基于胰腺周围脂肪间隙CT影像组学预测早期急性胰腺炎进展的价值

The value of CT based radiomics of peripancreatic adipose space in predicting progression of early acute pancreatitis

目的:探讨基于胰周脂肪间隙CT影像组学预测早期急性胰腺炎(AP)进展的价值。方法:回顾性分析123例根据新修订的亚特兰大分类诊断为AP的患者(进展组39例,非进展组84例),所有患者均接受腹部平扫及增强CT扫描。采用完全随机方法将患者按7:3的比例分为训练组和验证组。对各期手动勾画距炎症胰腺前缘3~5 mm范围的单层感兴趣区(ROI),采用AK软件提取CT纹理特征,使用最小冗余最大相关(mRMR)和最小绝对值收敛和选择算子(LASSO)回归分析对纹理特征降维、建立影像组学标签,并运用100次留组交叉验证(LGOCV)对模型的可靠性进行验证。将临床资料、CT特征及影像组学标签采用多因素Logistic回归分析建立影像组学模型。利用受试者工作特征(ROC)曲线评价模型的预测效能,应用决策曲线分析(DCA)评估模型的临床应用价值。结果:临床模型、平扫序列影像组学标签、联合序列影像组学标签、平扫序列个性化模型及联合序列个性化模型在训练组中的AUC分别为0.70、0.94、0.94、0.94及0.97,在验证组中的AUC分别为0.83、0.95、0.96、0.99、0.98。DCA显示平扫、联合序列影像组学标签及平扫、联合序列个性化模型均具有较好的净收益。结论:基于胰周脂肪间隙CT影像组学模型预测早期AP患者病情进展具有较高的诊断效能,明显优于临床模型,平扫序列影像组学标签简便有效,是有潜在应用前景的生物标志物。

Objective:To investigate the value of CT based radiomics of peripancreatic adipose space in predicting progression of early acute pancreatitis(AP).Methods:123 AP patients diagnosed according to the new revised Atlanta classification(39 case of progressive group and 84 case of non-progressive group)were retrospectively analyzed.All patients underwent plain and enhanced abdominal CT scan.The patients were divided into training group and verification group in a ratio of 7:3 by completely random method.The single-layer ROI was manually drawn of the range of 3~5mm anterior edge of the inflammation of the pancreas on each phase.AK software was used to extract texture features.The minimum redundancy maximum relevance(mRMR)and least absolute shrinkage and selection operator(LASSO)regression analysis were used to select feature and establish radiomics label,and 100 leave-group-out cross validation(LGOCV)was used to verify the reliability of the model.Then radiomics model was established with clinical data,CT characteristics,radiomics label by using the multiple Logistic regression analysis.Receiver operating characteristic(ROC)curve was used to evaluate the prediction of the model and the decision curve analysis(DCA)was used to evaluate the clinical application value of the model.Results:In the training group,the AUC of clinical model,plain sequence radiomics label,combined sequence radiomics label,plain sequence personalized model and combined sequence personalized model was 0.70,0.94,0.94,0.94 and 0.97,respectively,and in the validation group,the AUC was 0.83、0.95、0.96、0.99 and 0.98,respectively.DCA showed that both the plain scan and the combined sequence radiomics label,the plain scan and the combined sequence personalized model had better benefits.Conclusion:CT radiomics model had good diagnostic efficiency for predicting progression of acute pancreatitis,which is obviously better than the clinical model.The plain scan sequence radiomics label is a simple and effective biomarker with potential application.