Radiation treatment planning, which involves decisions about irradiation parameters based on the optimal beam placement and irradiation amount, is an essential process that affects the doses delivered to the tumor and the surrounding normal tissues. However, the planning process is challenging: it consumes a lot of time and energy and the quality depends largely on the expertise of the planner. Predicting feasible treatment plans for individual patients is expected to improve the planning efficiency and eliminate quality disparities between different plans.

Our laboratory studies a method for predicting treatment plans to estimate the doses delivered to the tumor and the surrounding normal tissue. We estimate doses by performing the modeling of a dose-volume histogram (DVH), which relates radiation dose to tissue volume, applying functions and then determining the irradiation parameters from the training data. We are endeavoring to improve the accuracy of DVH predictions by taking the dose distribution characteristics of proton beam therapy into account during the modeling and parameter determination processes.