A 2-year investigation of the impact of the computed tomography–derived fractional flow reserve calculated using a deep learning algorithm on routine decision-making for coronary artery disease management

• Published in: European radiology Vol. 31; no. 9; pp. 7039 - 7046
• Main Authors: Liu, Xin, Mo, Xukai, Zhang, Heye, Yang, Guang, Shi, Changzheng, Hau, William Kongtou
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.09.2021; Springer Nature B.V
• Subjects: Algorithms; Angiography; Cardiovascular disease; Computed tomography; Computed Tomography Angiography; Coronary Angiography; Coronary artery; Coronary artery disease; Coronary Artery Disease - diagnostic imaging; Coronary Artery Disease - therapy; Coronary Stenosis - diagnostic imaging; Coronary Stenosis - therapy; Coronary vessels; Decision making; Deep Learning; Diagnostic Radiology; Diagnostic systems; Fractional Flow Reserve, Myocardial; Heart diseases; Hemodynamics; Humans; Imaging; Imaging Informatics and Artificial Intelligence; Internal Medicine; Intervention; Interventional Radiology; Lesions; Machine learning; Mathematical analysis; Medical imaging; Medicine; Medicine & Public Health; Neuroradiology; Patients; Predictive Value of Tests; Radiology; Retrospective Studies; Severity of Illness Index; Stenosis; Tomography; Tomography, X-Ray Computed; Ultrasound; Computed tomography angiography; Deep learning; Myocardial fractional flow reserve; Myocardial revascularisation; Coronary artery disease
• ISSN: 0938-7994; 1432-1084; 1432-1084
• DOI: 10.1007/s00330-021-07771-7

Objective This study aims to investigate the safety and feasibility of using a deep learning algorithm to calculate computed tomography angiography–based fractional flow reserve (DL-FFRCT) as an alternative to invasive coronary angiography (ICA) in the selection of patients for coronary intervention. Materials and methods Patients ( N = 296) with symptomatic coronary artery disease identified by coronary computed tomography angiography (CTA) with stenosis over 50% were retrospectively enrolled from a single centre in this study. ICA-guided interventions were performed in patients at admission, and DL-FFRCT was conducted retrospectively. The influences on decision-making by using DL-FFRCT and the clinical outcome were compared to those of ICA-guided care for symptomatic CAD at the 2-year follow-up evaluation. Result Two hundred forty-three patients were evaluated. Up to 72% of diagnostic ICA studies could have been avoided by using a DL-FFRCT value > 0.8 as a cut-off for intervention. A similar major adverse cardiovascular event (MACE) rate was observed in patients who underwent revascularisation with a DL-FFRCT value ≤ 0.8 (2.9%) compared to that of ICA-guided interventions (3.3%) (stented lesions with ICA stenosis > 75%) ( p = 0.838). Conclusion DL-FFRCT can reduce the need for diagnostic coronary angiography when identifying patients suitable for coronary intervention. A low MACE rate was found in a 2-year follow-up investigation. Key Points • Seventy-two percent of diagnostic ICA studies could have been avoided by using a DL-FFRCT value > 0.8 as a cut-off for intervention. • Coronary artery stenting based on the diagnosis by using a 320-detector row CT scanner and a positive DL-FFRCT value could potentially be associated with a lower occurrence rate of major adverse cardiovascular events (2.9%) within the first 2 years. • A low event rate was found when intervention was performed in tandem lesions with haemodynamic significance based on DL-FFRCT < 0.8 as a cut-off value.