Value of machine learning–based coronary ct fractional flow reserve applied to triple-rule-out ct angiography in acute chest pain

Abstract

Purpose: To evaluate the additional value of noninvasive artificial intelligence (AI)–based CT-derived fractional flow reserve (CT FFR), derived from triple-rule-out coronary CT angiography for acute chest pain (ACP) in the emergency department (ED) setting. Materials and Methods: AI-based CT FFR from triple-rule-out CT angiography data sets was retrospectively obtained in 159 of 271 eligible patients (102 men; mean age, 57.0 years ± 9.7 [standard deviation]) presenting to the ED with ACP. The agreement between CT FFR (≤ 0.80) and stenosis at triple-rule-out CT angiography (≥ 50%), as well as downstream cardiac diagnostic testing, was in-vestigated. Furthermore, the predictive value of CT FFR for coronary revascularization and major adverse cardiac events (MACE) was assessed over a 1-year follow-up period. Results: CT FFR and triple-rule-out CT angiography demonstrated agreement in severity of coronary artery disease (CAD) in 52% (82 of 159) of all cases. CT FFR of 0.80 and less served as a better predictor for coronary revascularization and MACE than stenosis of 50% and greater at triple-rule-out CT angiography (odds ratio, 3.4; 95% confidence interval: 1.4, 8.2 vs odds ratio, 2.2; 95% confidence interval: 0.9, 5.3) (P