- Email: [email protected]
RayTracing and MonteCarlo Dose Calculation Might Lead to Clinically Relevant Differences for CyberKnife Lung SABR
Volume 93 Number 3S Supplement 2015 volumes. Analysis was done using computer algorithm software. CTVs of the prostate bed were compared using non-parametric statistics. Results: Contouring on CT alone showed a significantly (pZ0.001, Wilcoxon signed-rank test) higher similarity between observers with a mean DSC of 0.76 (SD 0.05) compared to MRI with a mean of 0.66 (SD 0.05). A good level of agreement (mean DSC 0.80 - 1) was seen in 4 patients with CT and none with MRI. A poor level of agreement (mean DSC <0.60) was seen in only one patient with MRI (DSC 0.59) and in none with CT contouring. While this patient had a poor DSC on MRI, his DSC on CT was 0.75. Mean intra-observer variability between CT and MRI was 0.68, 0.75 and 0.78 for the three observers. The CTV was on average 14%-58% larger on CT than on MRI. The intra-observer difference in CTV-volume between CT and MRI was significant in two observers and non-significant in the third one (pZ0.09). Conclusion: For prostate bed RT planning, there was less inter-observer variability when contouring on CT than on MRI. RTOG contouring guidelines are based on anatomical landmarks readily visible on CT. These landmarks are more inter-observer dependent on MRI. Therefore contouring guidelines might not be applicable to MRI planning. Author Disclosure: M. Barkati: None. G. Delouya: None. D. Simard: None. D. Taussky: None.
3418 Dosimetric Effect of Photon Beam Energy on VMAT Treatment Plan Quality Due to Body Habitus for Advanced Prostate Cancer D.N. Stanley,1 T. Popp,1 C.S. Ha,1 G.P. Swanson,2 T.Y. Eng,1 N. Papanikolaou,3 and A.N. Gutierrez3; 1The University of Texas Health Science Center San Antonio, San Antonio, TX, 2Baylor Scott & White Healthcare Temple Clinic, Temple, TX, 3University of Texas Health Science Center San Antonio, San Antonio, TX Purpose/Objective(s): The purpose of this study was to dosimetrically compare 6MV and 10MV photon beam energies in high-risk prostate cancer patients of varying body habitus using a VMAT radiation delivery technique. The objectives of the study were to evaluate if dosimetric differences exist and to investigate if differences are dependent on patient body habitus. Materials/Methods: Forty (nZ40) patients previously treated to the prostate and pelvic lymph nodes with VMAT techniques with varying body habitus were chosen. Patients were planned in a TPS using single or double VMAT plans with 6 and 10 MV photon energies. All patients were optimized with the same planning objectives and normalized such that 95% of the PTV received the prescription dose. Patients were evaluated for PTV and organ at risk (OAR) parameters for the bladder, rectum, small bowel, penile bulb and sigmoid colon. Metrics used for comparison were D2%, D98%, homogeneity, conformity (CN), and dose falloff for the PTV and D2%, Dmean, V80%, V60%, V40%for OARs. Statistical differences were evaluated using a paired sample Wilcoxon signed rank test with significance level of 0.05. Results: For the PTV, there were no statistically significant differences in Dmean, D2cc ,CN and HI values, but the R50 and R25 showed a median improvement of 6.7% (p<0.01) and 6.2% (p<0.01), respectively, with 10MV. A correlation between the patient anterior-posterior distance (dAP) and percent reduction in R50 of 0.436%/cm (p<0.01) was determined. For OARs, statistically significant reductions in dose metrics were found in the small bowel and bladder, but an increase in the D2ccof the penile bulb of 3.5% (p <0.01) was shown with 10MV. Conclusion: The study showed that 10MV provides a faster dose fall off in comparison to 6MV for patients treated with prostate and pelvic lymph nodes using a VMAT technique irrespective of body habitus; however, the improvement in dose fall off is dependent on body habitus and increases as the patient body habitus increases. Acknowledgements: This work was funded in part by the Cancer Prevention Research Institute of Texas Pre doctoral fellowship training grant (RP140105) to Dennis N. Stanley M.Sc. Author Disclosure: D.N. Stanley: Research Grant; Cancer Prevention Research Institute of Texas. T. Popp: None. C.S. Ha: None. G.P.
