增强MRI纹理分析在鉴别高级别胶质瘤和单发性脑转移瘤中的价值  

The Value of Texture Analysis of Contrast-Enhanced MRI in Differential Diagnosis of High-Grade Gliomas and Solitary Brain Metastases

作　　者：孙英杰[1] 刘晓梅[1] 李光辉 周山[1] 李娅[2] SUN Ying-jie;LIU Xiao-mei;LI Guang-hui;ZHOU Shan;LI Ya(Department of Radiology,the Second Affiliated Hospital of Luohe Medical College,Luohe 462000,Henan Province,China;Department of Radiology,the Second Affiliated Hospital of Zhengzhou University,Zhengzhou 450014,Henan Province,China)

机构地区：[1]漯河医学高等专科学校第二附属医院影像科,河南漯河462000 [2]郑州大学第二附属医院影像科,河南郑州450014

出　　处：《罕少疾病杂志》2022年第7期15-17,共3页Journal of Rare and Uncommon Diseases

基　　金：漯河医学高等专科学校2019年度创新创业发展能力提升工程团队项目(2019-LYZTD012)。

摘　　要：目的探讨MRI增强图像纹理分析对鉴别脑实质内高级别胶质瘤(HGG)和单发性脑转移瘤(SBM)的临床应用价值。方法回顾性分析67例HGG和72例SBM患者的MRI增强图像,用Image J软件勾画肿瘤的感兴趣区(ROI),获取增强图像的灰度共生矩阵纹理特征参数,比较两者各特征参数之间的差异性。采用支持向量机(SVM)的结构风险最小化理论进行分类器设计,并对纹理参数进行训练和测试。结果利用灰度共生矩阵提取出的能量、对比度、相关性、逆差距等4类纹理参数在HGG和SBM的数据集之间具有统计学意义(P<0.05),并确定其为最优特征参数。应用SVM对最优特征参数组合进行测试,其鉴别HGG和SBM的敏感度88.9%、特异度93.1%、准确率91.1%。结论增强MRI纹理分析可以提供量化的信息特征,为HGG和SBM的鉴别诊断提供了新的思路和方法。Objective To investigate the value of texture analysis of MRI enhanced images in differentiating diagnosis between the high grade gliomas(HGG)and solitary brain metastases(SBM).Methods The MRI enhanced imagings data of 67 cases of HGG and 72 cases of SBM were analyzed retrospectively.The region of interest(ROI)of tumor was delineated by Image J software,the texture feature parameters of gray level cooccurrence matrix were obtained,then the differences between the feature parameters of tumors were compared.Support vector machine(SVM)structure risk minimization theory was used to design a classifier,then the texture parameters were trained and tested.Results Energy,Contrast,Correlation and Inverse Difference Moment were statistically significant between the data of HGG and SBM(P<0.05),then they were determined as the optimal feature parameters.By using SVM to test the Optimal combination of characteristic parameter,the sensitivity,specificity and accuracy of HGG and SBM were 88.9%,93.1% and 91.1%,respectively.Conclusion The analysis of texture features has certain clinical application value for the differential diagnosis of HGG and SBM.

关 键 词：高级别胶质瘤 脑转移瘤 纹理分析 

分 类 号：R445.2[医药卫生—影像医学与核医学] R739.41[医药卫生—诊断学]