基于CCTA的影像组学列线图用于慢性心肌缺血预测的构建与价值  

The construction and value of CCTA-based radiomics nomogram for the prediction of chronic myocardial ischemia

作　　者：徐玉芸[1] 何晓东[1] 余苔痕 李杰[1] XU Yuyun;HE Xiaodong;YU Taihen;LI Jie(Department of Radiology,Zhejiang Provincial People's Hospital,Hangzhou 310014,China)

机构地区：[1]浙江省人民医院放射科,杭州310014

出　　处：《心电与循环》2020年第5期442-448,452,I0001,共9页Journal of Electrocardiology and Circulation

基　　金：浙江省医药卫生科技计划项目(2020KY402)。

摘　　要：目的探讨基于冠状动脉CT血管造影(CCTA)的影像组学列线图用于慢性心肌缺血预测的构建与价值。方法回顾性分析2017年9月至2020年1月在浙江省人民医院行CCTA和单光子发射计算机化断层显象心肌灌注成像(SPECT-MPI)检查的202例患者的临床资料,其中心肌缺血99例。根据7∶3比例按诊断时间分为训练组(141例),测试图像特征的稳健性并构建列线图,和验证组(61例),验证所构建的列线图。利用新定量显像技术建立影像组学标签模型,使用logistic回归模型构建预测心肌缺血的影像组学列线图。使用受试者工作特征曲线(ROC)评估影像组学标签模型及列线图的诊断效能。结果心肌缺血与非缺血组高脂血症、高血压、钙化程度、斑块总体积和影像组学标签分数比较,差异均有统计学意义(均P<0.05)。影像组学标签模型预测训练组和验证组发生心肌缺血的曲线下面积(AUC)、灵敏度、特异度分别为0.852、0.743、0.873和0.832、0.818、0.785。列线图预测训练组和验证组发生心肌缺血的准确度、灵敏度、特异度分别为0.894、0.829、0.859和0.879、0.828、0.844。列线图、影像组学标签分数、钙化程度、高血压预测所有患者发生慢性心肌缺血的准确度分别为0.856、0.816、0.722、0.596。结论基于CCTA构建的影像组学列线图有助于识别慢性心肌缺血人群。Objective To explore the construction and its value of radiomics nomogram based on coronary computed tomography angiography(CCTA)for the prediction of chronic myocardial ischemia(MIS).Methods The clinical data of 202 patients underwent CCTA and single-photon emission computed tomography myocardial perfusion imaging(SPECT-MPI)examinations at Zhejiang Provincial People's Hospital from September 2017 to January 2020 were retrospectively analyzed.Among them,99 cases were diagnosed as MIS.According to the diagnostic time and in a ratio 7∶3,they were divided into training set(n=141)for examining robustness of radiomic features and constructing a nomogram,and validation set(n=61)for verifying the constructed radiomics monogram.A radiomics signature model was established using a novel quantitative imaging technique.Logistic regression model was used to construct radiomics nomogram for predicting MIS.The ROC curve was used to evaluate the diagnostic performance of the radiomics signature model and the nomogram.Results There were significant differences of hyperlipidemia,hypertension,degree of calcification,total plaque volume and radiomics signature scores between patients with and without MIS(all P<0.05).The AUC,sensitivity,and specificity of the radiomics signature model for predicting MIS in the training set and the validation set were 0.852,0.743,0.873 and 0.832,0.818,0.785,respectively.The accuracy,sensitivity,and specificity of the nomogram to predict MIS in the training set and the validation set were 0.894,0.829,0.859 and 0.879,0.828,0.844,respectively.The accuracy of nomogram,radiomics score,degree of calcification and hypertension in predicting MIS in all patients were 0.856,0.816,0.722 and 0.596,respectively.Conclusion The radiomics nomogram constructed based on CCTA maybe benefit in identifying people with MIS.

关 键 词：影像组学 冠状动脉CT血管造影 心肌缺血 列线图 

分 类 号：R542.2[医药卫生—心血管疾病]