Room: AAPM ePoster Library
Purpose: Prostate bed with nodal involvement cases are often challenging to plan due to more overlap between PTV and organs at risk (OAR) volumes. We investigate the performance of Varian’s Rapidplan for 3 difficult cases where PTV overlaps both rectum and bladder by >40%.
Methods: Treatment planning system used is Eclipse version 15.6. Model Configuration was used to build a library of training plans consisting 57 prostate bed patients previously treated with Volumetric Modulated Arc Therapy (VMAT). Plans were generated between 2017-2019 and screened for Model Configuration DVH estimation criteria such as regression plot outliers, outlier statistics and dosimetric quality. To test our Rapidplan model’s ability to handle overlaps, three dosimetrically-challenging patients with similar overlap volumes by three separate planners were identified (Bladder/PTV overlap of 42%, 51% 54% and Rectum/PTV overlap of 51%, 57%, 41%). These patients were retrospectively re-planned using Rapidplan and results compared with original plans.
Results: Patient A – manual plan achieved higher PTV V45 (98.4% vs 96.9%). Rectal wall V40 and V30 was higher for manual plan (70.8%, 80.4%) than Rapidplan (64.7%, 76.8%). Small bowel V48 was also higher in manual plan (1cc) than Rapidplan (0.03cc). Patient B – Rapidplan had slightly better PTV V45 (93.1% vs 91.8%). Rectal wall V40 and V30 was higher for Rapidplan (40.0%, 56.2%) than manual plan (35.4%, 53.8%). Patient C – Rapidplan and manual plan achieved similar PTV V45 (96.6% vs 96.5%). Rectal wall V40 and V30 was higher for Rapidplan (59.7%, 71.0%) than manual plan (54.8%, 66.9%).
Conclusion: In all 3 cases, compared with manual planning, Rapidplan achieved similar PTV coverage and OAR sparing. Manual planning had slightly better dose statistics overall – however, this usually required twice the planning time. Rapidplan results could be improved further as more challenging cases are added to modelling library in the future.
VMAT, Knowledge Based Planning, Rapidplan
TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation