Knowledge Based Planning for Retrospective Quality Study of RTOG 0822 Radiotherapy Plans

Thumbnail Image
Penn collection
Master of Science in Medical Physics Thesis Projects
Degree type
medical physics
radiation oncology
Medicine and Health Sciences
Grant number
Copyright date
Related resources

Purpose: To create a clinically viable dose-volume histogram (DVH) estimation model using the Varian RapidPlan (Varian Medical Systems, Palo Alto USA) knowledge based planning (KBP) platform. This model aims to evaluate locally advanced rectal cancer with 6X IMRT, and was developed on a plan database taken from the RTOG 0822 national clinical trial. This is the first multi-institutional 6X IMRT dose estimation model designed using RapidPlan. The effectiveness of the model as a dosimetry quality assurance (QA) tool was evaluated. Methods: Treatment plans submitted to the RTOG 0822 clinical trial were dosimetrically evaluated for plan quality. Plans whose DVH statistics met RTOG 0822 target criteria were identified as high-quality, and were used in the initial training sample for the model. Of the 97 IMRT plans enrolled in the trial, 58 were treated with only 6X photons, and 26 of those were identified as high-quality plans. All 6X enrolled plans were iteratively re-optimized with the model to test clinical effectiveness, evaluate the model as a tool for treatment planning QA, and continuously expand the model’s training sample. Re-optimized plans which met target criteria were added to the training sample, resulting in a total of 40 geometries in the training sample. Results: The rectal IMRT RapidPlan model created in this paper was shown to accurately predict estimated DVH bands for all viable high-quality plans enrolled in the clinical trial. The model was also able improve the DVH statistics for a significant majority of the low-quality plans enrolled in the clinical trial. Conclusion: The RapidPlan rectal 6X IMRT model created in this study can be used as an effective tool for dosimetry QA and initial plan creation.

Date Range for Data Collection (Start Date)
Date Range for Data Collection (End Date)
Digital Object Identifier
Series name and number
Publication date
Volume number
Issue number
Publisher DOI
Journal Issue
Recommended citation