基于门脉期CT影像组学列线图早期预测重症急性胰腺炎  

The radiomics nomogram based on portal venous phase CT for early prediction of severe acute pancreatitis

作　　者：蒋宪 肖波 赵艳梅[1] 祝元仲 陈晓文 何汶静 JIANG Xian;XIAO Bo;ZHAO Yanmei;ZHU Yuanzhong;CHEN Xiaowen;HE Wenjing(School of Medical Imaging of North Sichuan Medical College,Nanchong,Sichuan Province 637000,China;Department of Radiology,the Affiliated Hospital of North Sichuan Medical College,Nanchong,Sichuan Province 637000,China)

机构地区：[1]川北医学院医学影像学院,四川南充637000 [2]川北医学院附属医院放射科,四川南充637000

出　　处：《实用放射学杂志》2023年第11期1792-1796,共5页Journal of Practical Radiology

基　　金：南充市市校科技战略合作专项(川北医学院)资金资助项目(20SXQT0315);四川省科技厅自然科学基金资助项目(2023NSFSC0646)。

摘　　要：目的基于门脉期CT图像建立早期预测急性胰腺炎(AP)严重程度的影像组学列线图模型。方法本研究共纳入215例AP患者,以时间为界分为训练集和测试集。手动勾画胰腺实质感兴趣区(ROI)后提取影像组学特征,并使用LightGBM算法筛选特征。采用逻辑回归算法建立影像组学模型并计算影像组学得分(Radscore),再联合临床特征构建影像组学列线图模型。最后在测试集中验证该模型预测性能。结果共筛选出11个最佳影像组学特征。Radscore联合C反应蛋白和胰周积液构建的影像组学列线图在训练集、验证集以及测试集中的曲线下面积(AUC)分别为0.978、0.954和0.893。结论基于门脉期CT图像的影像组学列线图模型早期预测重症AP的性能良好,可为临床诊疗提供辅助性决策参考。Objective To establish a radiomics nomogram model based on CT images in the portal venous phase for early prediction of the severity of acute pancreatitis(AP).Methods A total of 215 patients with AP were included in this study.They were divided into a training set and a test set according to a time point during the study.The radiomics features were extracted after manually outlining the regions of interest(ROI)of pancreatic parenchyma.Then the LightGBM algorithm was used to screen these features.The logistic regression algorithm was used to develop the radiomics model and to calculate the Radscore,which was then combined with clinical features to construct the radiomics nomogram model.The predictive performance of the model was finally validated in the test set.Results A total of 11 best radiomics features were selected.The area under the curve(AUC)of the radiomics nomogram which was constructed by Radscore in combination with C-reactive protein and peripancreatic fluid was 0.978,0.954,and 0.893 in the training set,validation set,and test set respectively.Conclusion The radiomics nomogram model based on portal venous phase CT images has excellent performance for early prediction of severe AP,which can provide an adjunctive decision-making reference for clinical treatment.

关 键 词：影像组学 重症急性胰腺炎 机器学习 列线图 

分 类 号：R445[医药卫生—影像医学与核医学] R657.51[医药卫生—诊断学] P114.57[医药卫生—临床医学]