Application of CTA-based Radiomic Phenotyping of Peri-coronary Adipose Tissue and Fluid Dynamics in Atherosclerotic Disease

A multicenter, retrospective, observational trial will be conducted (APPLE study). To investigate whether a combined model constructed on the basis of pericoronary adipose tissue (PCAT) radiomics, fluid dynamics and clinical risk factors can predict the formation of atherosclerotic plaque. It will be carried out in 11 hospitals in 4 provinces in China. The Boruta algorithm and correlation proof clustering analysis were used to screen the imaging histological features, and a random forest model was used to construct an imaging histological prediction model for PCAT and fluid dynamics and to construct radiomics' score. To investigate the incremental value of the radiomics' score beyond the traditional prediction model, the radiomics' score was combined with the traditional logistic regression prediction model. Receiver operating characteristic (ROC) curve analysis with integrated discrimination improvement (IDI) and category net reclassification index (NRI) were used to compare the performance of the predictive models. A ML-prediction model incorporates FAI, fluid dynamics and patient clinical characteristics to identify high-risk patients in advance for patients receiving routine CCTA and guide the more precise use of preventative treatments, including anti-inflammatory therapies.