Knowledge-Based Planning for Robustly-Optimized Intensity Modulated Proton Therapy of Prostate Cancer
Y Xu1*, N Brovold2, E Bossart1, K Padgett1, M Butkus1, T Diwanji1, A King1, A Dal Pra1, M Abramowitz1, A Pollack1, N Dogan1, (1) University of Miami, Miami, FL (2) Henry Ford Health System, Southfield, MI
Presentations
PO-GePV-T-117 (Sunday, 7/25/2021) [Eastern Time (GMT-4)]
Purpose: To assess the performance of a Proton-specific Knowledge Based Planning (KBP) model in creation of robustly optimized IMPT plans for treatment of prostate cancer patients.
Methods: Forty localized prostate cancer patients previously treated with volumetric modulated arc therapy (VMAT) were included. All the patients were replanned with robustly optimized Intensity Modulated Proton Therapy (IMPT). Thirty prostate patients were included in creation of proton KBP model. The remaining ten patients were used for model validation. KBP model quality and accuracy was evaluated based on the model-provided organ at risk (OAR) regression plots and metrics. To compare the robustness of the IMPT plans, we simulated the ±3% proton range uncertainty with 5mm translational error in six directions, resulting in twelve uncertainty scenarios for each plan. The expert and KBP generated IMPT plans were compared using the clinical dose volume constraints for clinical target volume (CTV), bladder, and rectum.
Results: The resulting R² (0.87± 0.07) between dosimetric and geometric features, as well as χ² (1.17± 0.07) between the original and estimated data, showed good quality of the model. All the KBP plans were clinically acceptable. Looking only at the nominal plans, KBP plans delivered marginally higher dose to the rectum (0.42± 0.56% for rectum V65Gy) than the expert plan, but otherwise there was no statistically significant difference in dose-volume indices for CTV and bladder. When all uncertainty scenarios are included, the KBP plans showed much greater CTV V100 (11.64± 20.09%) and lower HI (-1.81± 1.91) than the expert plan did.
Conclusion: This study demonstrated that the robustly optimized IMPT plans created by a proton KBP model are of high quality and are at least comparable to the expert plans. Furthermore, the KBP model are able to generate more robust and homogenous plans compared to the expert plans.
ePosters
Keywords
Taxonomy
Contact Email