Poster Viewing Session E565 Swanson: None. T.Y. Eng: None. N. Papanikolaou: None. A.N. Gutierrez: None.
3419 One Size Does Not Fit All: Evaluating the Need for Separate Unilateral and Bilateral Radiation Therapy Normal Tissue Treatment Planning Goals for Oropharynx Cancer L. Wang, P. Flanagan, R. Gleason, C.T. Murphy, T. Churilla, and T.J. Galloway; Fox Chase Cancer Center, Philadelphia, PA Purpose/Objective(s): Dosimetric constraints for head and neck organs at risk (OARs) are based on bilateral neck radiation treatment. Common constraints for the contralateral parotid and submandibular gland are mean <26Gy and mean <39Gy respectively, and unilateral neck plans are able to meet these goals easily. However, studies suggest that relatively low mean doses to the salivary structures (<10Gy) result in better function. We hypothesize that more stringent OAR dose limits can be placed in unilateral neck radiation therapy plans to avoid contralateral and midline neck normal tissues without compromising quality of the plan. Materials/Methods: We identified 10 patients with oropharyngeal primaries treated with unilateral radiation. The treatment plans were recreated using whole neck IMRT. Prescriptions were 70/63/56Gy in 35 fractions to gross disease, high risk neck, and low risk neck, respectively. At least 95% of each planning target volume (PTV) received prescription dose. No more than 1 cc was allowed to receive 110% and at least 99.5% of each PTV was required to receive 95% of the prescription dose. Achievable doses to the contralateral salivary structures and the supraglottic larynx/glottic larynx (GSL) were evaluated. The GSL was cropped 5mm from the PTV. The Mann-Whitney U test was used to detect differences between mean doses. Results: All patients had T1 tumors of the tonsil or glossotonsillar sulcus. There were 2 N0, 2 N1, 3 N2A, and 3 N2B patients. All gross nodal disease was in level 2/3. Target coverage was achieved in all circumstances. All plans used fixed beam IMRT with either 5 ipsilateral beams (nZ3) or 5 ipsilateral plus 1 contralateral beam (nZ7). All plans were able to achieve mean doses to the contralateral parotid and submandibular gland of <10Gy. Mean contralateral parotid dose was 7.94Gy (range 6.429.50) and mean contralateral submandibular dose was 8.94Gy (range 7.689.86). Mean dose to the GSL was 25.28Gy (range 14.33-29.78). Lymph node involvement significantly increased mean dose to the GSL (N0: 16.79 vs N+: 27.40Gy, pZ0.04) but did not significantly increase mean dose to the contralateral salivary glands: parotid (N0: 7.56 vs N+: 8.09Gy, pZ0.53), submandibular (N0:8.10 vs N+:9.15Gy, pZ0.18). Conclusion: Established OAR dose limits for contralateral salivary structures and the GSL can be further optimized for unilateral neck radiation and much greater sparing of these structures can be achieved. For patients treated with unilateral radiation, contralateral parotid and submandibular gland mean doses <10Gy and GSL <30Gy can be consistently achieved with IMRT plans without compromising tumor coverage. Separate unilateral and bilateral neck IMRT goal sheets should be used to reflect different OAR constraints. Author Disclosure: L. Wang: None. P. Flanagan: None. R. Gleason: None. C.T. Murphy: None. T. Churilla: None. T.J. Galloway: None.
3420 RayTracing and MonteCarlo Dose Calculation Might Lead to Clinically Relevant Differences for CyberKnife Lung SABR D. Dechambre,1 L. Janvary,2 N. Jansen,1 S. Cucchiaro,1 C. Ernst,1 M. Devillers,1 V. Baart,1 P. Coucke,1 and G. Akos1; 1Lie`ge University Hospital, Lie`ge, Belgium, 2Institute of Oncology, Debrecen University Clinical Center, Debrecen, Hungary Purpose/Objective(s): Various calculation algorithms could lead to significant differences in dose distribution especially at large inhomogeneities within the irradiated volume (up to 20%). Such variation could be even more important for lung stereotactic ablative radiation therapy (SABR).
