BACKGROUND: Advances in ablation for atrial fibrillation (AF) continue to be hindered by ambiguities in mapping, even between experts. We hypothesized that convolutional neural networks (CNN) may enable objective analysis of intracardiac activation in AF, which could be applied clinically if CNN classifications could also be explained. METHODS: We performed panoramic recording of bi-atrial electrical signals in AF. We used the Hilbert-transform to produce 175 000 image grids in 35 patients, labeled for rotational activation by experts who showed consistency but with variability (kappa [κ]=0.79). In each patient, ablation terminated AF. A CNN was developed and trained on 100 000 AF image grids, validated on 25 000 grids, then tested on a separate 50 000 grids. RESULTS: In the separate test cohort (50 000 grids), CNN reproducibly classified AF image grids into those with/without rotational sites with 95.0% accuracy (CI, 94.8%-95.2%). This accuracy exceeded that of support vector machines, traditional linear discriminant, and k-nearest neighbor statistical analyses. To probe the CNN, we applied gradient-weighted class activation mapping which revealed that the decision logic closely mimicked rules used by experts (C statistic 0.96). CONCLUSIONS: CNNs improved the classification of intracardiac AF maps compared with other analyses and agreed with expert evaluation. Novel explainability analyses revealed that the CNN operated using a decision logic similar to rules used by experts, even though these rules were not provided in training. We thus describe a scaleable platform for robust comparisons of complex AF data from multiple systems, which may provide immediate clinical utility to guide ablation. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT02997254. Graphic Abstract: A graphic abstract is available for this article.
BACKGROUND: Advances in ablation for atrial fibrillation (AF) continue to be hindered by ambiguities in mapping, even between experts. We hypothesized that convolutional neural networks (CNN) may enable objective analysis of intracardiac activation in AF, which could be applied clinically if CNN classifications could also be explained. METHODS: We performed panoramic recording of bi-atrial electrical signals in AF. We used the Hilbert-transform to produce 175 000 image grids in 35 patients, labeled for rotational activation by experts who showed consistency but with variability (kappa [κ]=0.79). In each patient, ablation terminated AF. A CNN was developed and trained on 100 000 AF image grids, validated on 25 000 grids, then tested on a separate 50 000 grids. RESULTS: In the separate test cohort (50 000 grids), CNN reproducibly classified AF image grids into those with/without rotational sites with 95.0% accuracy (CI, 94.8%-95.2%). This accuracy exceeded that of support vector machines, traditional linear discriminant, and k-nearest neighbor statistical analyses. To probe the CNN, we applied gradient-weighted class activation mapping which revealed that the decision logic closely mimicked rules used by experts (C statistic 0.96). CONCLUSIONS: CNNs improved the classification of intracardiac AF maps compared with other analyses and agreed with expert evaluation. Novel explainability analyses revealed that the CNN operated using a decision logic similar to rules used by experts, even though these rules were not provided in training. We thus describe a scaleable platform for robust comparisons of complex AF data from multiple systems, which may provide immediate clinical utility to guide ablation. Registration: URL: https://www.clinicaltrials.gov; Unique identifier: NCT02997254. Graphic Abstract: A graphic abstract is available for this article.
Entities:
Keywords:
atrial fibrillation; machine learning; pulmonary veins; support vector machine
Authors: Brian J Hansen; Jichao Zhao; Ning Li; Alexander Zolotarev; Stanislav Zakharkin; Yufeng Wang; Josh Atwal; Anuradha Kalyanasundaram; Suhaib H Abudulwahed; Katelynn M Helfrich; Anna Bratasz; Kimerly A Powell; Bryan Whitson; Peter J Mohler; Paul M L Janssen; Orlando P Simonetti; John D Hummel; Vadim V Fedorov Journal: JACC Clin Electrophysiol Date: 2018-11-01
Authors: Atul Verma; Andrea Sarkozy; Allan Skanes; Mattias Duytschaever; Alan Bulava; Roy Urman; Yariv A Amos; Tom de Potter Journal: J Cardiovasc Electrophysiol Date: 2018-10-08
Authors: Barbara Bellmann; Marit Zettwitz; Tina Lin; Peter Ruppersberg; Selma Guttmann; Verena Tscholl; Patrick Nagel; Mattias Roser; Ulf Landmesser; Andreas Rillig Journal: Int J Cardiol Date: 2019-02-11 Impact factor: 4.164
Authors: Rui Providência; Pier D Lambiase; Neil Srinivasan; Girish Ganesh Babu; Konstantinos Bronis; Syed Ahsan; Fakhar Z Khan; Anthony W Chow; Edward Rowland; Martin Lowe; Oliver R Segal Journal: Circ Arrhythm Electrophysiol Date: 2015-06-16
Authors: Aleksei V Mikhailov; Anuradha Kalyanasundaram; Ning Li; Shane S Scott; Esthela J Artiga; Megan M Subr; Jichao Zhao; Brian J Hansen; John D Hummel; Vadim V Fedorov Journal: J Mol Cell Cardiol Date: 2020-10-29 Impact factor: 5.000
Authors: Chayakrit Krittanawong; Albert J Rogers; Kipp W Johnson; Zhen Wang; Mintu P Turakhia; Jonathan L Halperin; Sanjiv M Narayan Journal: Nat Rev Cardiol Date: 2020-10-09 Impact factor: 32.419
Authors: Venkat D Nagarajan; Su-Lin Lee; Jan-Lukas Robertus; Christoph A Nienaber; Natalia A Trayanova; Sabine Ernst Journal: Eur Heart J Date: 2021-10-07 Impact factor: 29.983
Authors: Jordi Heijman; Henry Sutanto; Harry J G M Crijns; Stanley Nattel; Natalia A Trayanova Journal: Cardiovasc Res Date: 2021-06-16 Impact factor: 10.787