Room: AAPM ePoster Library
Purpose: To calculate the impact of knowledge-based planning and multicriteria optimization (MCO) on efficiency in treatment planning workflows.
Methods: Four custom RapidPlan® models were created utilizing historical plans in a network of centers. The focus was on generalized prostate, lung, hippocampal avoidance whole brain, and head and neck sites. Specialized data mining tools were developed to locate the treatment plans that were candidates for the model. The models were adjusted until most optimizations with the applied DVH Estimations gave clinically acceptable solutions after one full run. In addition, Clinical Protocols in Eclipse were created for these sites to aid in the further optimization with MCO trade-off analysis. Plan quality metrics were analyzed by DVH analysis to our site’s protocols. The new treatment planning workflows were tracked by analyzing the Care Paths in ARIA compared to plans optimized without RapidPlan, MCO, or both.
Results: The impact of RapidPlan and MCO on the treatment planning workflow was significant. On average, the turnaround on the planning task in the Care Path was reduced on all cases evaluated. In some circumstances, plans were finalized in optimization, calculated, and sent for physics review in under an hour. In addition, the plans that utilized the new workflow had better overall plan quality metrics than cases planned without RapidPlan or MCO.
Conclusions: With custom models in RapidPlan and trade-off explorations in MCO, there can be a large clinical benefit in workflow and plan quality. There is some skill required to build the models, and it is highly recommended to have the dosimetrists involved with the physicists during the building and training of the RapidPlan models. Finally, further gains can be obtained with the Eclipse Scripting API, which exposes both RapidPlan and MCO for automation.
Treatment Planning, Optimization
TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation