The Official Journal of EuroPCR and the European Association of Percutaneous Coronary Interventions (EAPCI)
Deep Learning For Prediction of Fractional Flow Reserve From Resting Coronary Pressure Curves (ARTIST study)
Frederik M. Zimmermann1; Thomas P. Mast1; Nils P. Johnson2; Ivo Everts3; Barry Hennigan4; Colin Berry5; Daniel T. Johnson2; Bernard De Bruyne6; William F. Fearon7; Keith G. Oldroyd4; Nico H.J. Pijls8; Pim A.L. Tonino1; Marcel van 't Veer8;
1. Department of Cardiology, Catharina Hospital, Eindhoven, The Netherlands, Netherlands 2. Weatherhead PET Center, Division of Cardiology, Department of Medicine, McGovern Medical School at UTHealth and Memorial Hermann Hospital, Houston, TX, USA 3. GoDataDriven, Amsterdam, The Netherlands 4. British Heart Foundation Glasgow Cardiovascular Research Centre,Institute of Cardiovascular and Medical Sciences,University of Glasgow, Glasgow,UK;West of Scotland Heart and Lung Centre,Golden Jubilee National Hospital,Clydebank,Agamemnon Street,Glasgow UK 5. British Heart Foundation Glasgow Cardiovascular Research Centre, Institute of Cardiovascular and Medical Sciences, University of Glasgow, Glasgow, UK. West of Scotland Heart and Lung Centre, Golden Jubilee National Hospital, Clydebank, Agamemnon Street, Gl 6. Cardiovascular Center Aalst, OLV-Clinic, Aalst, Belgium. Department of Cardiology, Lausanne University Center Hospital, Switzerland 7. Division of Cardiovascular Medicine, Stanford University School of Medicine, Stanford, CA 8. Department of Cardiology, Catharina Hospital, Eindhoven, The Netherlands. Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, the Netherlands
As a public service to our readership, this article - peer reviewed by the Editors of EuroIntervention - has been published immediately upon acceptance as it was received. The content of this article is the sole responsibility of the authors, and not that of the journal or its publishers.
Please note that supplementary movies are not available online at this stage. Once a paper is published in its edited and formatted form, it will be accompanied online by any supplementary movies.
To read the full content of this article, please log in to download the PDF.
Aims It would be ideal for a non-hyperemic index to predict fractional flow reserve (FFR) more accurately, given FFR’s extensive validation in a multitude of clinical settings. The aim of this study was to derive a novel non-hyperemic algorithm based on deep learning and to validate it in an internal validation cohort against FFR.
Methods and Results The ARTIST study is a post hoc analysis of 3 previously published studies. In a derivation cohort (random 80% sample of the total cohort) a deep neural network was trained (deep learning) with paired examples of resting coronary pressure curves and their FFR values. The resulting algorithm was validated against unseen resting pressure curves from a random 20% sample of the total cohort. The primary endpoint was diagnostic accuracy of the deep learning-derived algorithms against binary FFR≤0.8. To reduce the variance in the precision, we used a 5-fold cross-validation procedure. A total of 1666 patients with 1718 coronary lesions and 2928 coronary pressure tracings were included. Diagnostic accuracy of our convolutional neural network (CNN) and recurrent neural networks (RNN) against binary FFR≤0.80 were 79.6±1.9%, and 77.6±2.3%, respectively. There was no statistically significant difference between the accuracy of our neural networks to predict binary FFR and the most accurate non-hyperemic pressure ratio (NHPR).
Conclusions Compared to standard derivation of resting pressure ratios, we did not find a significant improvement in FFR prediction when resting data is analysed using artificial intelligence approaches. Our findings strongly suggest that a larger class of hidden information within resting pressure traces is not the main cause for the known disagreement between resting indices and FFR. Therefore, if clinicians want to use FFR for clinical decision making, hyperaemia induction should remain the standard practice.