- Email: [email protected]
A technique for the routine verification of treatment accuracy
216
Radiation Oncology, Biology,
Physics
Volume
24, Supplement
1
alignment method was noticeably more reuroduclble between users aI about i I .Onun. The variation between users was lareer uslnt! thefiducial point method at aboit * 1.5 mm. similar to the mantle field. Larger user varlabdlty was in pan due to poor imagi quality: non-rigidity of the anatomical features. and the difficult in locating an exact point on a contmuous anatomical structure. Results for rhe head and neck sites are presently under evaluation. 7
Conclusions: Quantitative verificanon requxes accurauz alignment of penal images with a reference or a prescripuon image. In well controlled phantom studies, both the fiducial points and template methods provide ~millar and adequate results. The phantom studies are also usefulin showing the changes in image features due to various setup variation, and the variability of the alignment due to different choice of anatomical features. In more realistic clinical situations, the template ahgnment method appears to be more precise, particularly when the image contrasr IS poor and only the &TOSSoutline of an anatomxal feature is shown.
154 CLINICAL Michael
USE OF ON-LINE G. Herman,
Division
PORTAL
Ph.D.,
of Radiation
IMAGING
FOR DAILY
Ross A. Abrams,
Oncology,
The Johns
Purpose: To determine whether electronic in radiation therapy. accuracy
M.D.,
PATIENT Kam-Shing
Hopkins
megavoltage
TREATMENl
Oncology portal
Lam, Ph.D., Center,
imaging
Rulon
R. Mayer,
Baltimore,
can be used
to
Ph.D.,
Wing-Chee
Ph.D.
MD enhance
Lam,
daily
treatment
Materials and Methods: A user friendly computer-controlled fluoroscopic screen mirror CCD camera imaging system (Philips SRI100) has been installed on a dual energy x-ray machine. High quality images comparable to those obtained by film require less than 5 cGy dose administration and are generated in a few seconds. A protocol was developed for clinical use of the system where images were obtained with short exposures at the beginning of external beam treatment. Patient set-up accuracy was determined by comparison with the set-up film. If misalignment was found, patient set-up was adjusted and the portal image repeated. The final accepted image was stored for the treatment record. Error analysis was facilitated by overlaying digital landmarks from a reference portal on the on-line image. Over 600 patient images were collected and analyzed. Results: Initial evaluation of 211 on-line images showed 70% were comparable or superior to standard film images. Sixty-five percent of the treatment fields fit completely within the imaging area. About 5% of the images have been rejected due to poor image quality. With a criterion of 5 mm variation, 15% of the daily treatments required set-up adjustment. Analysis of the accepted images shows a definite improvement in geometric variation attained with daily imaging. Correlations between frequency of required adjustment and anatomical site indicate which treatments benefit from daily imaging. Conclusion: Electronic on-line portal imaging facilitates daily portal alignment, verification and improved treatment accuracy. Ease of operation, almost instantaneous viewing and ability to record localization landmarks confirm the potential to replace film based systems. Results of future clinical studies on the effects of improved set-up accuracy on disease control will determine whether electronic portal imaging is accepted as a daily requisite procedure.
155 EFFICIENCY
AND DECISION
MAKING IN ON-LINE PORTAL VERIFICATION.
R. Bissett. S. Ryder,
K. Leszczynski,
Northeastern
Regional Cancer Centre. Sudhury.
Ontario
S. Cosby. R. Swart and P Dunscomhe
Ontario.
P3E SJ I
Purpose: On-line portal imaging offers the potential for detection and rectification of localization errors in patient set-up at the commencement of each radiotherapy treatment. The purpose of this study was to investigate possible approaches to routine treatment verification with an on-line portal imaging device, and to devise efficient ways of arriving at decisions on approving or otherwise a treatment field. Materials and Methods: accelerator.
Daily fraction
On-line portal images were acquired using the BEAMVIEW” system mounted on a MEVATRON” MX-2 6MV linear images for 38 patients (one field per patient) have been acquired to date: treatment sites included brain, head and neck.