E566
International Journal of Radiation Oncology Biology Physics
Historically algorithm calculation discrepancies were addressed by excluding homogeneity correction (RTOG 0236, 0618) or changes in prescription (ROSEL), lately allowing only advanced algorithms to be used (RTOG 0813, 0915). Drawn conclusions were mainly based on experience with Linac based SABR technique (max 15-20 beams), but never extensively investigated for specific robotic machines with more freedom for beam arrangements (up to 200). Our aim is to present the physical, biological dose and tumor control probability (TCP) differences between MonteCarlo (MC) and Raytracing (RT) calculation algorithms for CyberKnife lung SABR. Materials/Methods: One hundred and thirty early-stage NSCLC patients were treated by SABR. Clinical target volume (CTV) to planning target volume (PTV) margin varied between 2 and 4 mm. Intended prescription dose was 3x20 Gy with risk-adaptation when required, based on RT algorithm, followed by a retrospective MC calculation. Biologically relevant dose differences were analyzed using CTV’s mean and generalized equivalent uniform dose (gEUD, a Z -10). Physical dose differences were collected combined with the corresponding biological equivalent dose (BED10) using a method suggested previously followed by an estimation of TCP differences. Results from the two algorithms were tested using a paired t-test with a significance level of p<0.05. Results: Between 04/2010 and 05/2012, 157 lesions were treated. Focusing on physical dose differences (RT vs. MC) of the CTV, average mean dose and gEUD (SD) were 63.6 (9.5) vs. 49.9 (8.1) and 61.5 (10.1) vs. 43.3 (9.9) Gy. Median mean dose loss was about 17%. 42% and 59% of the patient population showed mean and gEUD differences higher than 20%. Corresponding biological dose differences were on average 190 (53) vs. 130 (38) and 185 (55) vs. 105 (39) Gy respectively for mean and gEUD doses. Correlation was found between BED10 loss and smaller tumor size as well as tumor’s peripheral location for both mean and gEUD. Associated TCP were on average 97 (5)% vs. 93(7)% and 96(7)% vs. 87(12)%. Median TCP losses were about 2 and 6.5% for mean and gEUD-based TCP. All of the differences were statistically significant in every aspect. Conclusion: Dose calculation differences between RT and MC algorithm could lead to severe differences in both physical (up to 60% for small, peripheral lesion) and biological doses which could dramatically worsen the TCP. Follow-up is ongoing to assess the effect of underdosage (RT based treatment) on clinical outcomes. Author Disclosure: D. Dechambre: None. L. Janvary: None. N. Jansen: None. S. Cucchiaro: None. C. Ernst: None. M. Devillers: None. V. Baart: None. P. Coucke: None. G. Akos: None.
Materials/Methods: A phenomenological model that describes proton RBE in the framework of the linear-quadratic model was proposed. In the model, effect of local energy spectrum is characterized by the dose-averaged-local-linear-energy-transfer (DALLET). The calculation of DALLET is usually a time consuming task, and therefore not well suited to clinical application. In this study, we are developing a method that can calculate the RBE within the clinically accepted time frame. We first developed and validated a Monte Carlo (MC) program in Geant4 to model the proton pencil-beam-scanning (PBS) system installed in our institute. Then the MC program is employed to generate DALLET data for selected human tissues. These DALLET data depends on tissue type, initial energy, and the actual energy at the point where the proton interacts with tissue. The chemical composition and material density of all selected human tissues are taken from ICRP pub. 23 (1975). The DALLET data are then incorporated into an in-house developed computer code to calculate variable RBE-weighted dose distribution within the patient represented by CT. Since the DALLET data are pre-calculated and bundled as a look-up-table, the RBE computation time is reduced substantially. Results: The MC validation result is first presented, which includes (1) transverse profile in air at 5 different locations along the beam central axis for 5 selected energies, (2) the nozzle water-equivalent-thickness, and (3) all 107 proton PBS beam characteristics (range, peak width and peak-toentrance ratio) in water. The DALLET data, generated by the validated MC program, is then presented. After incorporating the pre-calculated DALLET data, the in-house computer code is employed to calculate variable RBE weighted dose distributions. As the application of the fast method, a couple of clinical cases are chosen to be recalculated, and the focus is on the distal end region of PBS beam. The comparison between fixed RBE and variable RBE is presented. Conclusion: We are developing a fast method to calculate 3-D RBE distribution for proton treatment. The method is expected to perform the proton’s RBE calculation with the accuracy and computation time accepted in a busy clinical setting. Author Disclosure: X. Ding: None. W. Liu: None. M. Fatyga: None. A. Anand: None. J. Shen: None. J. Stoker: None. M. Bues: None.
