基于临床和CCTA影像组学特征冠心病患者冠状动脉易损斑块鉴别诊断模型的构建OA

Aim:To construct a differential diagnosis model for vulnerable coronary plaques in patients with coronary heart disease based on clinical and coronary computed tomography angiography(CCTA)radiomics features.Methods:A retrospective analysis was conducted on the clinical data and CCTA imaging data of 206 coronary heart disease patients(318 plaques)who underwent CCTA at the First Affiliated Hospital of Zhengzhou University from January 2024 to January 2025.The plaques were randomly divided into a training set(223 plaques)and a test set(95 plaques)in a 7∶3 ratio.The features were screened using LASSO regression,and a clinical model,a radiomics model,a multimodal feature quantification model were constructed.The efficacy of different models was evaluated using ROC curve,calibration curve,and decision curve a-nalysis.Results:Five clinical features including gender,minimum luminal area,plaque length,plaque volume,and calcified component volume were screened through LASSO regression(P＜0.05).One first-order feature(mean grayscale),1 mor-phological feature(plaque irregularity),and 5 texture features(small area low grayscale enhancement,regional entropy,cor-relation,deficit moment,and second-order moment)were selected using LASSO regression(P＜0.05).In the training set and the test set,the multimodal feature quantification model had the highest efficiency,with AUC(95%CI)of 0.992(0.970-0.999)and 0.963(0.903-0.991),respectively.Moreover,the original curve of this model was closer to the standard curve,and it had good net returns.Conclusion:The constructed multimodal feature quantification model based on clinical and CCTA radiomics features could accurately identify vulnerable coronary plaques.