Layout optimization of corrector magnets for a proton therapy beamline via genetic algorithm

Abstract

Proton therapy is one of the most effective radiotherapy methods for cancer and the dose distribution of protons on patients is superior to conventional rays. However, it is likewise more sensitive to the beam misalignment raised from errors in the magnets, girders and beam diagnostic devices. According to the clinical treatment guidelines, beam position accuracy at the iso-center should be controlled within ±0.5 mm. While the response matrix method is widely used in beam orbit correction, the correction performance relies largely on the locations of corrector magnets. This paper introduces an optimization approach for the layout of corrector magnets by using the multi-objective genetic algorithm NSGA-II and this technique is applied to a cyclotron-based proton therapy facility. All potential locations of the corrector magnets are encoded with an integer sequence. After 300 generations of crossover, mutation and selection, seventeen Pareto-front solutions were obtained and the best configuration was selected by taking the overall beam misalignment along the beamline into considerations. Finally, the maximum deviation of the corrected beam orbit in the beamline is less than 4.5 mm and the beam position at the iso-center can be optimized within ±0.3 mm, fulfilling the clinical requirement.