中文摘要：

目的：研究基于几何不确定性的鲁棒性优化对肺癌容积旋转调强（VMAT）治疗计划的剂量分布影响以及不确定性大小对剂量分布的影响。方法：选取非小细胞肺癌病例10例,利用Raystation计划系统设计基于Minimax原理随机几何误差为4 mm的鲁棒性优化计划（P2.8）,同传统基于靶区外放的计划（Pnormal）进行剂量学比较,并分别对误差减半（P1.4）和翻倍（P5.6）两种情况考察不确定性大小对剂量学的影响。结果：P2.8和Pnormal靶区覆盖率相似,二者平均差值为-0.3 Gy,肿瘤控制概率均为78%,均无统计学意义;但P2.8肺部大于20 Gy的体积（V20）比Pnormal平均降低1.1%,平均剂量（Dmean）减少0.6 Gy,并发症发生概率平均降低0.4%,均有统计学意义;食管的Dmean则平均降低5 Gy,有统计学意义。P5.6的肺、食管及心脏的各剂量指标均增加,有统计学意义,尤其食管Dmean增加4.4 Gy,肺V5平均增加4%以上。P1.4的食管Dmean减少1.1 Gy,肺部V40减少0.4%,均具有统计学意义。P5.6和P1.4的肿瘤控制概率均为78%。结论：基于几何不确定性的鲁棒性优化降低了危及器官的照射剂量,VMAT对随机几何不确定性剂量影响较敏感。

英文摘要：

Objective To investigate the effect of robust optimization based on geometric uncertainty on the dose distribution in volumetric modulated arc therapy（VMAT） plan for lung cancer, and the effect of uncertainty on the dose distribution. Methods Ten patients with non-small cell lung cancer were selected. Based on Minimax principle, the robust optimization plan（P2.8） was designed with the random error of 4 mm using Raystation planning system, and was compared with the traditional plan（Pnormal）.The robust plans with random errors of 2 mm（P1.4） and 8 mm（P5.6） were compared with P2.8 to investigate the effect of uncertainty on the dosimetry. Results P2.8 and Pnormal plans had similar target coverage, with an average difference of-0.3 Gy, and their tumor control probabilities were both 78%, without statistical differences. Compared with those in Pnormalplan, the volume of the lung receiving more than 20 Gy in P2.8plan was averagely reduced by 1.1%, and the mean lung dose was decreased by 0.6 Gy, and the normal tissue complication probability was averagely lowered by 0.4%; mean esophageal dose was reduced by 5%. All the above-mentioned differences showed statistical significance. Compared with Pnormal plan, P5.6 plan was increased in the doses of lung, esophagus and heart, with statistical significances, especially mean esophageal dose and lung-V5,which were increased by more than 4.4 Gy and 4%, while the mean esophageal dose in P1.4 plan was reduced by 1.1 Gy, and the volume of lung receiving more than 40 Gy was lowered by 0.4%, with statistical significance. Both the tumor control probabilities of P5.6 plan and P1.4 plan were 78%. Conclusion Robust optimization based on geometric uncertainty reduced doses of the organs-at-risk and the random uncertainty had obvious dose effects on normal tissues.