217
Proceedings of the 34th Annual ASTRO Meeting
breast. lung. mantle and abdomen. The same technologists who operated the treatment unit were operating the portal imager and evaluating its impact on the patient set-up, In addition the adequacy of the acquired on-line portal images for verification of the treatment was assessed. The accuracy of each imaged treatment was verified and, if deemed necessary, treatments were interrupted and set-up discrepancies corrected. The impact of using the portal imaging system on patient throughput was evaluated from treatment times recorded by a TDMS” record and verify system. Off-line
and independently
Radiation
Oncologists
repeated the evaluation
of field placement accuracy.
LResults’ To date images of 3 15 treatments have been collected and evaluated. The image quality was judged by technologists to be adequate for the purpose treatment verification in 95% of cases. Discrepancies requiring adjustments in the treatment set-up were indicated in 3.7% of cases. There was no negative impact on patient throughput as the average treatment time remained essentially unchanged whether or not BEAMVIEW was used. In the oncologist’s evaluation the image quality was inadequate. due to the poor contrast in fewer than I % of images. Comparison between the oncologist’s and the technologists’ evaluation of the accuracy of individual treatments indicated significant discrepancies in their judgments and a dependence on the decision making threshold. High sensitivity (over 90%) of the technologists’ assessment in detection of discrepancies indicated a posreriori by the oncologist was only possible with a very strict acceptance criterion and with a reduction in specificity.
of
Conclusions: On-line portal imaging was determined to he useful tool in evaluating accuracy of the treatment set-up and no negative impact on patient throughput was observed with the adopted method of use. Concerted efforts are required to establish the treatment acceptance criteria which are necessary for achieving high accuracy in recognizing localization errors.
156 A
TECHNIQUE
T.
Radcliffe,
Manitoba
FOR THE ROUTINE VERIFICATION
Cancer
G.
Gluhchev, Treatment
M. and
Hoppensack, Research
OF TREATMENT ACCURACY. K.
McGee,
Foundation,
S.
Shalev
Winnipeg,
Manitoba
Purpose: We have developed an objective, computerized technique display, and analysis of field positioning errors (FPE’s) using on-line portal images. Both physical and dosimetric criteria significant FPE’s requiring immediate corrective action.
R3E OV9,
Canada
for the determination, portal films or digital are used to identify
Materials & Methods: Images were acquired with an on-line portal imaging system’ mounted on a 6/23 MV linear accelerator, during routine treatments to the head and neck, thorax and pelvis. In some cases portal films were also exposed and digitized, for comparison with the on-line images. Field size and shape were determined automatically with a very fast (cl sec.) robust algorithm, and compared with the prescribed field parameters. Patient location within the field was determined by delineating a set of anatomical landmarks, and also by using fiducial markers placed on the immobilization A 2-D rigid cast. body transformation aligned the treatment field with the digitized simulator image, and FPE’s were displayed as colored overlays showing target underdose areas (TU) and normal tissue overdose areas (NTO). Results: FPE’s were quantified in terms of 2-D field displacements and rotations, and as values of TU and NTO. No significant differences were found between data obtained from film or on-line images. However, FPE’s determined from anatomical landmarks were much larger than from fiducial markers, indicating that patient movement within the cast is a significant problem. TU and NT0 can normally be maintained within about 5 cm* for head and neck treatments, within 15 cm2 for thorax, and up to 30 cm2 for treatments to the pelvis. Action levels have been determined for each site, indicating the need for corrective action before continuing treatment. The type of error, and the correction required, are displayed as a guide to the radiotherapist. Conclusion: Routine evaluation of FPE’s has been shown to be possible with and digital portal imaging, although the latter is much faster and easier. field size or shape are determined automatically, while FPE’s are quantified need for corrective action is indicated for each individual treatment.
