- Email: [email protected]
An Alternative Solution for Adaptive IMRT Planning for Head and Neck Cancer Patients
I. J. Radiation Oncology d Biology d Physics
S740
Volume 78, Number 3, Supplement, 2010
prostate cancer patients. The dosimeter measured doses were compared with calculated doses based on volumetric images of both planning CT and kV-CBCT acquired for patient positioning before each treatment. The prostate was treated to 3625 cGy at 725 cGy per fraction. Results: The accuracy and reproducibility of the dosimeters were shown to be within the manufacture stated uncertainties of \ 5.5% based on the phantom study. In each treatment fraction, the patient was accurately positioned based on the dosimeters which were used as fiducial markers. Large measured dose variations (range, from -4% to +35%) compared to treatment planned dose were observed between fractions for the dosimeter that was implanted in a region with dose gradients. However, the calculated dose to the dosimeters based on the kV-CBCT volumetric images was found to be consistent with that based on planning CT images with beam isocenter setting according to fiducial markers. The SBRT radiation treatment delivery took 30-40 minutes to complete. The measured large dose fluctuation was considered to real because the magnitude of observed fluctuation is much greater than that of the uncertainty of the dosimeters. This result indicated that there was difference between target position set by the image guidance and that during the radiation treatment delivery due to the variation in bowel distention and bladder filling especially for prostate treatment. Conclusions: Direct tumor dose measurement using implanted DVS-HFT dosimeters provide a very effective tool to confirm the actual dose delivered to the target in each treatment fraction as well as the integrated total dose. The system is capable to provide the most direct and rigorous radiotherapy treatment quality assurance, especially for hypo-fractionated SBRT radiation treatments. Author Disclosure: G.X. Ding, None; E. Yang, None; C. Coffey, None; A. Malcolm, None.
3196
Dosimetric Errors Caused by Couch Shifts: An Investigation with Phantom Measurements and Computer Simulation
J. Duan, S. Shen, R. A. Popple, X. Wu, I. A. Brezovich University of Alabama Medical Center, Birmingham, AL Purpose/Objective(s): Modern treatment planning systems are capable of including couch attenuation in dose calculations. However, clinically it is difficult to accurately reproduce the patient position relative to the couch during the treatment. Such couch position discrepancies can result in significant dose deviations in treatment delivery. We evaluated the magnitude of the dosimetric errors caused by couch lateral shifts with phantom measurements and computer simulation. Materials/Methods: Radiation doses in a rectangular acrylic phantom was measured for 6-MV beams passing through a Varian Exact Couch with couch lateral shifts of up to 5 cm in each direction. The doses were measured at various gantry angles including those for 5, 7 and 9 equi-spaced beam arrangements, and were compared to those measured without couch shifts. Computer simulation was performed on a commercial treatment planning system for IMRT plans of a prostate patient. Treatment plans were generated with 7 equi-spaced beams (IEC 0 , 52 , 103 , 154 ,206 , 258 and 309 ) to cover both the low and high dose PTvs. (PTV56 and PTV70) while sparing the critical structures. Couch shifts were simulated by laterally moving the couch support structures (couch top and rails). Dose discrepancies between treatment plans with and without couch shifts were calculated for the PTvs., body and critical structures to assess the potential dose delivery errors. Results: Measurements showed that lateral couch shifts could cause dose discrepancies greater than 20%. While a shift of 1 cm could introduce dose errors as large as 10.9%, the largest dose errors occurred when couch rails were moved into or out of the beam. Computer simulation results exhibited dose deviations from -5% to 6% of the prescription dose (70 Gy) due to couch lateral shifts. Doses to the PTvs. varied by -3.9% to 3.0% for PTV56 and -2.3% to 2.1% for PTV70, respectively. Doses for individual fields deviated as much as 26.5% for the two affected beams (IEC 154 and 206 ). It was observed that for treatment plans with 5, 7 and 9 equi-spaced beams, moving the support rails to the sides of the couch generally reduced dose discrepancies. Conclusions: Attenuation by the support structures of the couch can cause clinically significant dose deviations if the couch geometry in the treatment plan is not accurately reproduced. In cases where the couch cannot be positioned accurately, moving the supporting rails to the sides of the couch can mitigate the problem for commonly used 5, 7 and 9 equi-spaced beam arrangements. Author Disclosure: J. Duan, None; S. Shen, None; R.A. Popple, None; X. Wu, None; I.A. Brezovich, None.
