中文摘要：

目的：以统计图表和参数为依据,分析处理瓦里安Rapid Plan模型离群值,并比较处理前后计划之间及其与人工优化的剂量学差异。方法：（1）选取80例直肠癌术前同步推量计划建立模型;（2）修正明显原因所致的离群值;（3）利用拟合曲线、Z值和库克距离寻找几何离群值,检查原计划并分情况处理;（4）利用残差分布图、剂量体积直方图、学生化残差值和库克距离判断剂量离群值并酌情处理;（5）利用20例同类计划测试验证前后模型的优化效果。结果：统计学验证使模型构成计划中的离群参数最大值显著降低（P〈0.05）。相比原人工优化的测试计划,Rapid Plan使用任一模型均能显著改善靶区均匀性指数（P〈0.05,幅度2.7%~19.0%）和膀胱平均剂量（P〈0.05,降幅12.8%~13.2%）。但利用验证前后模型分别优化的计划质量差异不大：肿瘤计划靶区（PGTV）均匀性指数相差0.5%,PGTV适形指数相差0.1%,计划靶区适形指数相差0.5%,股骨头和膀胱平均剂量分别相差0.3%和0.4%（P〉0.05）。结论：基于知识的Rapid Plan计划可以改善靶区剂量均匀性和保护危及器官。依据统计学参数删改模型构成计划并不一定能取得更好的剂量学效果。

英文摘要：

Objective Based on the statistical diagram and parameter, to analyze and process the outliers in Varian Rapid Plan model, and to compare dosimetric differences before and after processing, and to compare the dosimetric differences between model optimization with the manual optimization. Methods Totally, 80 preoperative simultaneous integrated boosting plans for rectal cancer were selected to construct the model. The outliers caused by obvious reasons were firstly corrected. The geometric outliers were detected by using fitting curves, Z- score value and Cook＇s distance（CD）, and the original plan were tested and processed according to different situations. The dosimetric outliers were judged and accordingly processed based on the residual plots, dose- volume histogram, studentized residual and CD. The optimization results of model before and after statistical verification were tested by 20 similar plans. Results The maximum outlier indicators were significantly reduced by statistical verification（P0.05）. Compared with the original manual optimization plan, any model of Rapid Plan significantly improved the homogeneity index（HI） of target volume（P0.05, with a range of2.7%- 19.0%） and reduced the mean dose of urinary bladder（P0.05, with a range of 12.8%- 13.2%）. No significant differences were found in the quality of plans optimized by model before and after verification. The differences in the HI of planning gross target volume（PGTV）, conformity index（CI） of PGTV, CI of planning target volume（PTV）, mean dose of femoral head and urinary bladder were respectively 0.5%, 0.1%, 0.5%, 0.3% and 0.4%（P0.05）. Conclusion The knowledge- based Rapid Plan improves the HI of target volume and the protection for organs at risk. The adjustment of model components based on the statistical indicators does not guarantee better dosimetric consequences.