- Email: [email protected]
An Assessment of a Film Enhancement System for Use in a Radiation Therapy Department
Copynghr
0
1990 Amencan
AN ASSESSMENT OF A FILM ENHANCEMENT SYSTEM IN A RADIATION THERAPY DEPARTMENT
0739-021 l/90 Asscaal~on of Medxal
$3.00 + .lM Dosimelrists
FOR USE
ELLENL. SOLOWSKY, B.S.,C.M.D.,LAWRENCEE.REINSTEIN,PH.D., ~~~ALLENG.MEEK,M.D. Department of Radiation Oncology, University Hospital, State University of New York at Stony Brook, Stony Brook, New York 11794-7028, U.S.A. Abstract-The clinical uses of a radiotherapy film enhancement system are explored. The primary functions of the system are to improve the quality of poorly exposed simulator and portal films, and to perform comparisons between the two films to determine whether patient or block positioning errors are present. Other features include: the production of inexpensive, high quality hardcopy images of simulation films and initial portal films for chart documentation, the capacity to overlay lateral simulation films with sag&al MRI films to aid in field design, and a mode to zoom in on individual Cf or MRI images and enlarge them for video display during chart rounds or instructional sessions. This commercially available system is comprised of a microcomputer, frame grabber, CCD camera with zoom lens, and a high-resolution thermal printer. The user-friendly software is menu driven and utilizes both keyboard and track ball to perform its functions. At the heart of the software is a very fast Adaptive Histogram Equalization (AHE) routine, which enhances and improves the readability of most portal films. The system has been evaluated for several disease sites, and its advantages and limitations will be presented. Key Words: Video imaging, Portal films, Positioning.
INTRODUCTION
One of the primary functions of the system is to improve the readability of poor quality simulator and portal films by using different image enhancement routines. Once the films are enhanced, portal films are compared to corresponding simulator films to determine whether errors in patient positioning or block cutting are present. Several other uses of the RFES include the production of hardcopy images of simulator and portal films for documentation, the ability to overlay sagittal MRI images onto lateral simulator films to aid in field design, and the large screen display of individual CT or MRI images for video display.
Portal imaging is an important part of the quality assurance of patient treatment in radiation therapy. It is generally assumed that missing even a small portion of the target volume for a few treatments may greatly reduce the chance of controlling the disease.im3 To determine whether the given treatment is the same as the treatment intended by the physician, the portal film must be compared to the simulator film. If differences do exist, it is then possible to isolate the cause of these discrepancies, and take appropriate actions (such as recutting a block or adding an immobilization device). Portal films themselves are often difficult to interpret because of their poor resolution and contrast. Visual comparison between simulator and portal films is complicated by scaling differences, making it almost impossible to overlay or superimpose them. As a result, it is often difficult to determine the source of a particular discrepancy. Another problem we have encountered is the time wasted in locating films, especially in our busy department. Simulator or initial portal films may be needed by others, such as mold room or dosimetry staff, and, therefore, not available to technologists when the weekly portal is taken. We believe most of these problems can be overcome through use of a digital Radiotherapy Film Enhancement System (RFES), which has recently been acquired for testing and evaluation.*
SYSTEM DESIGN Hardware The RFES (Fig. 1) is comprised of an IBM-AT compatible microcomputer, a video digitizer board, a solid state, high resolution video camera with zoom lens, and a high resolution thermal printer. The computer has a 40 megabyte hard disk and a S&inch, highdensity, floppy drive for flexible storage capabilities. There are two display units: a 13-inch RGB monitor for image display and a monochrome screen for control menu display. The RFES also contains two builtin illumination panels for film viewing and imaging. The thermal printer provides very high quality hardcopy images, 4 inches wide by 3 inches long. All components are housed in a specially designed cabinet that can be easily moved if needed. Software The RFES software is user friendly and mostly self-explanatory. It is menu driven and functions are
* Atomic Products Corporation. Shirley. New York. 107
Medical Dosimetry
108
Fig. I. Overview of the system. On the left is the illumination panel for viewing, with the imaging panel below. Along the top row is the computer and thermal printer (left to right). Continuing right to left on the next row is the menu screen and video display screen, with the keyboard and track-ball below them.
controlled by either the keyboard or the track-ball. Film digitizing is done in a single keystroke, because the range of densities on any film is automatically handled by software control of the “gain” and “offset” of the digitizer board. Enhancements are performed with one or two additional keystrokes. Image comparisons require more user interaction through the use of a moving cursor controlled by either the keyboard or the track-ball.
