Value of radiomics nomogram based on CCTA in identifying vulnerable plaques

WANG Linyuan; XIONG Xin; YANG Kun; DENG Yongzhi

Abstract：Objective To develop and verify a radiomics nomogram of radiomics signature and clinical factors to differentiate vulnerable plaques from stable plaques.Methods The clinical data of patients with coronary artery disease who underwent CCTA examination were retrospectively collected,with a total of 108 plaques in 102 patients,including 35 vulnerable plaques and 73 stable plaques.These plaques were randomly placed in a training or a test cohort at a ratio of 8:2.The plaque regions of interest were manually created from CCTA images and the radiomics features were extracted.The Mann-Whitney U test,Pearson correlation coefficient and least absolute shrinkage and selection operator(LASSO)method were used for screening radiomics features and constructing the radiomics model and calculating the radiomics score(Radscore).A combined diagnostic model and nomogram combining the Radscore and clinical factors were constructed.Finally,the receiver operating characteristic(ROC)curve,calibration curve and decision curve analysis were performed to estimate the performance of each model.Results Eight optimal radiomics features were finally filtered out.The radiomics nomogram,which combined clinical factors with Radscore achieved an area under the curve(AUC)of 0.999 versus 0.929 in the training and test cohorts.Calibration curves indicated that there was good consistency between actual observation and prediction in the identification of vulnerable plaques.Decision curves demonstrated the nomogram had potential in clinical application.Conclusion Nomogram,which combines Radscore and clini-cal factors,has demonstrated superior diagnostic performance in objectively and accurately identifying plaque stability and can provide a reference for future clinical decisions.