美国数学建模竞赛特等奖论文——2003 B O The Gamma Knife Problem【数学建模】.pdfVIP

The Gamma Knife Problem 351 The Gamma Knife Problem Darin W. Gillis David R. Lindstone Aaron T. Windfield University of Colorado Boulder, CO Advisor: Anne M. Dougherty Abstract Noninvasive gamma-knife radiosurgery treatment attacks brain tumors using spherical radiation dosages (shots). We develop methods to design optimized treatment plans using four fixed-diameter dosages. Our algorithms strictly adhere to the following rule: Shots cannot violate tumor boundaries or overlap each other. From a mathematical perspective, the problem becomes a matter of filling an irregularly-shaped target volume with a conglomeration of spheres. We make no assumptions about the size and shape of the tumor; by maintaining complete generality, our algorithms are flexible and robust. The basic strategies of the algorithms are deepest-sphere placement, steepest descent, and adaptation. We design representative 3D models to test our algorithms. We find that the most efficient packing strategy is an adaptive algorithm that uses steepest descent, with an average coverage percentage of 40% over 100 test cases while not threatening healthy tissue. One variation covered 56% of one test case but had a large standard deviation across 100 test cases. It also produced results four times as fast as the adaptive method. Background Brain Tumors The average volume of a tumor operable by radiosurgery is about 15 cm3 [Lee et al. 2002]. We generate 3D tumor models of approximately this volume with varying physical dimensions. The Gamma Knife The gamma knife unit consists of 201 individual cobalt-60 radiation sources situated in a helmet. The 201 beams converge at an isocenter creating a spherical c The UMAP Journal 24 (3) (2003) 351–363. Copyright 2003 by COMAP, Inc. All rights