3197
An Alternative Solution for Adaptive IMRT Planning for Head and Neck Cancer Patients
1
Y. Feng , K. Sun2, L. Wang3, C. Yu1 1
University of Maryland School of Medicine, Baltimore, MD, 2Nankai University, Tianjin, China, 3Tianjin 4th Hospital, Tianjin, China Purpose/Objective(s): To access the effectiveness of adaptive IMRT plan modification using Direct Aperture Deformation (DAD) scheme for head and neck cancer patients using repeated CT scans. Materials/Methods: Fifteen patients underwent repeated CT scans during IG-IMRT treatment courses are studied. The plans use 7 to 9 coplanar beams to deliver the prescribed dose to the PTV which includes tumor/tumor bed and bilateral lymph nodes. All contours are delineated by the same physician. A deformable image registration is performed to align the new image set with the original image set used for planning. The resultant 3D deformation vectors are projected to all beam directions and used to deform the segment apertures of the original plan to the new CT (DAD plan). For comparison, a plan is re-optimized with the new CT for each patient (Re-Op plan) and used as the ‘‘gold standard’’. To simulate IMRT treatment with the CT image guidance but without replanning, the original plan is copied to the new image set with its isocenter shifted to the new PTV centroid (SHIFT plan). The plans are compared with measures of V95 and Conformity Index(CI) for PTV, D0.1cc for spinal cord and V30 for parotid glands. Student ttest is used for the statistical analysis.
Results: DAD improves V95% by 6% and CI by 0.04 with statistical significance as compared with SHIFT. On average, V95 and CI of DAD are 1.8% and 0.01 less than in Re-Op, respectively. No statistical significance is found in comparing of DAD vs. Re-Op for both PTV and avoidance structures.
Proceedings of the 52nd Annual ASTRO Meeting Conclusions: For head-and-neck patients with obvious anatomical changes during their IMRT treatment courses, image-guided repositioning is far from optimal and re-planning is needed. The DAD adaptation scheme can achieve similar plan quality as the Re-Op scheme without the lengthy re-planning process. It offers an alternative solution for adaptive IMRT planning. Author Disclosure: Y. Feng, None; K. Sun, None; L. Wang, None; C. Yu, None.
3198
Estimating a Delineation Uncertainty Margin to Account for Inter-observer Variability in Breast Cancer Radiotherapy
L. C. Holloway1,2, M. G. Jameson1,3, V. Batumalai1, E. Koh1,4, G. Papadatos1, D. Lonergan1,4, G. P. Delaney1,4 1
Liverpool and Macarthur Cancer Therapy Centres, Sydney, Australia, 2Institute of Medical Physics, University of Sydney, Sydney, Australia, 3Centre for Medical Radiation Physics, University of Wollongong, Wollongong, Australia, 4University of New South Wales, Sydney, Australia Purpose/Objective(s): Inter-observer variability (IOV) in target delineation in 3D conformal radiotherapy (RT) for breast cancer has been increasingly acknowledged, with its resultant dosimetric and clinical impact. This study aimed to, propose methodology for and quantify the margin to account for delineation uncertainty (MDU). Materials/Methods: Eight observers specializing in breast RT from a single institution independently delineated the breast CTV utilizing a departmental delineation protocol on the same CT images of ten representative breast cancer patients. The union volume for a given patient dataset was determined as the volume encompassing all generated contours. The mean position and SD of the most anterior(A), posterior(P), medial(M), lateral(L), superior(S) and inferior(I) points for all CTvs., excluding significant outliers, was determined. The MDU for a specific patient dataset was calculated as MDUid = 4 x SDid where i is the patient number and d is the direction of the margin (A,P,M,L,S or I), such that the expanded volume should cover 95% of the union volume. A margin to be used for future patients was then calculated by averaging for all patients; MDUfd = sum(MDUi = 1,2,.nd)/n, where n is the number of patients in the dataset. To test this methodology, MDUfd to be used for each patient dataset was calculated for all directions based on the other nine datasets (n = 9). The calculated MDUfd was then added to each original