Volume 15. Number 3. 1990
using the f-stop of the camera and the focus is adjusted. The RFES is equipped with several enhancement routines, including Histogram Equalization (HE). Edge Enhancement, Adaptive Histogram Equalization (AHE), and Smoothing. Two forms of the AHE algorithm are available, yielding different levels of enhancement. We have found that the AHE algorithm used in conjunction with the Smooth algorithm is ideal for this application. These routines are very fast and the entire procedure can be done in about a minute. Once the film is enhanced and smoothed, a hardcopy is printed by pushing a button on the thermal printer. (Whatever is on the video display screen is dumped to the printer.) These printouts are glued into the patient chart on a special page, and labeled with the patient’s name, field description, field number, and date. They are used as a quick reference for technologists, dosimetrists, physicists, and physicians to aid in the review of the treatment site and general field arrangement. It is especially useful during chart review and helps dosimetrists to visualize the extent of blocking in a particular field to determine whether special equivalent square or other special calculations are warranted. In Fig. 2, the original simulation film is shown as viewed on the light box (a), and the enhanced image is shown as viewed on the display screen (b).
Portal film enhancement CLINICAL
APPLICATIONS
Simulator film enhancement Although the quality of simulator films in general is acceptable, it is surprising how much these images can be improved by using enhancement techniques. Also, our film images are further degraded when copied for placement in the treatment chart. (Good copies of the simulator and portal films are also important for transmittal of patient data for cooperative group studies, such as RTOG.) In the past, we made these copies by taking Polaroid pictures of the simulation films and stapling them into the chart to document the treatment site. The Polaroids were usually of very poor quality, especially if the quality of the original simulation film was poor. The RFES improves the readability of the simulator film, and provides a high quality printout of the enhanced image for placement into the treatment chart. The procedure for enhancing films is simple. The film is placed on the illumination panel directly beneath the camera. The control menu indicates that the image is LIVE, and the film is centered by using video image display as a guide. By controlling the zoom lens, the area of interest on the film is magnified to fill the display. The optimal brightness is selected
We use the same procedure for the initial (approved) first day portal films as for the simulator films. The initial portal is enhanced, printed, and placed into the chart directly beside the corresponding simulator film. In Fig. 3 the simulator image (a) is shown along side the original portal as viewed on the light box (b) in comparison to the enhanced image (c) obtained from the same film. Figure 4 is an example of a page in the chart, complete with simulator and portal image prints.
Portal-simulatorjilm comparison Perhaps the most important function of the system is the comparison of the simulator image to the portal image, in order to determine whether the field shape and patient position are as planned by the physician. To perform the comparison it is first necessary to digitize the simulator and corresponding portal films, or if previously digitized, retrieve them from storage on the hard disk or floppies. Because the films may have been entered into the RFES with different scale factors and orientation, it is necessary for the software to align them through a “transformation” routine. The user has a choice of using either anatomical landmarks or the corners of the radiation field for three “transformation points”
109
Use of a film enhancement system 0 E. L. SOLOWSKY c/ a/. sen as the three field corners,
shown as dots with a circle around them (Fig. 5). (These are actually in color on the screen and are easily visible.)
a
a
b
Fig. 2(a). Lung simulator film as viewed on the light box. (b). Lung simulator film as viewed on the image screen after
enhancing and smoothing.
b
required by the software. (If a “graticule” is used during simulation and filming on the treatment machine, any corresponding points on the graticule may be used.) The three points are chosen first on one film, then the other, and are entered in the same sequence to obtain the correct translation, rotation, and scaling. Clinical example The following is a description of a case where review of the initial portal film seemed to suggest that the blocks were incorrectly cut. However, follow-up study using the RFES demonstrated that in fact, it was the patient set-up and not the block shape that was in error. For this case the process of film comparison began with the selection of transformation points cho-
Fig. 3(a). Lateral pelvic simulator film image from screen. (b). Corresponding lateral pelvic portal film as viewed on the light box. (c). Same lateral portal after enhancing and smoothing.
Medical Dosimetry
Volume 15, Number 3, 1990
Fig. 6. The same para-aortic setup, with the vertebral body outlined confirming that the error was in patient positioning rather than block cutting.
