摘要
目的探讨动态增强MRI三维纹理分析鉴别透明细胞肾细胞癌(clear cell renalcell carcinoma,ccRCC)、乳头状肾细胞癌(papillary renal cell carcinoma,pRCC)和嫌色肾细胞癌(chromophobe renal cell carcinoma,ChRCC)的价值。材料与方法回顾性分析经病理证实的74例ccRCC、22例pRCC和17例ChRCC患者资料,分别提取肾皮髓质期、实质期及延迟期增强T1WI上肿瘤三维纹理特征,筛选后应用原始数据分析、主成分分析、线性判别分析和非线性判别分析方法进行分类分析,计算诊断准确率、敏感度、特异度及受试者工作特征曲线下面积(area under the curve,AUC)。结果动态增强MRI三维纹理分析鉴别ccRCC与pRCC的诊断准确率、敏感度、特异度达88.54%、91.89%、77.27%(AUC=0.846);鉴别ccRCC与ChRCC的诊断准确率、敏感度、特异度达95.60%、97.30%、88.24%(AUC=0.928);鉴别pRCC与ChRCC的诊断准确率、敏感度、特异度达79.49%、72.73%、88.24%(AUC=0.805)。非线性判别分析方法均获得较高的诊断效能(AUC 0.707~0.928)。结论 DCE-MRI三维纹理分析有助于鉴别ccRCC、pRCC和ChRCC。
Objective:To evaluate the diagnostic performance of three-dimensional (3D) texture analysis (TA) in dynamic contrast-enhanced (DCE) MRI for the classification of clear cell (ccRCC),papillary (pRCC) and chromophobe renal cell carcinoma (ChRCC).Materials and Methods:A retrospective review was performed on patients with ccRCCs (n=74),pRCCs (n=22) or ChRCCs (n=17) confirmed by pathology.Corticomedullary phase,nephrographic phase and delayed phase CE-MR images obtained from all the patients were used for texture analysis.314 3D texture features were extracted from each of the three image series,and 30 important features were selected separately for each pair of ccRCCs,pRCCs and ChRCCs.Texture analysis was performed using raw data analysis,principle component analysis and linear discriminant analysis,and nonlinear discriminant analysis.Classification accuracy,sensitivity,specificity and area under the receiver operator characteristics curve (AUC) were calculated.Results:For ccRCC vs pRCC,the classification accuracy,sensitivity and specificity of 3D TA in DCE-MRI were up to 88.54%,91.89% and 77.27%(AUC=0.846),for ccRCC vs ChRCC,the classification accuracy,sensitivity and specificity were up to 95.60%,97.30% and 88.24%(AUC=0.928),for pRCC vs ChRCC,the classification accuracy,sensitivity and specificity were up to 79.49%,72.73% and 88.24%(AUC=0.805).For all the pairs of ccRCCs,pRCCs and ChRCCs,classification performed the best in nonlinear discriminant analysis (AUC 0.707-0.928) within each of the three image series.Conclusions:The three-dimensional texture analysis in DCE-MRI can be a reliable quantitative approach for differentiating ccRCC from pRCC or ChRCC and pRCC from ChRCC.
作者
周智
陈杰
潘靓
周菲菲
邢伟
ZHOU Zhi;CHEN Jie;PAN Liang;ZHOU Feifei;XING Wei(Department of Radiology,No.904 Hospital of Joint Logistics Unit,Changzhou 213001,China;Department of Radiology,the Third Affiliated Hospital of Soochow University,Changzhou 213001,China)
出处
《磁共振成像》
CAS
2019年第7期525-529,共5页
Chinese Journal of Magnetic Resonance Imaging
基金
国家自然科学基金面上项目(编号:81771798)
常州市科技计划资助项目(编号:CJ20179021)~~
关键词
肾肿瘤
磁共振成像
纹理分析
kidney neoplasms
magnetic resonance imaging
texture analysis
