用于逆向计划设计的整合优化方法  

Integrated optimization method to design inverse planning

作　　者：闵志方[1] 宋恩民[1,2] 金人超[1] 刘宏[1] 

机构地区：[1]华中科技大学计算机科学与技术学院.湖北武汉430074 [2]江西中医学院计算机学院,江西南昌330006

出　　处：《华中科技大学学报（自然科学版）》2010年第1期5-8,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家高技术研究发展计划资助项目(2006AA02Z347);广东省教育部产学研结合资助项目(2006D90304024)

摘　　要：针对调强放射治疗逆向计划设计中两步法优化算法在临床需求和硬件需求之间不能很好协调的问题,提出了一种新的整合优化方法.在强度图优化中,将传统的在目标函数添加平滑项改为在约束中添加总机器跳数约束项;另外在两个优化过程中加入反馈过程,依次求取各射野的子野序列.与目前已有的基于全变差平滑的算法在临床测试病例上进行了对比实验,结果表明,当总机器跳数和总子野个数分别为50和30时,新方法的目标值(43.195 2)远低于基于全变差平滑算法的目标值(72.154 7).A new integrated optimization method was proposed to solve the problem that the hardware requirements and the clinical needs are not well coordinated for the two steps optimization in intensity modulated radiation therapy （IMRT） inverse planning. First, the total monitor unit constraints are added directly into the fluence map optimization rather than being added into the smoothing objective function. Second, a feedback process is added into these two step method to get the segmentation of each beam, which is carried out alternately with the fluence map optimization. The new method and the existing smoothing strategy based on total variance were both tested on clinical test case. The re sults show that the new objective value of the new method（43. 195 2） is much lower than that of hte total variance smoothing method （72. 154 7） while the total number of moniter units and the total number of segments are 50 and 30 respectively.

关 键 词：计算机辅助设计 逆向问题 计划设计 二次规划 约束优化 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]