’ BEAMVI EW, Siemens
Medical
Laboratories,
Concord,
CA.
both film Errors in and the
Radiation Oncology, Biology,
Physics
Volume
24, Supplement
1
alignment method was noticeably more reuroduclble between users aI about i I .Onun. The variation between users was lareer uslnt! thefiducial point method at aboit * 1.5 mm. similar to the mantle field. Larger user varlabdlty was in pan due to poor imagi quality: non-rigidity of the anatomical features. and the difficult in locating an exact point on a contmuous anatomical structure. Results for rhe head and neck sites are presently under evaluation. 7
Conclusions: Quantitative verificanon requxes accurauz alignment of penal images with a reference or a prescripuon image. In well controlled phantom studies, both the fiducial points and template methods provide ~millar and adequate results. The phantom studies are also usefulin showing the changes in image features due to various setup variation, and the variability of the alignment due to different choice of anatomical features. In more realistic clinical situations, the template ahgnment method appears to be more precise, particularly when the image contrasr IS poor and only the &TOSSoutline of an anatomxal feature is shown.
154 CLINICAL Michael
USE OF ON-LINE G. Herman,
Division
PORTAL
Ph.D.,
of Radiation
IMAGING
FOR DAILY
Ross A. Abrams,
Oncology,
The Johns
Purpose: To determine whether electronic in radiation therapy. accuracy
M.D.,
PATIENT Kam-Shing
Hopkins
megavoltage
TREATMENl
Oncology portal
Lam, Ph.D., Center,
imaging
Rulon
R. Mayer,
Baltimore,
can be used
to
Ph.D.,
Wing-Chee
Ph.D.
MD enhance
Lam,
daily
treatment
Materials and Methods: A user friendly computer-controlled fluoroscopic screen mirror CCD camera imaging system (Philips SRI100) has been installed on a dual energy x-ray machine. High quality images comparable to those obtained by film require less than 5 cGy dose administration and are generated in a few seconds. A protocol was developed for clinical use of the system where images were obtained with short exposures at the beginning of external beam treatment. Patient set-up accuracy was determined by comparison with the set-up film. If misalignment was found, patient set-up was adjusted and the portal image repeated. The final accepted image was stored for the treatment record. Error analysis was facilitated by overlaying digital landmarks from a reference portal on the on-line image. Over 600 patient images were collected and analyzed. Results: Initial evaluation of 211 on-line images showed 70% were comparable or superior to standard film images. Sixty-five percent of the treatment fields fit completely within the imaging area. About 5% of the images have been rejected due to poor image quality. With a criterion of 5 mm variation, 15% of the daily treatments required set-up adjustment. Analysis of the accepted images shows a definite improvement in geometric variation attained with daily imaging. Correlations between frequency of required adjustment and anatomical site indicate which treatments benefit from daily imaging. Conclusion: Electronic on-line portal imaging facilitates daily portal alignment, verification and improved treatment accuracy. Ease of operation, almost instantaneous viewing and ability to record localization landmarks confirm the potential to replace film based systems. Results of future clinical studies on the effects of improved set-up accuracy on disease control will determine whether electronic portal imaging is accepted as a daily requisite procedure.
155 EFFICIENCY
AND DECISION
MAKING IN ON-LINE PORTAL VERIFICATION.
R. Bissett. S. Ryder,
K. Leszczynski,
Northeastern
Regional Cancer Centre. Sudhury.
Ontario
S. Cosby. R. Swart and P Dunscomhe
Ontario.
P3E SJ I
Purpose: On-line portal imaging offers the potential for detection and rectification of localization errors in patient set-up at the commencement of each radiotherapy treatment. The purpose of this study was to investigate possible approaches to routine treatment verification with an on-line portal imaging device, and to devise efficient ways of arriving at decisions on approving or otherwise a treatment field. Materials and Methods: accelerator.
Daily fraction
On-line portal images were acquired using the BEAMVIEW” system mounted on a MEVATRON” MX-2 6MV linear images for 38 patients (one field per patient) have been acquired to date: treatment sites included brain, head and neck.
