the aSi detector (25 cm vs. 22 cm) at the imaging position used here. The greater longitudinal extent of the SLIC device, combined with a more favorable sampling of treatment portals, probably led to the result that SLIC-EPID was comparable to film for imaging lung (⌬ ⫽ 0.23, p ⫽ 0.494), whereas the aSi-EPID was somewhat less effective than film (⌬ ⫽ ⫺0.65, p ⫽ 0.100). Both EPID systems were compared to Kodak PP-L ready-pack film in the image quality studies. Commercially available portal films such as Kodak ECL have been shown to exhibit superior contrast resolution to the ready-pack film, and could be expected to compare more favorably to EPID systems. However, because of issues associated with the reduced latitude of ECL film and the convenience of the ready-pack format of PP-L film, ECL has not replaced PP-L as our standard clinical film verification method. Because the goal of the present work is to evaluate EPIDs’ effectiveness as clinical tools, they were compared against PP-L film, our current standard.
Volume 54, Number 2, 2002
CONCLUSIONS This study objectively compared two commercial EPID systems with portal film during clinical treatment field portal imaging using quantitative measures of landmark visibility and treatment review accuracy. The rated clarity of anatomic landmarks varied widely from patient to patient in both film and EPID portals. The average difference in clarity between film and SLIC-EPID was calculated for each landmark, and no statistically significant (p ⬍ 0.05) differences were found for any of the 15 landmarks. This result suggests that image quality alone is not a significant barrier to implementation of routine electronic portal imaging with an SLIC-EPID. In contrast, 11 of 12 anatomic landmarks were clearer (average ⌬ ⫽ 1.37) in aSi-EPIs than in films of the same treatment portals; 8 of 11 p values were less than 0.002. The only landmark found to be more visible in films than in aSi-EPIs was the lung, in abdominal treatments. In this case, the limited spatial extent of the aSi detector impaired its performance relative to film, although the ability to move the detector longitudinally could be exploited to eliminate this problem. For all other landmarks compared here, the aSi-EPID represents a significant advancement in image quality. Finally, the reviewers were asked to detect and quantify setup errors in portal images. We found that users detected and quantified treatment portal errors more accurately with both commercial EPID systems than they did with film. Port films have long been the accepted means of assessing and documenting geographic accuracy of patient treatments. Replacing this system with an EPID represents a significant shift in the way all clinical personnel—therapists, physicists, and physicians—perform their roles. The transition to EPID is not without pitfalls, but this work suggests that image quality need not be one of them, even with older, less advanced imaging systems. Conversely, the superior image quality afforded by a new-generation aSi imager is no substitute for careful planning in implementing the device with success. Successful implementation of EPID requires the development of specific goals and protocols, as well as thorough training for all users (30). Commitment to a careful transition can pay off in quicker, more frequent portal images, digital storage and enhancement, and, as demonstrated here, more accurate portal review, without increasing work load.
REFERENCES 1. Marks JE, Haus AG, Sutton HG, et al. The value of frequent treatment verification films in reducing localization error in the irradiation of complex fields. Cancer 1976;37:2755–2761. 2. Goitein M, Busse J. Immobilization error: Some theoretical considerations. Radiology 1975;117:407– 412. 3. Killoran JH, Kooy HM, Gladstone DJ, et al. A numerical simulation of organ motion and daily setup uncertainties: Implications for radiation therapy. Int J Radiat Oncol Biol Phys 1997;37:213–221. 4. Byhardt RW, Cox JD, Hornburg A, et al. Weekly localization
5.
6. 7. 8.
films and detection of field placement errors. Int J Radiat Oncol Biol Phys 1978;4:881– 887. Meertens H, van Herk M, Bijhold J, et al. First clinical experience with a newly developed electronic portal imaging device. Int J Radiat Oncol Biol Phys 1990;18:1173–1181. Boyer AL, Antonuk L, Fenster A, et al. A review of electronic portal imaging devices (epids). Med Phys 1992;19:1–16. Leong J. Use of digital fluoroscopy as an on-line verification device in radiation therapy. Phys Med Biol 1986;31:985–992. Munro P, Rawlinson JA, Fenster A. A digital fluoroscopic
Image quality of film and EPID
9. 10. 11. 12. 13. 14. 15.
16.
17.
18.
