Optimization of Gamma Knife treatment planning via guided evolutionary simulated annealing

Abstract

We present a method for generating optimized Gamma Knife (Elekta, Stockholm, Sweden) radiosurgery treatment plans. This semiautomatic method produces a highly conformal shot packing plan for the irradiation of an intracranial tumor. We simulate optimal treatment planning criteria with a probability function that is linked to every voxel in a volumetric (MR or CT) region of interest. This sigmoidal P+ parameter models the requirement of conformality (i.e., tumor ablation and normal tissue sparing). After determination of initial radiosurgery treatment parameters, a guided evolutionary simulated annealing (GESA) algorithm is used to find the optimal size, position, and weight for each shot. The three-dimensional GESA algorithm searches the shot parameter space more thoroughly than is possible during manual shot packing and provides one plan that is suitable to the treatment criteria of the attending neurosurgeon and radiation oncologist. The result is a more conformal plan, which also reduces redundancy, and saves treatment administration time.