美国数学建模竞赛特等奖论文——2003 B O The Genetic Algorithm-Based Optimization Approach for Gamma Unit Treatment【数学建模】.pdfVIP

Genetic Algorithm-Based Optimization Approach 323 The Genetic Algorithm-Based Optimization Approach for Gamma Unit Treatment Sun Fei Yang Lin Wang Hong Donghua University Shanghai, China Advisor: Ding Yongsheng Abstract The gamma knife is used to treat brain tumors with radiation. The treatment planning process determines where to center the shots, how long to expose them, and what size focusing helmets should be used, to cover the target with sufficient dosage without overdosing normal tissue or surrounding sensitive structures. We formulate the optimal treatment planning for a gamma-knife unit as a sphere-packing problem and propose a new genetic algorithm (GA)-based op- timization approach for it. Considering the physical limitations and biological uncertainties involved, we outline a reasonable, efficient and robust solution. First, we design a geometry-based heuristic to produce quickly a reasonable configuration of shot sizes, locations, and number. We first generate the skeleton using a 3D-skeleton algorithm. Then, along the skeleton, we use the GA-based shot placement algorithm to find a best location to place a shot. By continuously iterating the algorithm, we obtain the number, sizes, and the locations of all shots. After that, we develop a dose-based optimization method. Then we implement simulations of our models in Matlab. We did numerous computer simulations, using different shape or size targets, to examine the effec- tiveness of our model. From the simulation results, we know that the geometry- based heuristic with the GA optimization approach is a useful tool for in the selection of the appropriate number of shots and helmet sizes. Generally, all of the optimized plans for various targets provide full-target coverage