常规CT纹理分析鉴别孤立性肺结节的价值初探  

The value of unenhanced CT texture analysis in the differentiation of solitary pulmonary nodules:a preliminary study

作　　者：吴艳 谢元亮[1] 王翔 马锋 李友 刘子豪 WU Yan;XIE Yuanliang;WANG Xiang;MA Feng;LI You;LIU Zihao(Department of Radiology, the Central Hospital of Wuhan, Tongji Medical College,Huazhong University of Science and Technology, Wuhan 430014,China)

机构地区：[1]华中科技大学同济医学院附属武汉中心医院影像科,湖北武汉430014

出　　处：《东南大学学报（医学版）》2020年第2期169-174,共6页Journal of Southeast University(Medical Science Edition)

摘　　要：目的:探讨基于CT平扫的纹理分析技术鉴别孤立性肺结节(SPNs)的价值。方法:回顾性分析经病理证实的138例SPNs患者的资料。其中恶性组89例,良性组49例。采用MaZda软件手动描绘结节感兴趣区(ROI)并提取其纹理特征,分别通过费希尔系数、分类错误概率联合平均相关系数、交互信息及上述3种方法联合(FPM)来选取最佳纹理参数集合。运用机器学习(主要成分分析、线性判别分析及K最邻近分类算法)及人工智能(非线性判别分析、人工神经网络)的方法对纹理特征进行分类,结果以错判率的形式表示。结果:良恶性SPNs组间鉴别FPM联合人工神经网络错判率最低(为11.59%);恶性SPNs组内鉴别FPM联合人工神经网络错判率最低(为5.62%);良性SPNs组内鉴别FPM联合线性判别分析错判率最低(为0)。结论:常规CT纹理分析对鉴别SPNs具有一定价值。Objective:To investigate the value of texture analysis derived from conventional CT imaging in differentiating solitary pulmonary nodules.Methods:One hundred and thirty-eight patients with solitary pulmonary nodules confirmed by pathology were enrolled in this retrospective study,of whom 89 cases were in malignant group and 49 cases in benign group.Texture features were calculated from manually drawn ROIs by using MaZda software.The feature selection methods included Fishers coefficient,classification error probability combined with average correlation coefficients(PA),mutual information(MI)and the combination of the above three(Fishers+PA+MI,FPM).Machine learning(principal component analysis,linear discriminant analysis,K nearest neighbor classification)and artificial intelligence(nonlinear discriminant analysis,artificial neural networks)were performed for texture classification.The results were shown by misclassification rate.Results:In the differentiation between benign and malignant nodules,the misclassification rate of FPM combined with artificial neural networks was the lowest(11.59%).In the differentiation of malignant nodules,the misclassification rate of FPM combined with artificial neural networks was the lowest(5.62%).In the differentiation of benign nodules,the misclassification rate of FPM combined with linear discriminant analysis was the lowest(0).Conclusion:Texture analysis of conventional CT imaging is valuable for the differentiation of solitary pulmonary nodules.

关 键 词：孤立性肺结节 计算机体层摄影术 纹理分析 影像组学 

分 类 号：R734.2[医药卫生—肿瘤] R814.42[医药卫生—临床医学]