3421 A Fast Method to Calculate RBE Weighted Dose Distribution X. Ding, W. Liu, M. Fatyga, A. Anand, J. Shen, J. Stoker, and M. Bues; Mayo Clinic Arizona, Phoenix, AZ Purpose/Objective(s): It is well known that proton RBE is not a constant, but depends on several factors such as dose per fraction, biological endpoint, tissue type and local energy spectrum. However at the moment, the variable RBE in clinical application is limited, due to the fact that determination of the effect of local energy spectrum is time consuming. In this study, we present a fast method to calculate 3-D RBE weighted proton dose distribution.
3422 Quantitative Radiomics: Effects of Stochastic Variability on PET Textural Features and Implications for Clinical Trials M. Nyflot,1 S.R. Bowen,2 F. Yang,2 D. Byrd,2 G.A. Sandison,2 and P.E. Kinahan2; 1Department of Radiation Oncology, University of Washington, Seattle, WA, 2University of Washington, Seattle, WA Purpose/Objective(s): Image heterogeneity metrics such as textural features are an active area of research for evaluating clinical outcomes with PET imaging; however, the effects of stochastic image acquisition noise on these metrics are poorly understood. We tested the hypothesis that the majority of textural features would have greater sensitivity to stochastic effects (as measured by coefficient of variation [COV] and sample size estimation) than the standard deviation of the intensity histogram (SDIH). Materials/Methods: 50 statistically independent PET images of the NEMA IQ phantom were simulated with realistic noise and resolution properties. The phantom incorporated lesions with diameters of 17, 22, 28,
Poster Viewing Abstracts 3422; Table 1 Estimated sample size (patients) for given clinical effect size
30% effect
Lesion size Min Q1 Median Q3 Max
37 mm 3 3 5 6 22
15% effect 28 mm 3 3 5 6 313
22 mm 3 3 6 10 587
17 mm 3 4 7 14 52
37 mm 3 4 10 13 80
28 mm 3 5 10 16 1244
22 mm 3 5 14 30 2337
17 mm 3 6 17 44 200
3418 Dosimetric Effect of Photon Beam Energy on VMAT Treatment Plan Quality Due to Body Habitus for Advanced Prostate Cancer D.N. Stanley,1 T. Popp,1 C.S. Ha,1 G.P. Swanson,2 T.Y. Eng,1 N. Papanikolaou,3 and A.N. Gutierrez3; 1The University of Texas Health Science Center San Antonio, San Antonio, TX, 2Baylor Scott & White Healthcare Temple Clinic, Temple, TX, 3University of Texas Health Science Center San Antonio, San Antonio, TX Purpose/Objective(s): The purpose of this study was to dosimetrically compare 6MV and 10MV photon beam energies in high-risk prostate cancer patients of varying body habitus using a VMAT radiation delivery technique. The objectives of the study were to evaluate if dosimetric differences exist and to investigate if differences are dependent on patient body habitus. Materials/Methods: Forty (nZ40) patients previously treated to the prostate and pelvic lymph nodes with VMAT techniques with varying body habitus were chosen. Patients were planned in a TPS using single or double VMAT plans with 6 and 10 MV photon energies. All patients were optimized with the same planning objectives and normalized such that 95% of the PTV received the prescription dose. Patients were evaluated for PTV and organ at risk (OAR) parameters for the bladder, rectum, small bowel, penile bulb and sigmoid colon. Metrics used for comparison were D2%, D98%, homogeneity, conformity (CN), and dose falloff for the PTV and D2%, Dmean, V80%, V60%, V40%for OARs. Statistical differences were evaluated using a paired sample Wilcoxon signed rank test with significance level of 0.05. Results: For the PTV, there were no statistically significant differences in Dmean, D2cc ,CN and HI values, but the R50 and R25 showed a median improvement of 6.7% (p<0.01) and 6.2% (p<0.01), respectively, with 10MV. A correlation between the patient anterior-posterior distance (dAP) and percent reduction in R50 of 0.436%/cm (p<0.01) was determined. For OARs, statistically significant reductions in dose metrics were found in the small bowel and bladder, but