Fig. 4. A typical chart page for a three-field head and neck setup. Once the points were entered, using either the keyboard or the track-ball to move the cursor about the screen, the image on the right side of the video display (simulator image) was automatically aligned by the software to match the image on the left (portal image). After image alignment, any outline drawn on one image can be transferred to the other image, in order to check the position of a block edge or an anatomical landmark. (Note that in this case, the physician indicated in black on the portal film that the block needed to be recut because it appeared to be insufficient (Fig. 5).) Using the track ball, the outline of the block shadow was traced onto the portal image and then overlaid onto the simulator image. (These lines are white in the illustration, but are shown in color on the screen for easy viewing.) Notice how closely the white line follows the black outline of the block as designed
Fig. 5. Posterior para-aortic setup, simulator film image on the left and portal film image on the right. The black line on the portal is an indication made by the physician to recut the block. The white line is the block shadow as outlined on the portal image and transferred to the simulator image using the compare routine. This block was cut correctly.
on the simulator film. In retrospect, this indicates that, in fact, the block was cut correctly, and it was the patient position that was in error. This conclusion was confirmed by further investigation using the patient’s anatomical landmarks. The same overlay procedure was used (Fig. 6), this time outlining one of the T-spine vertebral bodies. Once the vertebral body outline was transferred from the portal image on the right to the simulator image on the left, it became apparent that the patient was not set up accurately. (Note the discrepancy of about 1 cm on the position of this vertebra.) Figure 7 illustrates the corrected field, with the block outline matching. The lower left image is an overlay of the simulator and portal images. Note that the block shadow from the portal film is matching the black outline of the block drawn on the simulator film. Figure 8 confirms the correct patient position, again by using one of the vertebral bodies,
A4RI assistedfield design Another important application of the RFES is the incorporation of MRI data into the design of the treatment fields. By scaling and aligning the MRI
Fig. 7. The same para-aortic setup with the corrected portal film. Notice the overlay of the two images in the lower left corner. where the block shadow from the portal matches the block outline in black on the simulator image.
Use of a film enhancement
system 0 E. L.
!~LOWSKY
CI a/.
111
courages participation from those seated in the room. The size comparison is illustrated in Fig. 10. CONCLUSIONS
Fig. 8. The same setup now showing the corrected patient position as indicated by the outline of the vertebral body.
image to the simulator image, it allows the radiotherapist to check the block design for cone-downs to the tumor volume. In the example shown (Fig. 9), the two images were aligned using three anatomical landmarks. As can be seen, it was difficult in this case to identify three points in an area with no sharp intersections of bones or well defined comers. (It may be desirable to use “MRI opaque” fiducial markers for future “planning” MRI scans. The accuracy of this technique is also affected by the fact that the simulator film is a projected radiograph, while the MRI image is a single slice.) The tumor volume as seen on the MRI was outlined with the track-ball and transferred to the lateral simulation film. The overlay of the MRI tumor volume on the simulator image was then used to help guide the radiotherapist in designing field shaping blocks for the cone-down fields. Use of RFES during treatment planning conferences CT scan films and MRI films are usually printed in a format containing nine images per page of film, each image measuring approximately 3.5 inches square. Although this size is suitable for viewing by one or two people standing close to the film, in a conference or chart rounds, it is not easy for everyone in the room to see what is being discussed. By connecting the RFES to a large screen monitor (17 inches wide by 12 inches high), we have provided a very readable full screen display of the CT or MRI image for use during conferences or teaching sessions. This is a definite advantage over the small film panels, and en-
Fig. 9. Example of a lateral whole brain simulator film image with a corresponding sagittal MRI film image. The tumor volume is outlined on the MRI, and transferred to the simulator image.
The Radiotherapy Film Enhancement System (RFES) has been a worthwile addition to our department. Its ability to provide high quality simulator and portal image printouts for placement in the patient chart and the comparison portions of the program are the two functions we have found to be most promising. The enhancing routines are very fast (approximately 30 seconds), and the software is easy to run with a minimum of instruction. Acceptance of the system within the clinic has been steadily growing. The simulator technologists are already comfortable with its use, and are quite dependent on it for improving and copying their films. The reaction of the treatment technologists to having ready access to the simulator and portal images in the chart has been mixed. Some problems exist with the actual attachment of the printed images to the chart pages, in that they tend to be dislodged over time through normal wear and tear. (Special selfadhesive mounting sheets are under development, which should eliminate this problem.) The dosimetrists, however, rely on the images in the chart to aid in performance of a more comprehensive weekly chart review. While we have already demonstrated the usefulness of the RFES comparison routines for determining whether an apparent block cutting error requires a new block to be cut or simply the repositioning of the patient, it has thus far only been used occasionally for this purpose. This is because the comparison routines require more time and user interaction to perform. It is expected that its use will grow as the staff becomes more familiar with its operation. Several suggestions for improving the RFES and the ease of user interaction are:
Fig. 10. Photograph of the large screen video display, with the same CT film for size comparison.
