摘要目的:探讨后装放疗计划的鲁棒优化方法在宫颈癌放疗中的应用价值。方法:回顾性选择已完成治疗的根治性宫颈癌患者20例，用剂量体积直方图(DVH)参数对比常规优化和鲁棒优化计划差异，用DVH和DVH束评估常规优化和鲁棒优化计划的鲁棒性。鲁棒优化方法使用最差剂量分布考虑放疗中不确定因素存在时的剂量，每次优化迭代计算放射源沿人体左右( x轴)、前后( y轴)、头脚( z轴)方向偏移固定值(2 mm)时的剂量分布，再加上放射源位置无偏移情况；每个体元的最差剂量分布是这几种情况中靶区内剂量最低值和靶区外剂量最高值，迭代目标函数通过最差剂量分布计算。 结果:在没有放射源位置偏移情况下，鲁棒优化HR-CTV D 100%均值比常规优化低，而V 150%均值比常规优化高( P<0.05)。在考虑放射源位置偏移情况下，对比多种剂量分布的最差剂量学参数，鲁棒优化和常规优化的HR-CTV D 100%均值相近，但鲁棒优化的D 90%均值比常规优化高(范围0.02～0.03 Gy)( P<0.05)。鲁棒优化增加了膀胱和小肠的D 2cm 3，直肠剂量随着鲁棒优化中考虑的放射源位置偏移方向的增加而增加。所有患者的常规优化和鲁棒优化的DVH束宽度相近。 结论:基于最差剂量分布的鲁棒优化不能明显地提升宫颈癌后装放疗计划的鲁棒性，需要其他的方式减小后装放疗中的不确定因素对剂量的影响。

abstractsObjective:To evaluate the application value of robost optimization of brachytherapy for cervical cancer.Methods:Twenty patients who completed radical treatment were recruited in this study. The dose volume histogram (DVH) parameters were statistically compared between the conventional and robust optimization plans, and the robustness between the conventional and robust optimization plans was evaluated using DVH and DVH bands. The robust optimization method utilized the worst dose distribution to consider the dose in the presence of uncertainties. In each optimization iteration, the dose distributin when the radioactive source shifted along the X, Y, and Z directions (±2 mm), and the dose distribution when the radioactive source was not shifted were calculated. The worst dose distribution for each voxel was the lowest dose in the target and the highest dose outside the target under all circumstances. The iterative objective function was calculated by the worst dose distribution.Results:In the scenario of no shifting of radioactive source position, the mean value of robust optimization was significantly lower and that of V 150% was significantly higher than those of conventional optimization (both P<0.05). When considering the shifting of radioactive source position, the worst dosimetric parameters of multiple dose distributions were statistically compared. The mean HR-CTV D 100% values did not significantly differ between the robust and conventional optimization plans, whereas the mean D 90% value (range: 0.02-0.03 Gy) of robust optimization was significantly higher than that of conventional optimization ( P<0.05). Robust optimization increased the D 2cm 3 of the bladder and small intestine, and the rectum dose was increased with the shifting of the radioactive source position in the robust optimization. The DVH bands did not significantly differ between the conventional and robust optimization plans for all patients. Conclusions:Robust optimization based on the worst dose distribution fails to significantly improve the robustness of brachytherapy for cervical cancer. Alternative methods are required to minimize the dosimetric effect of uncertainties in brachytherapy.