contour to form the composite CTV (cCTV). The resultant cCTvs. were then compared with the union volumes to ensure the union volume was covered but the most A,P,M,L,S,I positions were not increased by greater than the margin beyond the union volume positions. Results: The MDUid values ranged from 0.6-1.8cm (A,P), 0.6-1.9cm (M,L) and 0.6-2.7cm (S,I). MDUfd margin values calculated for all 10 patients were 1.2cm (A,P), 1.2cm (M,L) and 1.6cm (S,I). Despite the range of MDUid values, when MDUfd values were used, 99% of cCTvs. covered the union volume. The distance between the cCTV and union volume for all directions was less than MDUfd for all contours except for one outlier. Conclusions: The methodology proposed here for determining delineation uncertainty margins warrants further validation in a larger dataset. Calculated MDUs ranged from 1.2-1.6cm for the dataset considered here. Consideration should be given to delineation uncertainty margins which account for IOV in breast target volume generation. These should be assessed for any given institution and delineation protocol. Author Disclosure: L.C. Holloway, None; M.G. Jameson, None; V. Batumalai, None; E. Koh, None; G. Papadatos, None; D. Lonergan, None; G.P. Delaney, None.
3199
Non-cluster TomoTherapy Treatment Planning System
W. Lu, Q. Chen, M. Chen, Y. Chen, G. Olivera TomoTherapy Inc., Madison, WI Purpose/Objective(s): Besides the Planning Station (PS) computer, the current TomoTherapy Treatment Planning System (TPS) includes a separate computer cluster with 7-14 nodes for optimization and dose calculation. We developed a novel Planning-Station-only system for TomoTherapy treatment planning. Compared with the cluster-TPS, the new solution will significantly reduce the cost and improve in both plan quality and planning throughput. Materials/Methods: The current TomoTherapy plan optimization requires pre-calculation and storage of large amount of beamlets. A computer cluster is used for both computation and data storage power to accommodate the very large scale (VSL) optimization problem. In this work, we developed a new direct-machine-parameter-optimization (DMPO) scheme that features a non-voxel and broad-beam (NVBB) representation and does not rely on beamlets. Low-memory, full computation and data parallelization nature of this scheme facilitate its efficient implementation on the graphic processing unit (GPU). We incorporated the NVBB approach into TomoTherapy TPS. The dose calculation and optimization engine runs on the same PS computer. The graphic card of the PS computer is replaced by a Navida GeForce GTX295 (GPU) card. Extensive verification and validation tests were performed in house and via third parties. Benchmarks on dose accuracy, plan quality and throughput were compared with the commercial TomoTherapy TPS with the 14-blade DC3 cluster. Clinical case studies include the prostate, lung, breast, H&N and TBM for both TomoHelical and TomoDirect planning. Results: Compared with the current cluster-TPS, the new GPU-TPS reduced the pre-processing time from 10-200 minutes to 10 seconds as no beamlet pre-calculation is needed any longer. The iteration time was reduced to 25-90%. Drivability and plan qualities were indistinguishable for most cases and less dose artifacts were observable for a few cases via GPU-TPS. For the same delivery plan in full dose and final dose calculation, the GPU-TPS had speedup of about 8-16 times compared with the cluster-TPS, while the dose differences between cluster-TPS and GPU-TPS were within 1%, 1 mm for all test cases. Conclusions: The DMPO nature of the NVBB framework eliminates the needs of beamlets and leads to better plan quality. Extensive planning and benchmark studies validate the GPU-TPS. The non-cluster solution results in significant savings on the hardware and service cost. Compared with the cluster-TPS, planning time was reduced in many folds with the new GPU-TPS. With this new technique, VSL TomoTherapy treatment planning can even be accomplished via a single laptop. Author Disclosure: W. Lu, TomoTherapy Inc., A. Employment; Q. Chen, TomoTherapy Inc., A. Employment; M. Chen, TomoTherapy Inc., A. Employment; Y. Chen, TomoTherapy Inc., A. Employment; G. Olivera, TomoTherapy Inc., A. Employment.