112
Medical Dosimetry
The three point entry needed for image alignment should be augmented by a routine that allows the entry of curved contour lines as well (such as a segment of the lung/soft tissue interface or rib outline). It is often difficult to find three distinct pairs of anatomical points on the films, and such an improvement would make the comparison procedure easier. ?? The RFES should have the option to generate larger format printouts, which can be used as templates for block cutting. 0 A data base capability should be provided so that images can be stored and retrieved by indexing according to physician or patient name. (Currently, stored images are identified by numbers only.) This would allow a radiotherapist to retrieve all his stored portal images for review at the end of the day. ??
There are many features of the system we have not yet explored, such as the measurement of errors in
Volume IS, Number 3. 1990
block or patient position and the other enhancement algorithms. We hope to investigate these as well in the near future.
Acknowledgements-Special thanks to Atomic Products Corporation for supplying the system and their support. and to the Staff of the Department of Radiation Oncology at Stony Brook for their assistance and cooperation during this study.
REFERENCES 1. Marks, J.E.; Bedwinek, J.M.; Lee. F.; Purdy, J.A.; Perez. CA. Dose response analysis in nasopharyngeal carcinoma. Cancer 50:1042; 1982. 2. Goitein, M.; Busse, J. Immobilization error: some theoretical considerations. Radiology 117:407-4 12; 1975. 3. Byhardt, R.W.; Cox, J.D.; Homburg, A.; Liermann, G. Weekly localization films and detection of field placement errors. Int. J. Radiat. Oncol. Biol. Phys. 4~88l-887: 1978. 4. Leszczynski, K.W.; Shalev, S. A robust algorithm for contrast enhancement by local histogram modification. Image And Vision Computing 7:205-209; 1989.
0
1990 Amencan
AN ASSESSMENT OF A FILM ENHANCEMENT SYSTEM IN A RADIATION THERAPY DEPARTMENT
0739-021 l/90 Asscaal~on of Medxal
$3.00 + .lM Dosimelrists
FOR USE
ELLENL. SOLOWSKY, B.S.,C.M.D.,LAWRENCEE.REINSTEIN,PH.D., ~~~ALLENG.MEEK,M.D. Department of Radiation Oncology, University Hospital, State University of New York at Stony Brook, Stony Brook, New York 11794-7028, U.S.A. Abstract-The clinical uses of a radiotherapy film enhancement system are explored. The primary functions of the system are to improve the quality of poorly exposed simulator and portal films, and to perform comparisons between the two films to determine whether patient or block positioning errors are present. Other features include: the production of inexpensive, high quality hardcopy images of simulation films and initial portal films for chart documentation, the capacity to overlay lateral simulation films with sag&al MRI films to aid in field design, and a mode to zoom in on individual Cf or MRI images and enlarge them for video display during chart rounds or instructional sessions. This commercially available system is comprised of a microcomputer, frame grabber, CCD camera with zoom lens, and a high-resolution thermal printer. The user-friendly software is menu driven and utilizes both keyboard and track ball to perform its functions. At the heart of the software is a very fast Adaptive Histogram Equalization (AHE) routine, which enhances and improves the readability of most portal films. The system has been evaluated for several disease sites, and its advantages and limitations will be presented. Key Words: Video imaging, Portal films, Positioning.
INTRODUCTION
One of the primary functions of the system is to improve the readability of poor quality simulator and portal films by using different image enhancement routines. Once the films are enhanced, portal films are compared to corresponding simulator films to determine whether errors in patient positioning or block cutting are present. Several other uses of the RFES include the production of hardcopy images of simulator and portal films for documentation, the ability to overlay sagittal MRI images onto lateral simulator films to aid in field design, and the large screen display of individual CT or MRI images for video display.
