基于MRI的纹理分析预测肝外胆管癌分化程度的应用价值

Evaluation of MRI-Based Texture Analysis in Predicting the Differentiation Degree of Extrahepatic Cholangiocarcinoma

目的探讨基于MRI的纹理分析预测肝外胆管癌(ECC)分化程度的价值。方法回顾性分析我院2011年1月至2018年12月经病理证实的ECC患者97例(低分化16例,中-高分化81例)临床资料,所有患者术前4周内均行MRI检查。使用MaZda软件勾画感兴趣区(ROI),并自动提取纹理特征。应用Mann-Whitney U检验、相关分析及Logistics回归分析筛选出具有独立危险因素的纹理特征,并建立预测模型。应用ROC曲线评估预测模型的效能。结果区分低分化与中-高分化ECC并具有独立危险因素的纹理特征参数共6个(均P<0.05)。Logistics回归模型的预测值预测低分化ECC的ROC曲线下面积为0.961(P<0.001),其敏感度、特异度、阳性预测值和阴性预测值分别为87.7%、100.0%、61.5%、100.0%。结论基于MRI的纹理分析可以准确地区分低分化与中-高分化ECC。

Objective To investigate the diagnostic value of texture analysis based on MRI in predicting the differentiation degree of Extrahepatic Cholangiocarcinoma(ECC).Methods 97 patients who underwent MRI scan within four weeks before surgery with pathological diagnosis of ECC were retrospectively analyzed(16 cases poorly differentiated,81 cases moderately or well differentiated).Texture features of the lesion were extracted by MaZda software on the region of interest(ROI).Mann-Whitney U test,Pearson correlation coefficient and Logistics regression were used to select the texture features with independent risk factors,and the prediction model was established.Receiver operating characteristics(ROC)curve were used to evaluate the effectiveness of the prediction model.Results A total of 6 texture features with independent risk factors were selected to establish the prediction model.Area under ROC,sensitivity,specificity,positive predictive value and negative predictive value of prediction model were 0.961,87.7%,100.0%,61.5%and 100.0%,respectively.Conclusion Texture analysis based on MRI could accurately distinguish poorly differentiated ECC from moderately or well differentiated ECC.