Objective To investigate the diagnostic value of artificial intelligence-assisted quantitative parameters of pericoronary adipose tissue(PCAT)combined with coronary CT angiography(CCTA)plaque characteristics in non-ob-structive coronary artery ischemic disease(INOCA).Methods A total of 176 patients diagnosed with non-obstructive coronary artery stenosis(stenosis degree＜50％)by CCTA in our hospital were retrospectively collected.They were divid-ed into an ischemia group and a non-ischemia group based on clinical symptoms of myocardial ischemia.The plaque charac-teristics of CCTA and quantitative parameters of PCAT(FAI,volume)assisted by artificial intelligence were compared be-tween the two groups.Logistic regression was used to analyze the risk factors for myocardial ischemia,and a receiver operat-ing characteristic(ROC)curve was drawn to obtain the best diagnostic model.Results There were no statistically signif-icant differences in clinical baseline parameters between the ischemia group(93 cases)and the non-ischemia group(83 cases).The ischemia group had higher values than the non-ischemia group for total plaque volume,calcified plaque volume,non-calcified plaque volume,lipid plaque volume,lipid-fibrous plaque volume,plaque length,remodeling index,plaque bur-den,percentage of non-calcified plaque volume,and FAI(all P＜0.05).Among these variables,independent predictors of myocardial ischemia included plaque length(OR=1.070;P=0.001),plaque burden(OR=1.144;P＜0.001),and proximal FAI(OR=1.111;P=0.002).The sensitivity and specificity for predicting myocardial ischemia using only plaque characteristics were 48.18％and 83.14％,respectively;AUC=0.696;P＜0.001.The sensitivity and specificity for predic-ting myocardial ischemiawith only FAIwere 83.2％and32.6％,respectively;AUC=0.586;P=0.003.The combinationof bothvariables yieldeda sensitivityof62.3％and aspecificityof75.6％;AUC=0753;P＜0001.The ROCcurve showed that the prediction efficiency of the combination model was better(AUC=0.753),higher than that of the plaque characteristics model(AUC=0.696)and FAI model(AUC=0.586),and the differences were statistically significant(P＜0.05).Conclusion Plaque length,plaque burden,and proximal FAI are independent predictors of myocardial ischemia in the IN-OCA population,and the combination model of FAI and plaque characteristics has a higher predictive efficiency for myocar-dial ischemia in the INOCA population.
Ischemic with non-obstructive coronary arteriesPlaquesPericoronary fatMyocardial ischemiaCom-puterized Tomography
NETL
NSTL
万方数据
