基于改良遗传算法的调强射野角度优化

Improved genetic algorithm in optimization of beam orientation in intensity modulated radiotherapy

目的基于调强放疗计划凭经验手动的现状探讨自动调强射野角度优化方法的可行性。方法用基于射野方向观视剂量测量（BEVD）的遗传算法（GA）射野方向的优化方法对临床脊柱肿瘤和肺肿瘤进行5个野等分隔、5个野GA射野优化方向、5个野基于BEVD的GA射野方向优化的调强放疗计划进行比较。结果由于该技术先对GA搜索范围进行限制，然后再进行遗传操作，不但减少了优化时间，而且提高了优化效果。对临床脊柱肿瘤和肺肿瘤进行5个野基于BEVD的GA射野方向优化的调强放疗计划方法最优。结论采用调强射野角度优化方法是可行的，避免了手工多次调整射野方向，提高了效率和效果。

Objective At present beam orientation selection in intensity-modulated radiotherapy （IMRT） is mainly based on empiric knowledge. This study is to evaluate the feasibility of automated beam angle selection. Methods Genetic algorithm technique which based on beam eye view dose measurement （BEVD-GA） was tested on two clinical cases, including a spine column cancer and a lung cancer. Three plans were obtained under the following different beam configurations：five equiangular-spaced beams, five beams with GA-seleeted, and five beams with BEVD-GA-selected beams. Then the dose distribution was compared among the three plans. Results The method, restricting the range of genetic algorithm followed by carrying through genetic operations, not only shortened the optimization time, but also improved the optimization effect. For spine column cancer and lung cancer, the best IMRT plans were obtained with BEVDGA-selected beams, which used automated beam orientation selection. Conclusions Comparing with the conventional manual beam orientation selection, beam orientation optimization which is feasible in IMRT planning may significantly improve the efficiency and result.