A memetic algorithm for body gamma knife stereotactic radiotherapy treatment planning

Body gamma knife stereotactic radiotherapy (SBRT) is an effective treatment for non-small-cell lung cancer (NSCLC). Treatment planning is both tedious and timeconsuming due to its complexity. In order to address this issue, we present a memetic optimization algorithm for body gamma knife SBRT treatment planning. Different from currently available methods, the memetic approach performs the optimization on relative dosimetric distribution, which results in an enlarged search space. Taking advantage of the prior knowledge about the interaction between the focuses, a heuristic initialization strategy and a set of genetic operators are developed. During optimization, the most time consuming part is accelerated by graphic processing unit (GPU). Three typical targets arising from real patient data are used to test the efficiency of this approach. Experimental results demonstrate that the memetic optimization could generate a feasible treatment plan quickly (94.6 sec for the largest case).