Comparative study on software DEEPVESSEL FFR and invasive FFR in assessing coronary ischemia

Abstract

Objective: To assess application value of artificial intelligence software DEEPVESSEL FFR(DV-FFR) in evaluating CTA (computed tomography angiography) of coronary ischemia. Methods: This was a prospective, single-center, and self-control study, and 21 vessels in 18 patients with coronary artery narrow (degree between 30%-90%) were included. Coronary ischemia was evaluated by DV-FFR and invasive FFR, and the efficiency of DV-FFR for diagnosing coronary ischemia was assessed by comparison with the results of golden criterion (invasive FFR). Based on data obtained from CTA DICOM, the blood vessel images were segmented and reconstructed by DV-FFR with deep learning technology, and vessel centerlines was obtained. FFR was calculated using3-D geometric auto-quantitative technology to assess ischemia in LAD (left anterior descending), LCX (left circumflex), RCA (right coronary artery). Results: When cut-off value of invasive FFR was taken as 0.8, DV-FFR≤0.8 was considered as functional coronary ischemia. On vessel level, the accuracy, specificity, sensitivity, positive predictive value, and negative predictive value of DV- FFR for diagnosing functional coronary ischemia were 90.5%, 88.9%, 91.7%, 91.7% and 88.9%, respectively. While on patient level, the above indexes were 88.9%, 87.5%, 90.0%, 90.0%and 87.5%, respectively. The results of the 2 methods were similar (P=0.787). Conclusions: The results of DV-FFR showed a good consistency with that of invasive FFR in evaluating coronary artery narrow degree between 30%-90%, which could be used as effective diagnostic approach.

Keywords

• |computed tomography angiography|coronary artery|artificial intelligence software|fractional flow reserve