S741
S740
Volume 78, Number 3, Supplement, 2010
prostate cancer patients. The dosimeter measured doses were compared with calculated doses based on volumetric images of both planning CT and kV-CBCT acquired for patient positioning before each treatment. The prostate was treated to 3625 cGy at 725 cGy per fraction. Results: The accuracy and reproducibility of the dosimeters were shown to be within the manufacture stated uncertainties of \ 5.5% based on the phantom study. In each treatment fraction, the patient was accurately positioned based on the dosimeters which were used as fiducial markers. Large measured dose variations (range, from -4% to +35%) compared to treatment planned dose were observed between fractions for the dosimeter that was implanted in a region with dose gradients. However, the calculated dose to the dosimeters based on the kV-CBCT volumetric images was found to be consistent with that based on planning CT images with beam isocenter setting according to fiducial markers. The SBRT radiation treatment delivery took 30-40 minutes to complete. The measured large dose fluctuation was considered to real because the magnitude of observed fluctuation is much greater than that of the uncertainty of the dosimeters. This result indicated that there was difference between target position set by the image guidance and that during the radiation treatment delivery due to the variation in bowel distention and bladder filling especially for prostate treatment. Conclusions: Direct tumor dose measurement using implanted DVS-HFT dosimeters provide a very effective tool to confirm the actual dose delivered to the target in each treatment fraction as well as the integrated total dose. The system is capable to provide the most direct and rigorous radiotherapy treatment quality assurance, especially for hypo-fractionated SBRT radiation treatments. Author Disclosure: G.X. Ding, None; E. Yang, None; C. Coffey, None; A. Malcolm, None.
3196
Dosimetric Errors Caused by Couch Shifts: An Investigation with Phantom Measurements and Computer Simulation
J. Duan, S. Shen, R. A. Popple, X. Wu, I. A. Brezovich University of Alabama Medical Center, Birmingham, AL Purpose/Objective(s): Modern treatment planning systems are capable of including couch attenuation in dose calculations. However, clinically it is difficult to accurately reproduce the patient position relative to the couch during the treatment. Such couch position discrepancies can result in significant dose deviations in treatment delivery. We evaluated the magnitude of the dosimetric errors caused by couch lateral shifts with phantom measurements and computer simulation. Materials/Methods: Radiation doses in a rectangular acrylic phantom was measured for 6-MV beams passing through a Varian Exact Couch with couch lateral shifts of up to 5 cm in each direction. The doses were measured at various gantry angles including those for 5, 7 and 9 equi-spaced beam arrangements, and were compared to those measured without couch shifts. Computer simulation was performed on a commercial treatment planning system for IMRT plans of a prostate patient. Treatment plans were generated with 7 equi-spaced beams (IEC 0 , 52 , 103 , 154 ,206 , 258 and 309 ) to cover both the low and high dose PTvs. (PTV56 and PTV70) while sparing the critical structures. Couch shifts were simulated by laterally moving the couch support structures (couch top and rails). Dose discrepancies between treatment plans with and without couch shifts were calculated for the PTvs., body and critical structures to assess the potential dose delivery errors. Results: Measurements showed that lateral couch shifts could cause dose discrepancies greater than 20%. While a shift of 1 cm could introduce dose errors as large as 10.9%, the largest dose errors occurred when couch rails were moved into or out of the beam. Computer simulation results exhibited dose deviations from -5% to 6% of the prescription dose (70 Gy) due to couch lateral shifts. Doses to the PTvs. varied by -3.9% to 3.0% for PTV56 and -2.3% to 2.1% for PTV70, respectively. Doses for individual fields deviated as much as 26.5% for the two affected beams (IEC 154 and 206 ). It was observed that for treatment plans with 5, 7 and 9 equi-spaced beams, moving the support rails to the sides of the couch generally reduced dose discrepancies. Conclusions: Attenuation by the support structures of the couch can cause clinically significant dose deviations if the couch geometry in the treatment plan is not accurately reproduced. In cases where the couch cannot be positioned accurately, moving the supporting rails to the sides of the couch can mitigate the problem for commonly used 5, 7 and 9 equi-spaced beam arrangements. Author Disclosure: J. Duan, None; S. Shen, None; R.A. Popple, None; X. Wu, None; I.A. Brezovich, None.
