血氧水平依赖MRI纹理分析预测肾透明细胞癌病理分级的价值  

The value of blood oxygen level dependent MRI texture analysis in predicting the pathological grade of clear cell renal cell carcinoma

作　　者：周菲菲 周智 潘靓[1] 邢伟[1] 陈杰[1] ZHOU Feifei;ZHOU Zhi;PAN Liang;XING Wei;CHEN Jie(Department of Radiology,the Third Affiliated Hospital of Soochmv University,Changzhou,Jiangsu Province 213003,China;Department of Radiology,No.904 Hospital of Jonit,Changzhou,Jiangsu Province 213003,China)

机构地区：[1]苏州大学附属第三医院影像科,江苏常州213003 [2]联勤保障部队第904医院常州医疗区放射诊断科,江苏常州213003

出　　处：《实用放射学杂志》2021年第10期1648-1650,共3页Journal of Practical Radiology

基　　金：江苏省科技计划资助项目(BE2018646);常州市科技计划资助项目(CE20185034).

摘　　要：目的探讨血氧水平依赖磁共振成像(BOLD-MRI)纹理分析对肾透明细胞癌(ccRCC)病理分级的预测价值.方法回顾性分析经病理证实为ccRCC且Fuhrman分级明确的84例患者资料.所有患者分为低级别组(Fuhrman Ⅰ+Ⅱ,n=56)和高级别组(Fuhrman Ⅲ+Ⅳ,n=28),术前2周均行肾脏BOLD-MRI扫描.提取肾肿瘤BOLD纹理特征,特征降维后采用多因素Logistic回归构建预测模型.采用受试者工作特征(ROC)曲线分析预测模型鉴别高、低级别ccRCC的诊断效能.结果提取出396个肾肿瘤BOLD纹理特征,降维后筛选出10个冗余度最低、相关性最高的特征.多因素Logistic回归分析构建的预测模型鉴别高、低级别ccRCC的曲线下面积(AUC)为0.84,最佳截断值为-0.80,其准确性、敏感性和特异性分别为73.81%、85.71%、67.86%.结论BOLD-MRI纹理分析可用于术前无创性预测ccRCC病理分级,为临床ccRCC手术方案的制订和预后预测提供依据.Objective To explore the predictive value of blood oxygen level dependent magnetic resonance imaging(BOLD-MRI)texture analysis for pathological grade of clear cell renal cell carcinoma(ccRCC).Methods The data of 84 patients with ccRCC confirmed by pathology and definite Fuhrman classification were analyzed retrospectively.Al patients were divided into the low-grade group(Fuhrman Ⅰ+Ⅱ,n=56)and the high-grade group(Fuhrman Ⅲ+Ⅳ,n=28).BOLD-MRI was performed within two weeks before the surgery.BOLD texture features of renal tumors were extracted.After dimension reduction of all the features,multivariate Logistic regression was used to generate the prediction model.The receiver operating characteristic(ROC)curve analysis was used to assess the diagnostic efficiency for differentiating high-grade ccRCC from low-grade ccRCC.Results 396 BOLD texture features from an individual renal tumor were extracted,and 10 features with the lowest redundancy and the highest correlation were selected after dimension reduction.The area under the curve(AUC)of the prediction model generated by multivariate Logistic regression analysis was 0.84,with the optimal cutoff value of -0.80,the accuracy of 73.81%,the sensitivity of 85.71%,and the specificity of 67.86%,respectively.Conclusion BOLD-MRI texture analysis can be a noninvasive approach for preoperative prediction of ccRCC pathological grade,and may provide basis for the optimizing operation scheme and prognostic prediction in clinic.

关 键 词：血氧水平依赖 磁共振成像 纹理分析 肾透明细胞癌 分级 

分 类 号：R445.2[医药卫生—影像医学与核医学] R737.11[医药卫生—诊断学]