an increase in the D2ccof the penile bulb of 3.5% (p <0.01) was shown with 10MV. Conclusion: The study showed that 10MV provides a faster dose fall off in comparison to 6MV for patients treated with prostate and pelvic lymph nodes using a VMAT technique irrespective of body habitus; however, the improvement in dose fall off is dependent on body habitus and increases as the patient body habitus increases. Acknowledgements: This work was funded in part by the Cancer Prevention Research Institute of Texas Pre doctoral fellowship training grant (RP140105) to Dennis N. Stanley M.Sc. Author Disclosure: D.N. Stanley: Research Grant; Cancer Prevention Research Institute of Texas. T. Popp: None. C.S. Ha: None. G.P.
Poster Viewing Session E565 Swanson: None. T.Y. Eng: None. N. Papanikolaou: None. A.N. Gutierrez: None.
3419 One Size Does Not Fit All: Evaluating the Need for Separate Unilateral and Bilateral Radiation Therapy Normal Tissue Treatment Planning Goals for Oropharynx Cancer L. Wang, P. Flanagan, R. Gleason, C.T. Murphy, T. Churilla, and T.J. Galloway; Fox Chase Cancer Center, Philadelphia, PA Purpose/Objective(s): Dosimetric constraints for head and neck organs at risk (OARs) are based on bilateral neck radiation treatment. Common constraints for the contralateral parotid and submandibular gland are mean <26Gy and mean <39Gy respectively, and unilateral neck plans are able to meet these goals easily. However, studies suggest that relatively low mean doses to the salivary structures (<10Gy) result in better function. We hypothesize that more stringent OAR dose limits can be placed in unilateral neck radiation therapy plans to avoid contralateral and midline neck normal tissues without compromising quality of the plan. Materials/Methods: We identified 10 patients with oropharyngeal primaries treated with unilateral radiation. The treatment plans were recreated using whole neck IMRT. Prescriptions were 70/63/56Gy in 35 fractions to gross disease, high risk neck, and low risk neck, respectively. At least 95% of each planning target volume (PTV) received prescription dose. No more than 1 cc was allowed to receive 110% and at least 99.5% of each PTV was required to receive 95% of the prescription dose. Achievable doses to the contralateral salivary structures and the supraglottic larynx/glottic larynx (GSL) were evaluated. The GSL was cropped 5mm from the PTV. The Mann-Whitney U test was used to detect differences between mean doses. Results: All patients had T1 tumors of the tonsil or glossotonsillar sulcus. There were 2 N0, 2 N1, 3 N2A, and 3 N2B patients. All gross nodal disease was in level 2/3. Target coverage was achieved in all circumstances. All plans used fixed beam IMRT with either 5 ipsilateral beams (nZ3) or 5 ipsilateral plus 1 contralateral beam (nZ7). All plans were able to achieve mean doses to the contralateral parotid and submandibular gland of <10Gy. Mean contralateral parotid dose was 7.94Gy (range 6.429.50) and mean contralateral submandibular dose was 8.94Gy (range 7.689.86). Mean dose to the GSL was 25.28Gy (range 14.33-29.78). Lymph node involvement significantly increased mean dose to the GSL (N0: 16.79 vs N+: 27.40Gy, pZ0.04) but did not significantly increase mean dose to the contralateral salivary glands: parotid (N0: 7.56 vs N+: 8.09Gy, pZ0.53), submandibular (N0:8.10 vs N+:9.15Gy, pZ0.18). Conclusion: Established OAR dose limits for contralateral salivary structures and the GSL can be further optimized for unilateral neck radiation and much greater sparing of these structures can be achieved. For patients treated with unilateral radiation, contralateral parotid and submandibular gland mean doses <10Gy and GSL <30Gy can be consistently achieved with IMRT plans without compromising tumor coverage. Separate unilateral and bilateral neck IMRT goal sheets should be used to reflect different OAR constraints. Author Disclosure: L. Wang: None. P. Flanagan: None. R. Gleason: None. C.T. Murphy: None. T. Churilla: None. T.J. Galloway: None.