217
Proceedings of the 34th Annual ASTRO Meeting
breast. lung. mantle and abdomen. The same technologists who operated the treatment unit were operating the portal imager and evaluating its impact on the patient set-up, In addition the adequacy of the acquired on-line portal images for verification of the treatment was assessed. The accuracy of each imaged treatment was verified and, if deemed necessary, treatments were interrupted and set-up discrepancies corrected. The impact of using the portal imaging system on patient throughput was evaluated from treatment times recorded by a TDMS” record and verify system. Off-line
and independently
Radiation
Oncologists
repeated the evaluation
of field placement accuracy.
LResults’ To date images of 3 15 treatments have been collected and evaluated. The image quality was judged by technologists to be adequate for the purpose treatment verification in 95% of cases. Discrepancies requiring adjustments in the treatment set-up were indicated in 3.7% of cases. There was no negative impact on patient throughput as the average treatment time remained essentially unchanged whether or not BEAMVIEW was used. In the oncologist’s evaluation the image quality was inadequate. due to the poor contrast in fewer than I % of images. Comparison between the oncologist’s and the technologists’ evaluation of the accuracy of individual treatments indicated significant discrepancies in their judgments and a dependence on the decision making threshold. High sensitivity (over 90%) of the technologists’ assessment in detection of discrepancies indicated a posreriori by the oncologist was only possible with a very strict acceptance criterion and with a reduction in specificity.
of
Conclusions: On-line portal imaging was determined to he useful tool in evaluating accuracy of the treatment set-up and no negative impact on patient throughput was observed with the adopted method of use. Concerted efforts are required to establish the treatment acceptance criteria which are necessary for achieving high accuracy in recognizing localization errors.
156 A
TECHNIQUE
T.
Radcliffe,
Manitoba
FOR THE ROUTINE VERIFICATION
Cancer
G.
Gluhchev, Treatment
M. and
Hoppensack, Research
OF TREATMENT ACCURACY. K.
McGee,
Foundation,
S.
Shalev
Winnipeg,
Manitoba
Purpose: We have developed an objective, computerized technique display, and analysis of field positioning errors (FPE’s) using on-line portal images. Both physical and dosimetric criteria significant FPE’s requiring immediate corrective action.
R3E OV9,
Canada
for the determination, portal films or digital are used to identify
Materials & Methods: Images were acquired with an on-line portal imaging system’ mounted on a 6/23 MV linear accelerator, during routine treatments to the head and neck, thorax and pelvis. In some cases portal films were also exposed and digitized, for comparison with the on-line images. Field size and shape were determined automatically with a very fast (cl sec.) robust algorithm, and compared with the prescribed field parameters. Patient location within the field was determined by delineating a set of anatomical landmarks, and also by using fiducial markers placed on the immobilization A 2-D rigid cast. body transformation aligned the treatment field with the digitized simulator image, and FPE’s were displayed as colored overlays showing target underdose areas (TU) and normal tissue overdose areas (NTO). Results: FPE’s were quantified in terms of 2-D field displacements and rotations, and as values of TU and NTO. No significant differences were found between data obtained from film or on-line images. However, FPE’s determined from anatomical landmarks were much larger than from fiducial markers, indicating that patient movement within the cast is a significant problem. TU and NT0 can normally be maintained within about 5 cm* for head and neck treatments, within 15 cm2 for thorax, and up to 30 cm2 for treatments to the pelvis. Action levels have been determined for each site, indicating the need for corrective action before continuing treatment. The type of error, and the correction required, are displayed as a guide to the radiotherapist. Conclusion: Routine evaluation of FPE’s has been shown to be possible with and digital portal imaging, although the latter is much faster and easier. field size or shape are determined automatically, while FPE’s are quantified need for corrective action is indicated for each individual treatment.
’ BEAMVI EW, Siemens
Medical
Laboratories,
Concord,
CA.
both film Errors in and the