LP formulations for optimizing radiation treatment strategies

The goal of radiation therapy is to destroy the tumor with no side-effects. The recently developed technique of intensity modulated radiation therapy (IMRT) uses spatially nonuniform intensities. While this allows more flexibility and better treatments, it also increases the search space and hence requires optimization. We demonstrate the success of linear programming formulations in quickly finding acceptable plans. We believe this success stems from the natural way the constraints correspond to the physician's goals. In addition we present a novel objective that has a more natural spatial dependence than conventional penalty function formulations.