Inter-operator reliability of an onsite machine learning-based prototype to estimate CT angiography-derived fractional flow reserve. (20th July 2021)

• Record Type: Journal Article
• Authors: Han, Y; Ahmed, AI; Schwemmer, C; Cocker, M; Alnabelsi, T; Ramirez-Giraldo, JC; Al Rifai, M; Nabi, F; Chang, SM; Al-Mallah, MH
• Is Part Of: European heart journal. Volume 22(2021)Supplement 3
• Publication Date: 2021-07-20
• Subjects: Cardiovascular system -- Imaging -- Periodicals; Heart -- Imaging -- Periodicals; 616.10754
• DOI: 10.1093/ehjci/jeab111.027
• Languages: English
• ISSNs: 2047-2404

Abstract

Abstract: Funding Acknowledgements: Type of funding sources: None. Background: Advances in computed tomography (CT) and machine learning have enabled on-site non-invasive assessment of fractional flow reserve (ML-FFRCT). However, reproducibility of measurements across operators is not well demonstrated. Purpose: This study was designed to measure the inter-operator variability and reproducibility of Coronary CT Angiography–derived fractional flow reserve values using a post-processing prototype based on a machine learning algorithm (ML-FFRCT). Methods: We included 60 symptomatic patients who underwent coronary CT angiography. FFRCT was calculated by 2 independent operators after training using a machine learning based on-site prototype. FFRCT was measured 1 cm distal to the coronary plaque or in the middle of the segments if no coronary lesions were present. Intraclass correlation coefficient (ICC) and Bland-Altman analysis were used to evaluate inter-operator variability effect in FFRCT estimates. Sensitivity analysis was done by cardiac risk factors, degree of stenosis and image quality. Results: A total of 535 coronary segments in 60 patients were assessed. The overall ICC was 0.986 per patient (95% CI: 0.977 - 0.992) and 0.972 per segment (95% CI: 0.967 - 0.977). The absolute mean difference in FFRCT estimates was 0.012 per patient (95% CI for limits of agreement: -0.035 - 0.039) and 0.02 per segment (95% CI for limits of agreement: -0.077 - 0.080). Tight limits of agreement were seen on Bland-Altman analysis. Distal segments had greater variability compared to proximal/mid segments (absolute mean difference 0.011 vs 0.025, p < 0.001). Sensitivity analysis showed similar results across degrees of stenosis, image quality and those with cardiac risk factors such as hypertension, diabetes and dyslipidemia. Conclusion: A high degree of inter-operator reproducibility can be achieved by onsite machine learning based ML-FFRCT assessment. Future research is required to evaluate the physiological relevance and prognostic value of ML-FFRCT.