基于增强CT影像组学预测糖尿病状态下急性胰腺炎复发

Prediction of Recurrence of Acute Pancreatitis in Diabetes Based on Enhanced CT Radiomics

目的:探讨基于双期增强CT影像组学在糖尿病状态下预测急性胰腺炎复发的价值.方法:回顾性收集川北医学院附属医院2016年1月至2019年12月2型糖尿病(type 2 diabetes mellitus,T2DM)合并急性胰腺炎(acute pancreatitis,AP)的患者129例,所有患者均行上腹部增强扫描,其中T2DM合并首发AP的62例,T2DM合并急性复发性胰腺炎(recurrent acute pancreatitis,RAP)患者67例,以7:3比例随机地将入组患者分为训练集和测试集,同期收集2组临床特征(年龄、性别、并发症、严重程度、结石、高脂血症)。利用3D Slicer软件分别勾画动脉期及静脉期胰腺实质并提取影像组学特征。采用随机森林建立T2DM状态下预测AP复发的动静脉联合模型和临床模型,使用受试者工作特征(receiver operating characteristic,ROC)曲线评价2种模型的预测能力并使用DeLong检验比较动静脉期联合模型与临床模型的预测效能。结果:动静脉期联合模型训练集ROC曲线下面积(AUC)为0.903(95%CI 0.842~0.964),特异度为0.767,精确度为0.804,灵敏度为0.872,测试集AUC为0.843(95%CI 0.720~0.967),特异度为0.737,精确度为0.762,灵敏度为0.800;基于年龄和高脂血症临床模型训练集的AUC为0.785(95%CI 0.691~0.879),特异度为0.581,精确度为0.660,灵敏度为0.745;测试集中AUC为0.758(95%CI 0.590~0.926),特异度为0.737,精确度为0.737,灵敏度为0.700。DeLong检验结果显示动静脉联合模型和临床模型预测效能之间的差异有统计学意义(P<0.05)。结论:基于双期增强CT影像组学模型可以较准确地预测T2DM状态下AP复发,并能为临床制定个性化治疗方案提供一定的参考价值。

Purpose:To explore the value of dual-phase enhanced CT radiomics in predicting the recurrence of acute pancreatitis in patients with diabetes.Methods:A total of 129 patients with type 2 diabetes mellitus(T2DM)complicated with acute pancreatitis(AP)in the Affiliated Hospital of North Sichuan Medical College from January 2016 to December 2019 were retrospectively collected.All patients underwent upper abdominal enhanced scan,including 62 cases of T2DM complicated with first-episode AP and 67 cases of T2DM complicated with recurrent acute pancreatitis(RAP).They were randomly divided into training set and test set in the ratio of 7:3.The clinical characteristics(age,gender,complications,severity,stones,hyperlipidemia)of the two groups were collected at the same time.The 3D Slicer software was used to outline the arterial and venous pancreatic parenchyma and extract the radiomics features.Random forest was used to establish the prediction model,and the receiver operating characteristic(ROC)curve was used to evaluate the prediction performance of the model.DeLong test was used to compare the efficiency of the combination of arterial and venous phases and the clinical models.Results:The area under the ROC curve(AUC)of the training set of the combined model in the arteriovenous phase was 0.903(95%CI 0.842-0.964),the specificity was 0.767,the accuracy was 0.804,and the sensitivity was 0.872.The AUC of the test set was 0.843(95%CI 0.720-0.967),the specificity was 0.737,the accuracy was 0.762,and the sensitivity was 0.800.The AUC of clinical model training set based on age and hyperlipidemia was 0.785(95%CI 0.691-0.879),the specificity was 0.581,the accuracy was 0.660,and the sensitivity was 0.745.The AUC of the test set was 0.758(95%CI 0.590-0.926),the specificity was 0.737,the accuracy was 0.737,and the sensitivity was 0.700.DeLong test results showed that there was a statistical significant difference between the prediction effectiveness of the combined model and clinical model(P<0.05).Conclusion:The dual-phase enhanced CT radiomics model can be used to accurately predict the recurrence of AP in T2DM,and provide certain reference value for the clinical formulation of personalized treatment regimens.