基于混合遗传算法的近距离放射剂量仿真优化  

Dose Optimization Algorithm in Brachytherapy Based on Hybrid Genetic Algorithm

作　　者：闵小平[1] 王博亮[1] 潘建基[2] 戴培山[3] 鞠颖[1] 

机构地区：[1]厦门大学计算机系,厦门361005 [2]福建省肿瘤医院,福州350003 [3]国防科技大学电子科学与技术系,长沙410083

出　　处：《系统仿真学报》2008年第12期3270-3273,共4页Journal of System Simulation

基　　金：国家自然科学基金(60371012);福建省科技重点项目(2004Y008)

摘　　要：提出了一种连续驻留位置和驻留时间的近距离放疗剂量优化算法。将驻留时间t看成是驻留位置x的一个连续函数,用积分方式计算每个参考剂量点的剂量。再用计算剂量和目标剂量值之间的差别为目标函数,用遗传算法和POWELL算法混合求解最优的曲线参数。曲线可根据具体情况进行分段。得到曲线后再根据积分的数值逼近方法原理将其离散化,得到最终的驻留位置和驻留时间。实验的结果表明,算法不仅避免了负的驻留时间问题,还让相邻驻留位置的驻留时间比较平滑。在最后的离散化过程中,还可以得到不同的驻留位置和驻留时间结果,使计划具有更好的灵活性。A dose optimization algorithm based on Hybrid Genetic Algorithm was proposed to optimize the dwell time in high dose brachytherapy. Firstly the dwell time was considered as a continuous function over the dwell position, the dose values of referenced dose points was computed by integral algorithm. The minimal difference between these dose values and demanded dose values was used as the objective of Hybrid Genetic optimization algorithm to get the optimized function parameters. The function curve could be separated to several sects. At last, the optimized curve was dispersed using compound trapezoid formula to get the dwell positions and dwell times. The experiments show this method not only avoids the emergence of negative dwell time, but also reduces the dwell time gradient. And the dwell times and dwell positions were not fixed. This makes the brachytherapy treatment plan more flexible.

关 键 词：近距离放疗 剂量优化 混合遗传算法 积分模型 

分 类 号：R815[医药卫生—放射医学]