Portal imaging is an important part of the quality assurance of patient treatment in radiation therapy. It is generally assumed that missing even a small portion of the target volume for a few treatments may greatly reduce the chance of controlling the disease.im3 To determine whether the given treatment is the same as the treatment intended by the physician, the portal film must be compared to the simulator film. If differences do exist, it is then possible to isolate the cause of these discrepancies, and take appropriate actions (such as recutting a block or adding an immobilization device). Portal films themselves are often difficult to interpret because of their poor resolution and contrast. Visual comparison between simulator and portal films is complicated by scaling differences, making it almost impossible to overlay or superimpose them. As a result, it is often difficult to determine the source of a particular discrepancy. Another problem we have encountered is the time wasted in locating films, especially in our busy department. Simulator or initial portal films may be needed by others, such as mold room or dosimetry staff, and, therefore, not available to technologists when the weekly portal is taken. We believe most of these problems can be overcome through use of a digital Radiotherapy Film Enhancement System (RFES), which has recently been acquired for testing and evaluation.*
SYSTEM DESIGN Hardware The RFES (Fig. 1) is comprised of an IBM-AT compatible microcomputer, a video digitizer board, a solid state, high resolution video camera with zoom lens, and a high resolution thermal printer. The computer has a 40 megabyte hard disk and a S&inch, highdensity, floppy drive for flexible storage capabilities. There are two display units: a 13-inch RGB monitor for image display and a monochrome screen for control menu display. The RFES also contains two builtin illumination panels for film viewing and imaging. The thermal printer provides very high quality hardcopy images, 4 inches wide by 3 inches long. All components are housed in a specially designed cabinet that can be easily moved if needed. Software The RFES software is user friendly and mostly self-explanatory. It is menu driven and functions are
* Atomic Products Corporation. Shirley. New York. 107
Medical Dosimetry
108
Fig. I. Overview of the system. On the left is the illumination panel for viewing, with the imaging panel below. Along the top row is the computer and thermal printer (left to right). Continuing right to left on the next row is the menu screen and video display screen, with the keyboard and track-ball below them.
controlled by either the keyboard or the track-ball. Film digitizing is done in a single keystroke, because the range of densities on any film is automatically handled by software control of the “gain” and “offset” of the digitizer board. Enhancements are performed with one or two additional keystrokes. Image comparisons require more user interaction through the use of a moving cursor controlled by either the keyboard or the track-ball.
Volume 15. Number 3. 1990
using the f-stop of the camera and the focus is adjusted. The RFES is equipped with several enhancement routines, including Histogram Equalization (HE). Edge Enhancement, Adaptive Histogram Equalization (AHE), and Smoothing. Two forms of the AHE algorithm are available, yielding different levels of enhancement. We have found that the AHE algorithm used in conjunction with the Smooth algorithm is ideal for this application. These routines are very fast and the entire procedure can be done in about a minute. Once the film is enhanced and smoothed, a hardcopy is printed by pushing a button on the thermal printer. (Whatever is on the video display screen is dumped to the printer.) These printouts are glued into the patient chart on a special page, and labeled with the patient’s name, field description, field number, and date. They are used as a quick reference for technologists, dosimetrists, physicists, and physicians to aid in the review of the treatment site and general field arrangement. It is especially useful during chart review and helps dosimetrists to visualize the extent of blocking in a particular field to determine whether special equivalent square or other special calculations are warranted. In Fig. 2, the original simulation film is shown as viewed on the light box (a), and the enhanced image is shown as viewed on the display screen (b).
Portal film enhancement CLINICAL
APPLICATIONS
Simulator film enhancement Although the quality of simulator films in general is acceptable, it is surprising how much these images can be improved by using enhancement techniques. Also, our film images are further degraded when copied for placement in the treatment chart. (Good copies of the simulator and portal films are also important for transmittal of patient data for cooperative group studies, such as RTOG.) In the past, we made these copies by taking Polaroid pictures of the simulation films and stapling them into the chart to document the treatment site. The Polaroids were usually of very poor quality, especially if the quality of the original simulation film was poor. The RFES improves the readability of the simulator film, and provides a high quality printout of the enhanced image for placement into the treatment chart. The procedure for enhancing films is simple. The film is placed on the illumination panel directly beneath the camera. The control menu indicates that the image is LIVE, and the film is centered by using video image display as a guide. By controlling the zoom lens, the area of interest on the film is magnified to fill the display. The optimal brightness is selected
We use the same procedure for the initial (approved) first day portal films as for the simulator films. The initial portal is enhanced, printed, and placed into the chart directly beside the corresponding simulator film. In Fig. 3 the simulator image (a) is shown along side the original portal as viewed on the light box (b) in comparison to the enhanced image (c) obtained from the same film. Figure 4 is an example of a page in the chart, complete with simulator and portal image prints.
