Clinical Diagnostic Value Of AI-based CT-FFR And TAG Analysis Of Critical Coronary Artery Lesions

BackgroundPatients with critical coronary artery lesions,easy to ignore,and thus develop into a serious MACE event^[1]（including cardiac death,nonfatal myocardial infarction,blood vessel revascularization）again,have harsh consequences,how to accurately assess myocardial perfusion in patients with dynamic change,and select the patient treatment,improve prognosis has been closely watched.With the update of CT machines and the change of high-tech new technologies,many innovative cardiac coronary imaging technologies have been studied and excavated,and used in the evaluation of coronary artery anatomy and hemodynamic significance.Artificial Intelligence（Artificial Intelligence,AI）post-processing CT-fractional flow reserve（CT-FFR）and Transluminal decay gradient（TAG）are gradually applied.To evaluate the diagnostic analysis of coronary artery stenosis and myocardial ischemia.ObjectiveUsing CCTA as the original data,CT-FFR and TAG were obtained by AI post-processing technology,which were used for the comparative diagnosis of coronary artery critical lesion-specific ischemia and gave guidance for clinical diagnosis and treatment.MethodsPatients with chest pain admitted from December 2018 to January 2020 who had undergone Coronary artery CT angiography（CCTA）and had Coronary artery stenosis detected were enrolled.All patients were performed Digital Subtraction Angiography（DSA）.sixty-nine patients with coronary artery stenosis of about 50%-70%^[2]were selected as the focus of the study.In the whole hospitalization period,the measurement of Fractional flow reserve（FFR）of coronary arteries has been completed,and the CT-FFR and TAG values of each vessel are obtained through the AI-post-processing auxiliary diagnosis system,and the ROC curves of CT-FFR,TAG and CCTA are drawn,and the AUC is obtained.The accuracy of various diagnostic methods for coronary artery stenosis was compared,and Spearman correlation analysis was used to draw correlation scatter plot,and the consistency of diagnosis of various methods was tested.ResultsOf the seventy-one vessels,twenty-eight vessel lesions（39.4%）had invasive FFR（≤0.80）.The sensitivity,specificity,Youden index,positive predictive value,and negative predictive value of TAG≤-0.34HU/mm for measuring ischemia were 35.7%,81.4%,0.16,55.6%,and 66.0%,respectively.CT-FFR were 80.8%,91.3%,0.72,84.0%,and 89.3%,respectively.Receiver operating characteristic curve analysis showed that the AUC of CT-FFR（0.906）was significantly greater than that of TAG（0.575）,and there were statistical differences between CCTA and CT-FFR,CCTA and TAG,CT-FFR and TAG.The area under the ROC curve of FFR combined with TAG was 0.949.AUC of CT-FFR（0.868）was not significantly different from TAG（0.783,P=0.011）when non-calcified or mixed plaques were present in coronary plaques.The AUC of TAG（0.923）was greater than CT-FFR（0.699）when calcified plaque was present in coronary vascular plaque.Spearman correlation analysis showed that there was a moderate correlation between the TAG of seventy-one coronary arteries and the coronary stenosis area,and there was a statistical significance（r=0.483,P<0.001）.The correlation between the TAG of non-ischemic vessels in coronary artery and the size of coronary stenosis was r=0.369,P=0.015.The correlation between vascular TAG area and coronary stenosis in patients with coronary ischemia was r=0.640,and the difference was statistically significant（P<0.001）.ConclusionTaking FFR as the gold standard,compared with AI-analyzed TAG,AI-derived CT-FFR provided better diagnostic performance for the diagnosis of lesion-specific ischemia;in vessels with calcified plaque,AI-analyzed TAG was better than AI-analyzed TAG.CT-FFR has certain advantages;in vessels with non-calcified or mixed plaque,there is no significant difference between AI-analyzed TAG and CT-FFR;AI-analyzed TAG is correlated with the size of coronary artery stenosis.