3420 RayTracing and MonteCarlo Dose Calculation Might Lead to Clinically Relevant Differences for CyberKnife Lung SABR D. Dechambre,1 L. Janvary,2 N. Jansen,1 S. Cucchiaro,1 C. Ernst,1 M. Devillers,1 V. Baart,1 P. Coucke,1 and G. Akos1; 1Lie`ge University Hospital, Lie`ge, Belgium, 2Institute of Oncology, Debrecen University Clinical Center, Debrecen, Hungary Purpose/Objective(s): Various calculation algorithms could lead to significant differences in dose distribution especially at large inhomogeneities within the irradiated volume (up to 20%). Such variation could be even more important for lung stereotactic ablative radiation therapy (SABR).
E566
International Journal of Radiation Oncology Biology Physics
Historically algorithm calculation discrepancies were addressed by excluding homogeneity correction (RTOG 0236, 0618) or changes in prescription (ROSEL), lately allowing only advanced algorithms to be used (RTOG 0813, 0915). Drawn conclusions were mainly based on experience with Linac based SABR technique (max 15-20 beams), but never extensively investigated for specific robotic machines with more freedom for beam arrangements (up to 200). Our aim is to present the physical, biological dose and tumor control probability (TCP) differences between MonteCarlo (MC) and Raytracing (RT) calculation algorithms for CyberKnife lung SABR. Materials/Methods: One hundred and thirty early-stage NSCLC patients were treated by SABR. Clinical target volume (CTV) to planning target volume (PTV) margin varied between 2 and 4 mm. Intended prescription dose was 3x20 Gy with risk-adaptation when required, based on RT algorithm, followed by a retrospective MC calculation. Biologically relevant dose differences were analyzed using CTV’s mean and generalized equivalent uniform dose (gEUD, a Z -10). Physical dose differences were collected combined with the corresponding biological equivalent dose (BED10) using a method suggested previously followed by an estimation of TCP differences. Results from the two algorithms were tested using a paired t-test with a significance level of p<0.05. Results: Between 04/2010 and 05/2012, 157 lesions were treated. Focusing on physical dose differences (RT vs. MC) of the CTV, average mean dose and gEUD (SD) were 63.6 (9.5) vs. 49.9 (8.1) and 61.5 (10.1) vs. 43.3 (9.9) Gy. Median mean dose loss was about 17%. 42% and 59% of the patient population showed mean and gEUD differences higher than 20%. Corresponding biological dose differences were on average 190 (53) vs. 130 (38) and 185 (55) vs. 105 (39) Gy respectively for mean and gEUD doses. Correlation was found between BED10 loss and smaller tumor size as well as tumor’s peripheral location for both mean and gEUD. Associated TCP were on average 97 (5)% vs. 93(7)% and 96(7)% vs. 87(12)%. Median TCP losses were about 2 and 6.5% for mean and gEUD-based TCP. All of the differences were statistically significant in every aspect. Conclusion: Dose calculation differences between RT and MC algorithm could lead to severe differences in both physical (up to 60% for small, peripheral lesion) and biological doses which could dramatically worsen the TCP. Follow-up is ongoing to assess the effect of underdosage (RT based treatment) on clinical outcomes. Author Disclosure: D. Dechambre: None. L. Janvary: None. N. Jansen: None. S. Cucchiaro: None. C. Ernst: None. M. Devillers: None. V. Baart: None. P. Coucke: None. G. Akos: None.
