Encrypted login | home

Program Information

Evaluation of a RapidPlanTM Model Trained with Mulit-Criteria Optimised Test Cases for Knowledge Based Planning of Pancreatic Radiotherapy Plans

no image available
P Houston

P Houston*, G Currie , NHS Greater Glasgow and Clyde


SU-I-GPD-T-330 (Sunday, July 30, 2017) 3:00 PM - 6:00 PM Room: Exhibit Hall

Purpose: To evaluate the effectiveness of a RapidPlanᵀᴹ [Varian Medical Systems] model trained with Multi-Criteria Optimised (MCO) test cases in producing highly optimized treatment plans for pancreatic planning.

Methods: 53 pancreas patients were replanned with a pre-clinical release of MCO in Eclipse to create highly optimised plans. These plans were then used to train a RapidPlanᵀᴹ model (RP_MCO). 10 separate pancreas cases were used as validation cases. The RP_MCO model was compared against a clinical pancreas RapidPlanᵀᴹ model (RP_Clinical), trained on manually optimised plans, by an experienced planner, from the same patient cohort.All patients were prescribed 50.4Gy to the target mean in 28 fractions. Plans were compared against clinical plan constraints, derived from national standards.

Results: Achieved PTV coverage was comparable between the two models, with no statistical difference in D98%, D2%, DMax or Paddick Conformity Index.The most significant organ sparing was achieved in the ipsilateral kidney when planning with RP_MCO, with an average reduction in V20Gy of 7.1% (Range:2.4%-13.9%; p=0.00009))and an additional reduction in mean dose of on average 5.0Gy (Range:0.6Gy-8.1Gy; p=0.0005).When planning with RP_MCO, the mean liver dose was reduced, in 7 of 10 cases and equivalent in 3 others, by an average of 5% (Range:0%-15%; p=0.005) in comparison to RP_ClinicalMaximum dose to spinal cord was reduced in 9 cases with RP_MCO by an average of 3.6Gy (Range:-0.6Gy-9.6Gy; p=0.003).The dose to bowel and stomach was not significantly altered by planning with either RapidPlanᵀᴹ model. Calculated MU was larger when planned RP_MCO in 9 of 10 cases and equivalent in another but not statistically significant.

Conclusion: Incorporating MCO plans into a RapidPlanᵀᴹ model can achieve significant reductions in dose to organs at risk, whilst maintaining PTV coverage in the pancreas setting in comparison to a model trained on manually optimised plans.

Contact Email: