Gaussian Process Based Water Equivalent Thickness Mapping (GPP-WET-MAP): An Alternative to 4D Robust Optimization in Proton Therapy

Currently, proton radiotherapy accounts for target motion with 4D optimization, which robustly optimizes the treatment plan using setup uncertainty in multiple geometric directions combined with density uncertainty on the treatment planning CT (TPCT) and additional images that represent organ motion scenarios. Optimizing on the additional images is slow, inefficient, and increases the size of the dose cloud, raising coverage but worsening organ at risk (OAR) sparing. For these reasons, this study aims to develop Gaussian Process Prediction Water Equivalent Thickness Map (GPP-WET-MAP), a planning strategy that maps effects of beam path changes from patient-specific motion margins to contour changes on the TPCT.

This work will apply GPP-WET-MAP (GWM) to prostate patients. Prostate inter-fraction motion can be addressed with 4D optimization using simulated organ motion (SOM) images that represent prostate motion scenarios. GP regression models will be trained on patient population data to predict patient-specific prostate inter-fraction motion distributions, which will be used to calculate motion bounds. GWM plans account for predicted inter-fraction motion with WET mapping, are optimized on just the TPCT, increasing efficiency, and use a smaller setup uncertainty. The goals for this work are creating GWM plans for normal fractionated and stereotactic body radiotherapy (SBRT) prostate patients, comparing them to 4D optimized plans, and evaluating GWM’s clinical viability.