3197
An Alternative Solution for Adaptive IMRT Planning for Head and Neck Cancer Patients
1
Y. Feng , K. Sun2, L. Wang3, C. Yu1 1
University of Maryland School of Medicine, Baltimore, MD, 2Nankai University, Tianjin, China, 3Tianjin 4th Hospital, Tianjin, China Purpose/Objective(s): To access the effectiveness of adaptive IMRT plan modification using Direct Aperture Deformation (DAD) scheme for head and neck cancer patients using repeated CT scans. Materials/Methods: Fifteen patients underwent repeated CT scans during IG-IMRT treatment courses are studied. The plans use 7 to 9 coplanar beams to deliver the prescribed dose to the PTV which includes tumor/tumor bed and bilateral lymph nodes. All contours are delineated by the same physician. A deformable image registration is performed to align the new image set with the original image set used for planning. The resultant 3D deformation vectors are projected to all beam directions and used to deform the segment apertures of the original plan to the new CT (DAD plan). For comparison, a plan is re-optimized with the new CT for each patient (Re-Op plan) and used as the ‘‘gold standard’’. To simulate IMRT treatment with the CT image guidance but without replanning, the original plan is copied to the new image set with its isocenter shifted to the new PTV centroid (SHIFT plan). The plans are compared with measures of V95 and Conformity Index(CI) for PTV, D0.1cc for spinal cord and V30 for parotid glands. Student ttest is used for the statistical analysis.
Results: DAD improves V95% by 6% and CI by 0.04 with statistical significance as compared with SHIFT. On average, V95 and CI of DAD are 1.8% and 0.01 less than in Re-Op, respectively. No statistical significance is found in comparing of DAD vs. Re-Op for both PTV and avoidance structures.
Proceedings of the 52nd Annual ASTRO Meeting Conclusions: For head-and-neck patients with obvious anatomical changes during their IMRT treatment courses, image-guided repositioning is far from optimal and re-planning is needed. The DAD adaptation scheme can achieve similar plan quality as the Re-Op scheme without the lengthy re-planning process. It offers an alternative solution for adaptive IMRT planning. Author Disclosure: Y. Feng, None; K. Sun, None; L. Wang, None; C. Yu, None.
3198
Estimating a Delineation Uncertainty Margin to Account for Inter-observer Variability in Breast Cancer Radiotherapy
L. C. Holloway1,2, M. G. Jameson1,3, V. Batumalai1, E. Koh1,4, G. Papadatos1, D. Lonergan1,4, G. P. Delaney1,4 1
Liverpool and Macarthur Cancer Therapy Centres, Sydney, Australia, 2Institute of Medical Physics, University of Sydney, Sydney, Australia, 3Centre for Medical Radiation Physics, University of Wollongong, Wollongong, Australia, 4University of New South Wales, Sydney, Australia Purpose/Objective(s): Inter-observer variability (IOV) in target delineation in 3D conformal radiotherapy (RT) for breast cancer has been increasingly acknowledged, with its resultant dosimetric and clinical impact. This study aimed to, propose methodology for and quantify the margin to account for delineation uncertainty (MDU). Materials/Methods: Eight observers specializing in breast RT from a single institution independently delineated the breast CTV utilizing a departmental delineation protocol on the same CT images of ten representative breast cancer patients. The union volume for a given patient dataset was determined as the volume encompassing all generated contours. The mean position and SD of the most anterior(A), posterior(P), medial(M), lateral(L), superior(S) and inferior(I) points for all CTvs., excluding significant outliers, was determined. The MDU for a specific patient dataset was calculated as MDUid = 4 x SDid where i is the patient number and d is the direction of the margin (A,P,M,L,S or I), such that the expanded volume should cover 95% of the union volume. A margin to be used for future patients was then calculated by averaging for all patients; MDUfd = sum(MDUi = 1,2,.nd)/n, where n is the number of patients in the dataset. To test this methodology, MDUfd to be used for each patient dataset was calculated for all directions based on the other nine datasets (n = 9). The calculated MDUfd was then added to each original