Materials/Methods: A phenomenological model that describes proton RBE in the framework of the linear-quadratic model was proposed. In the model, effect of local energy spectrum is characterized by the dose-averaged-local-linear-energy-transfer (DALLET). The calculation of DALLET is usually a time consuming task, and therefore not well suited to clinical application. In this study, we are developing a method that can calculate the RBE within the clinically accepted time frame. We first developed and validated a Monte Carlo (MC) program in Geant4 to model the proton pencil-beam-scanning (PBS) system installed in our institute. Then the MC program is employed to generate DALLET data for selected human tissues. These DALLET data depends on tissue type, initial energy, and the actual energy at the point where the proton interacts with tissue. The chemical composition and material density of all selected human tissues are taken from ICRP pub. 23 (1975). The DALLET data are then incorporated into an in-house developed computer code to calculate variable RBE-weighted dose distribution within the patient represented by CT. Since the DALLET data are pre-calculated and bundled as a look-up-table, the RBE computation time is reduced substantially. Results: The MC validation result is first presented, which includes (1) transverse profile in air at 5 different locations along the beam central axis for 5 selected energies, (2) the nozzle water-equivalent-thickness, and (3) all 107 proton PBS beam characteristics (range, peak width and peak-toentrance ratio) in water. The DALLET data, generated by the validated MC program, is then presented. After incorporating the pre-calculated DALLET data, the in-house computer code is employed to calculate variable RBE weighted dose distributions. As the application of the fast method, a couple of clinical cases are chosen to be recalculated, and the focus is on the distal end region of PBS beam. The comparison between fixed RBE and variable RBE is presented. Conclusion: We are developing a fast method to calculate 3-D RBE distribution for proton treatment. The method is expected to perform the proton’s RBE calculation with the accuracy and computation time accepted in a busy clinical setting. Author Disclosure: X. Ding: None. W. Liu: None. M. Fatyga: None. A. Anand: None. J. Shen: None. J. Stoker: None. M. Bues: None.
3421 A Fast Method to Calculate RBE Weighted Dose Distribution X. Ding, W. Liu, M. Fatyga, A. Anand, J. Shen, J. Stoker, and M. Bues; Mayo Clinic Arizona, Phoenix, AZ Purpose/Objective(s): It is well known that proton RBE is not a constant, but depends on several factors such as dose per fraction, biological endpoint, tissue type and local energy spectrum. However at the moment, the variable RBE in clinical application is limited, due to the fact that determination of the effect of local energy spectrum is time consuming. In this study, we present a fast method to calculate 3-D RBE weighted proton dose distribution.
3422 Quantitative Radiomics: Effects of Stochastic Variability on PET Textural Features and Implications for Clinical Trials M. Nyflot,1 S.R. Bowen,2 F. Yang,2 D. Byrd,2 G.A. Sandison,2 and P.E. Kinahan2; 1Department of Radiation Oncology, University of Washington, Seattle, WA, 2University of Washington, Seattle, WA Purpose/Objective(s): Image heterogeneity metrics such as textural features are an active area of research for evaluating clinical outcomes with PET imaging; however, the effects of stochastic image acquisition noise on these metrics are poorly understood. We tested the hypothesis that the majority of textural features would have greater sensitivity to stochastic effects (as measured by coefficient of variation [COV] and sample size estimation) than the standard deviation of the intensity histogram (SDIH). Materials/Methods: 50 statistically independent PET images of the NEMA IQ phantom were simulated with realistic noise and resolution properties. The phantom incorporated lesions with diameters of 17, 22, 28,
Poster Viewing Abstracts 3422; Table 1 Estimated sample size (patients) for given clinical effect size
30% effect
Lesion size Min Q1 Median Q3 Max
37 mm 3 3 5 6 22
15% effect 28 mm 3 3 5 6 313
22 mm 3 3 6 10 587
17 mm 3 4 7 14 52
37 mm 3 4 10 13 80
28 mm 3 5 10 16 1244
22 mm 3 5 14 30 2337
17 mm 3 6 17 44 200