Analysis Of Fat Attenuation Index And Radiomics Signature Around Plaques In Patients With Acute Coronary Syndrome Based On CCTA Imaging

Objective:To compare the attenuation and radiomics signature of pericoronary adipose tissue(PCAT)in patients with acute coronary syndrome(ACS)and matched controls with chronic coronary disease(CCD),and to explore the relationship between PCAT CT attenuation,radiomics signature and acute plaque instability,to improve the identification of vulnerable plaque in patients prone to coronary syndrome.Methods:The clinical data and images of 80 ACS patients who underwent CCTA in the affiliated Hospital of Yangzhou University from July 2019 to December 2022 were analyzed retrospectively.According to age,sex,BMI and risk factors,ACS patients was matched with 80 stable angina pectoris patients who underwent CCTA in the same period.Both groups were examined by CCTA within 72 hours before ICA,and the culprit plaque of ACS patients were determined by ICA.The plaque volume of culprit lesions and all main coronary arteries(RCA,LAD and LCX),PCAT attenuation at proximal 40 mm and radiomics signature parameters around culprit plaque in two groups were measured respectively.Using machine learning,the features selected after dimension reduction are divided into three categories: plaque,pericoronary and radiomics.the maximum weight parameters were selected to evaluate the recognition efficiency and clinical applicability of the three kinds of feature parameters and the combination of three categories by calculating the working characteristic curve and decision curve of the subjects.Finally,the machine learning classifier model is established based on the plaque volume,pericoronary fat attenuation index and radiomics,and the recognition efficiency of the combination of three categories is evaluated by the model efficiency.Results:The sum of calcified plaque volume in culprit plaque and all main coronary arteries in ACS patients was significantly less than that in control group,and the median volume of calcified plaque in responsible lesions was 33.79 mm 3and45.73 mm 3,respectively,P < 0.05,and the median volume of calcified plaques in all coronary arteries was 125.83 mm 3and 210.83 mm 3,respectively,P < 0.05.The 40 mm PCAT attenuation of culprit plaque and all proximal coronary arteries in ACS patients was higher than that in the control group(higher degree of inflammation),and there was statistical significance,in which the average PCAT attenuation in culprit plaque was-89.03 ±10.49,-96.26 ±9.23HU(P < 0.001),and the average PCAT attenuation in all coronary artery trunks(RCA,LAD and LCX)was RCA:-91.29 ±11.49,-94.77±8.13 HU respectively(P < 0.01).LAD:-89.11 ±8.04,-94.43 ±7.73HU(P < 0.001).The area under the ROC and 95% confidence interval of plaque,pericoronal and radiomics are plaque volume: 0.487(0.397,0.587),FAI:0.688(0.606,0.769),radiomics:0.880(0.827,0.933),combined with three categories,0.931(0.893,0.970).Among them,radiomics and the combination of three categories have achieved better recognition efficiency and greater clinical benefits.The area under curve(AUC)of machine learning combined models(SVM,KNN,DT,RF,ET and XGBoost)are 0.955,0.972,0.832,0.955,0.980 and 0.955,respectively,indicating that the combined model has a strong ability to distinguish between ACS plaque instability.Conclusions:PCAT attenuation at culprit plaque and whole coronary artery level in ACS patients is higher than CCD patients,indicating that ACS patients have higher levels of vascular inflammation;CCTA-based PCAT radiomics analysis can find subtle remodeling around coronary arteries other than vascular inflammation,combined with PCAT attenuation and radiomics signature may reliable identification of plaque instability.