contour to form the composite CTV (cCTV). The resultant cCTvs. were then compared with the union volumes to ensure the union volume was covered but the most A,P,M,L,S,I positions were not increased by greater than the margin beyond the union volume positions. Results: The MDUid values ranged from 0.6-1.8cm (A,P), 0.6-1.9cm (M,L) and 0.6-2.7cm (S,I). MDUfd margin values calculated for all 10 patients were 1.2cm (A,P), 1.2cm (M,L) and 1.6cm (S,I). Despite the range of MDUid values, when MDUfd values were used, 99% of cCTvs. covered the union volume. The distance between the cCTV and union volume for all directions was less than MDUfd for all contours except for one outlier. Conclusions: The methodology proposed here for determining delineation uncertainty margins warrants further validation in a larger dataset. Calculated MDUs ranged from 1.2-1.6cm for the dataset considered here. Consideration should be given to delineation uncertainty margins which account for IOV in breast target volume generation. These should be assessed for any given institution and delineation protocol. Author Disclosure: L.C. Holloway, None; M.G. Jameson, None; V. Batumalai, None; E. Koh, None; G. Papadatos, None; D. Lonergan, None; G.P. Delaney, None.
3199
Non-cluster TomoTherapy Treatment Planning System
W. Lu, Q. Chen, M. Chen, Y. Chen, G. Olivera TomoTherapy Inc., Madison, WI Purpose/Objective(s): Besides the Planning Station (PS) computer, the current TomoTherapy Treatment Planning System (TPS) includes a separate computer cluster with 7-14 nodes for optimization and dose calculation. We developed a novel Planning-Station-only system for TomoTherapy treatment planning. Compared with the cluster-TPS, the new solution will significantly reduce the cost and improve in both plan quality and planning throughput. Materials/Methods: The current TomoTherapy plan optimization requires pre-calculation and storage of large amount of beamlets. A computer cluster is used for both computation and data storage power to accommodate the very large scale (VSL) optimization problem. In this work, we developed a new direct-machine-parameter-optimization (DMPO) scheme that features a non-voxel and broad-beam (NVBB) representation and does not rely on beamlets. Low-memory, full computation and data parallelization nature of this scheme facilitate its efficient implementation on the graphic processing unit (GPU). We incorporated the NVBB approach into TomoTherapy TPS. The dose calculation and optimization engine runs on the same PS computer. The graphic card of the PS computer is replaced by a Navida GeForce GTX295 (GPU) card. Extensive verification and validation tests were performed in house and via third parties. Benchmarks on dose accuracy, plan quality and throughput were compared with the commercial TomoTherapy TPS with the 14-blade DC3 cluster. Clinical case studies include the prostate, lung, breast, H&N and TBM for both TomoHelical and TomoDirect planning. Results: Compared with the current cluster-TPS, the new GPU-TPS reduced the pre-processing time from 10-200 minutes to 10 seconds as no beamlet pre-calculation is needed any longer. The iteration time was reduced to 25-90%. Drivability and plan qualities were indistinguishable for most cases and less dose artifacts were observable for a few cases via GPU-TPS. For the same delivery plan in full dose and final dose calculation, the GPU-TPS had speedup of about 8-16 times compared with the cluster-TPS, while the dose differences between cluster-TPS and GPU-TPS were within 1%, 1 mm for all test cases. Conclusions: The DMPO nature of the NVBB framework eliminates the needs of beamlets and leads to better plan quality. Extensive planning and benchmark studies validate the GPU-TPS. The non-cluster solution results in significant savings on the hardware and service cost. Compared with the cluster-TPS, planning time was reduced in many folds with the new GPU-TPS. With this new technique, VSL TomoTherapy treatment planning can even be accomplished via a single laptop. Author Disclosure: W. Lu, TomoTherapy Inc., A. Employment; Q. Chen, TomoTherapy Inc., A. Employment; M. Chen, TomoTherapy Inc., A. Employment; Y. Chen, TomoTherapy Inc., A. Employment; G. Olivera, TomoTherapy Inc., A. Employment.
S741