(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: evaluate the planning efficiency and dosimetric quality improvement of knowledge-based planning models in the treatment of GI specific sites.
Methods: models were created using Varian’s RapidPlan™ for rectum, anal, pancreas, and liver site-specific planning. For each model greater than 70 prior patients were selected for model training. Models include both IMRT and VMAT techniques in training plans with a focus on institutional constraints for abdominal and pelvic critical structures. Each model was validated for PTV coverage (D95% and D2%) as well as dose-volume constraints to pelvic and abdominal critical structures relevant to each site. Planning efficiency was also studied by selecting 5 patients from each site and evaluating the number of iterations required for optimization and the active planning time required for model based planning and traditional manual planning.
Results: all sites, the model generated clinically acceptable plans meeting all planning constraints. PTV D95% improved across all sites with a minimal change in PTV D2% demonstrating increased homogeneity in target coverage. Dose to most critical structures remained consistent with manually optimized clinical plans. Notable changes were seen for the femoral heads with a 2-4Gy reduction in dose across D50%, D35%, and D5% and the spinal cord with up to a 5Gy max dose reduction for upper GI sites. Overall, plan quality remained consistent for all sites but planning efficiency was improved. RapidPlan optimization required 1 optimization (1050±287.6 sec) while manual planning required 2-5 optimizations (1941.4±996.9 sec) reducing the average active planning time by approximately half.
Conclusion: knowledge-based planning implementation through RapidPlan has been shown here to create clinically acceptable plans while maintaining plan quality seen in prior clinical plans for multiple GI sites. The planning efficiency is also significantly increased and the variation in active planning time is reduced when using knowledge-based models.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