19.
imaging device for radiotherapy localization. Int J Radiat Oncol Biol Phys 1990;18:641– 649. van Herk M, Meertens H. A matrix ionisation chamber imaging device for on-line patient setup verification during radiotherapy. Radiother Oncol 1988;11:369 –378. Visser AG, Huizenga H, Althof VG, et al. Performance of a prototype fluoroscopic radiotherapy imaging system. Int J Radiat Oncol Biol Phys 1990;18:43–50. Wong JW, Cheng AY, Binns WR, et al. Development of a second-generation fiber-optic on-line image verification system. Int J Radiat Oncol Biol Phys 1993;26:311–320. Antonuk LE, Yorkston J, Huang W, et al. Megavoltage imaging with a large-area, flat-panel, amorphous silicon imager. Int J Radiat Oncol Biol Phys 1996;36:661– 672. Falco T, Wang H, Fallone BG. Preliminary study of a metal/ a-se-based portal detector. Med Phys 1998;25:814 – 823. Michalski JM, Graham MV, Bosch WR, et al. Prospective clinical evaluation of an electronic portal imaging device. Int J Radiat Oncol Biol Phys 1996;34:943–951. Valicenti RK, Michalski JM, Bosch WR, et al. Is weekly port filming adequate for verifying patient position in modern radiation therapy? Int J Radiat Oncol Biol Phys 1994;30:431– 438. Herman MG, Khadivi KO, Gage I, et al. Effects of respiration on target and critical structure positions during treatment assessed with movie-loop electronic portal imaging. Int J Radiat Oncol Biol Phys 1997;39(2S):163. Fein DA, McGee KP, Schultheiss TE, et al. Intra- and interfractional reproducibility of tangential breast fields. A prospective on-line portal imaging study. Int J Radiat Oncol Biol Phys 1996;34:733–740. Creutzberg CL, Althof VG, Huizenga H, et al. Quality assurance using portal imaging: The accuracy of patient positioning in irradiation of breast cancer. Int J Radiat Oncol Biol Phys 1993;25:529 –539. Herman MG, Abrams RA, Mayer RR. Clinical use of on-line
● J. J. KRUSE et al.
20.
21. 22. 23. 24. 25. 26. 27. 28.
29. 30.
591
portal imaging for daily patient treatment verification. Int J Radiat Oncol Biol Phys 1994;28:1017–1023. Ezz A, Munro P, Porter AT, et al. Daily monitoring and correction of radiation field placement using a video-based portal imaging system: A pilot study. Int J Radiat Oncol Biol Phys 1992;22:159 –165. De Neve W, Van den Heuvel F, Coghe M, et al. Interactive use of on-line portal imaging in pelvic radiation. Int J Radiat Oncol Biol Phys 1993;25:517–524. Balter JM, Chen GT, Pelizzari CA, et al. Online repositioning during treatment of the prostate: A study of potential limits and gains. Int J Radiat Oncol Biol Phys 1993;27:137–143. Bel A, van Herk M, Bartelink H, et al. A verification procedure to improve patient set-up accuracy using portal images. Radiother Oncol 1993;29:253–260. Bissett R, Boyko S, Leszczynski K, et al. Radiotherapy portal verification: An observer study. Br J Radiol 1995;68:165–174. Perera T, Moseley J, Munro P. Subjectivity in interpretation of portal films. Int J Radiat Oncol Biol Phys 1999;45:529 –534. Munro P, Rawlinson JA, Fenster A. Therapy imaging: A signal-to-noise analysis of a fluoroscopic imaging system for radiotherapy localization. Med Phys 1990;17:763–772. Yin FF, Rubin P, Schell MC, et al. An observer study for direct comparison of clinical efficacy of electronic to film portal images. Int J Radiat Oncol Biol Phys 1996;35:985–991. Hatherly K, Smylie J, Rodger A. A comparison of field-only electronic portal imaging hard copies with double exposure port films in radiation therapy treatment setup confirmation to determine its clinical application in a radiotherapy center. Int J Radiat Oncol Biol Phys 1999;45:791–796. Lavertu S, Girouard LM, Pouliot J. Observation study of electronic portal images for off-line verification. Radiother Oncol 2000;54:47–55. Herman MG, Kruse JJ, Hagness C. A guide to clinical implementation of electronic portal imaging. J Appl Clin Med Phys 2000: March.