Portal-simulatorjilm comparison Perhaps the most important function of the system is the comparison of the simulator image to the portal image, in order to determine whether the field shape and patient position are as planned by the physician. To perform the comparison it is first necessary to digitize the simulator and corresponding portal films, or if previously digitized, retrieve them from storage on the hard disk or floppies. Because the films may have been entered into the RFES with different scale factors and orientation, it is necessary for the software to align them through a “transformation” routine. The user has a choice of using either anatomical landmarks or the corners of the radiation field for three “transformation points”
109
Use of a film enhancement system 0 E. L. SOLOWSKY c/ a/. sen as the three field corners,
shown as dots with a circle around them (Fig. 5). (These are actually in color on the screen and are easily visible.)
a
a
b
Fig. 2(a). Lung simulator film as viewed on the light box. (b). Lung simulator film as viewed on the image screen after
enhancing and smoothing.
b
required by the software. (If a “graticule” is used during simulation and filming on the treatment machine, any corresponding points on the graticule may be used.) The three points are chosen first on one film, then the other, and are entered in the same sequence to obtain the correct translation, rotation, and scaling. Clinical example The following is a description of a case where review of the initial portal film seemed to suggest that the blocks were incorrectly cut. However, follow-up study using the RFES demonstrated that in fact, it was the patient set-up and not the block shape that was in error. For this case the process of film comparison began with the selection of transformation points cho-
Fig. 3(a). Lateral pelvic simulator film image from screen. (b). Corresponding lateral pelvic portal film as viewed on the light box. (c). Same lateral portal after enhancing and smoothing.
Medical Dosimetry
Volume 15, Number 3, 1990
Fig. 6. The same para-aortic setup, with the vertebral body outlined confirming that the error was in patient positioning rather than block cutting.
Fig. 4. A typical chart page for a three-field head and neck setup. Once the points were entered, using either the keyboard or the track-ball to move the cursor about the screen, the image on the right side of the video display (simulator image) was automatically aligned by the software to match the image on the left (portal image). After image alignment, any outline drawn on one image can be transferred to the other image, in order to check the position of a block edge or an anatomical landmark. (Note that in this case, the physician indicated in black on the portal film that the block needed to be recut because it appeared to be insufficient (Fig. 5).) Using the track ball, the outline of the block shadow was traced onto the portal image and then overlaid onto the simulator image. (These lines are white in the illustration, but are shown in color on the screen for easy viewing.) Notice how closely the white line follows the black outline of the block as designed
Fig. 5. Posterior para-aortic setup, simulator film image on the left and portal film image on the right. The black line on the portal is an indication made by the physician to recut the block. The white line is the block shadow as outlined on the portal image and transferred to the simulator image using the compare routine. This block was cut correctly.
on the simulator film. In retrospect, this indicates that, in fact, the block was cut correctly, and it was the patient position that was in error. This conclusion was confirmed by further investigation using the patient’s anatomical landmarks. The same overlay procedure was used (Fig. 6), this time outlining one of the T-spine vertebral bodies. Once the vertebral body outline was transferred from the portal image on the right to the simulator image on the left, it became apparent that the patient was not set up accurately. (Note the discrepancy of about 1 cm on the position of this vertebra.) Figure 7 illustrates the corrected field, with the block outline matching. The lower left image is an overlay of the simulator and portal images. Note that the block shadow from the portal film is matching the black outline of the block drawn on the simulator film. Figure 8 confirms the correct patient position, again by using one of the vertebral bodies,
A4RI assistedfield design Another important application of the RFES is the incorporation of MRI data into the design of the treatment fields. By scaling and aligning the MRI
Fig. 7. The same para-aortic setup with the corrected portal film. Notice the overlay of the two images in the lower left corner. where the block shadow from the portal matches the block outline in black on the simulator image.
Use of a film enhancement
system 0 E. L.
!~LOWSKY
CI a/.
