遗传算法求解伽马刀治疗方案

Genetic Algorithm for Gamma Knife Treatment

伽马刀是治疗脑瘤的一种常用手段 .2 0 1束光束会聚形成球状的斑 ,其 5 0 %等剂量线以内是有伤杀力的 ,于是问题转化为求解一些不同大小球的堆积问题 .用遗传算法来求解伽马刀优化方案 ,用有效杀伤率 ,肿瘤轮廓与等剂面的符合情况 ,重叠辐射程度 ,总照射剂量等多项指标来衡量方案的优化程度 .先分别求出针对每一种指标单目标优化方案 ,然后再加权设计整体目标函数来求解 .为了能清楚地表示 ,在二维平面上对肿瘤进行了测试 ,可以达到 97%以上的杀伤率 .

Gamma knife treatment is a common therapy for the curing of brain tumor. 201 beams simultaneously intersect at the same location in the space. The spot within 50% isodose is its efficient “killing space”. Thus, the problem is transformed to the packing problem of some different-radius-spheres. We use the genetic algorithm to solve such a gamma knife treatment planning problem, considering the discrepancy among our contour and target volume, the overlapping volume, the integral dose to the brain, etc. , as our objectives to value our scheme. During the process, we start with seeking for optimal solutions to each individual objective respectively. After that, taking account of objectives of various kinds, we figure out a finale comprehensive solution by adding weight to each objective, and as a consequence, the results can satisfy most objectives. Later on, we test our model in 2-D space by working it on tumors in a 2-D plane and find that the execution can attain as much as 97% of the target volume.