自动勾画软件对鼻咽癌靶区和危及器官勾画结果对比分析

摘要

目的：分析和比较两种自动勾画软件对放射治疗中鼻咽癌患者CT图像的靶区和危及器官勾画效果。方法：选取50例鼻咽癌患者的CT图像，建立一个数据库，另选20例鼻咽癌患者的CT图像应用RS和MIM两种软件进行基于以上数据库的勾画，对两种软件勾画结果进行形状相似性指数(DSC)和交叉指数(OI)分析，并对两种勾画结果的差异进行配对t检验。结果：MIM和RS软件自动勾画的危及器官的DSC和OI结果大部分在0.7～0.9之间，两种软件的DSC值大部分无明显差别，而OI的差别相对比较大；靶区自动勾画的DSC和OI均无统计学差异。结论：两种软件勾画结果大部分无统计学差异，均会在危及器官勾画上给勾画者节省时间、减少工作量，同时靶区的自动勾画也可以为自适应技术提供便利。%Objective:To validate, in the context of adaptive radiotherapy, two commercial softwares for atlas-based auto-segmentation through a comparison with manual delineation of target or OAR.Methods: A total of 70 patients were enrolled in this study(50 patients set up the atlas, and the remainder were the target image). Dice similarity coefficient(DSC)and overlap index(OI)were introduced to evaluate the auto-contours. The discrepancy between the two software was evaluated through paired t-test.Results: The DSC and OI of OAR is between 0.7-0.9. The DSC of OAR is no significant difference between the MIM 5.4(by MIMVista crop)and the RS(Raystation v3.99.0.7 by Raysearch crop); The difference of OI is relatively large. There is no statistical difference of DSC and OI in target auto-contouring. Conclusion: Although editing of the auto-contouring is inevitable, a substantial time reduction was achieved using editing, instead of manual contouring. The result of auto-contouring of the target can give some reference for adaptive radiotherapy.