111
courages participation from those seated in the room. The size comparison is illustrated in Fig. 10. CONCLUSIONS
Fig. 8. The same setup now showing the corrected patient position as indicated by the outline of the vertebral body.
image to the simulator image, it allows the radiotherapist to check the block design for cone-downs to the tumor volume. In the example shown (Fig. 9), the two images were aligned using three anatomical landmarks. As can be seen, it was difficult in this case to identify three points in an area with no sharp intersections of bones or well defined comers. (It may be desirable to use “MRI opaque” fiducial markers for future “planning” MRI scans. The accuracy of this technique is also affected by the fact that the simulator film is a projected radiograph, while the MRI image is a single slice.) The tumor volume as seen on the MRI was outlined with the track-ball and transferred to the lateral simulation film. The overlay of the MRI tumor volume on the simulator image was then used to help guide the radiotherapist in designing field shaping blocks for the cone-down fields. Use of RFES during treatment planning conferences CT scan films and MRI films are usually printed in a format containing nine images per page of film, each image measuring approximately 3.5 inches square. Although this size is suitable for viewing by one or two people standing close to the film, in a conference or chart rounds, it is not easy for everyone in the room to see what is being discussed. By connecting the RFES to a large screen monitor (17 inches wide by 12 inches high), we have provided a very readable full screen display of the CT or MRI image for use during conferences or teaching sessions. This is a definite advantage over the small film panels, and en-
Fig. 9. Example of a lateral whole brain simulator film image with a corresponding sagittal MRI film image. The tumor volume is outlined on the MRI, and transferred to the simulator image.
The Radiotherapy Film Enhancement System (RFES) has been a worthwile addition to our department. Its ability to provide high quality simulator and portal image printouts for placement in the patient chart and the comparison portions of the program are the two functions we have found to be most promising. The enhancing routines are very fast (approximately 30 seconds), and the software is easy to run with a minimum of instruction. Acceptance of the system within the clinic has been steadily growing. The simulator technologists are already comfortable with its use, and are quite dependent on it for improving and copying their films. The reaction of the treatment technologists to having ready access to the simulator and portal images in the chart has been mixed. Some problems exist with the actual attachment of the printed images to the chart pages, in that they tend to be dislodged over time through normal wear and tear. (Special selfadhesive mounting sheets are under development, which should eliminate this problem.) The dosimetrists, however, rely on the images in the chart to aid in performance of a more comprehensive weekly chart review. While we have already demonstrated the usefulness of the RFES comparison routines for determining whether an apparent block cutting error requires a new block to be cut or simply the repositioning of the patient, it has thus far only been used occasionally for this purpose. This is because the comparison routines require more time and user interaction to perform. It is expected that its use will grow as the staff becomes more familiar with its operation. Several suggestions for improving the RFES and the ease of user interaction are:
Fig. 10. Photograph of the large screen video display, with the same CT film for size comparison.
112
Medical Dosimetry
The three point entry needed for image alignment should be augmented by a routine that allows the entry of curved contour lines as well (such as a segment of the lung/soft tissue interface or rib outline). It is often difficult to find three distinct pairs of anatomical points on the films, and such an improvement would make the comparison procedure easier. ?? The RFES should have the option to generate larger format printouts, which can be used as templates for block cutting. 0 A data base capability should be provided so that images can be stored and retrieved by indexing according to physician or patient name. (Currently, stored images are identified by numbers only.) This would allow a radiotherapist to retrieve all his stored portal images for review at the end of the day. ??
There are many features of the system we have not yet explored, such as the measurement of errors in
Volume IS, Number 3. 1990
block or patient position and the other enhancement algorithms. We hope to investigate these as well in the near future.
Acknowledgements-Special thanks to Atomic Products Corporation for supplying the system and their support. and to the Staff of the Department of Radiation Oncology at Stony Brook for their assistance and cooperation during this study.
REFERENCES 1. Marks, J.E.; Bedwinek, J.M.; Lee. F.; Purdy, J.A.; Perez. CA. Dose response analysis in nasopharyngeal carcinoma. Cancer 50:1042; 1982. 2. Goitein, M.; Busse, J. Immobilization error: some theoretical considerations. Radiology 117:407-4 12; 1975. 3. Byhardt, R.W.; Cox, J.D.; Homburg, A.; Liermann, G. Weekly localization films and detection of field placement errors. Int. J. Radiat. Oncol. Biol. Phys. 4~88l-887: 1978. 4. Leszczynski, K.W.; Shalev, S. A robust algorithm for contrast enhancement by local histogram modification. Image And Vision Computing 7:205-209; 1989.