Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

RawECGNet : Deep Learning Generalization for Atrial Fibrillation Detection From the Raw ECG

Ben-Moshe, Noam ; Tsutsui, Kenta ; Biton, Shany ; Zvuloni, Eran ; Sornmo, Leif LU and Behar, Joachim A. (2024) In IEEE Journal of Biomedical and Health Informatics p.1-10
Abstract

Introduction Deep learning models for detecting episodes of atrial fibrillation (AF) using rhythm information in long-term ambulatory ECG recordings have shown high performance. However, the rhythm-based approach does not take advantage of the morphological information conveyed by the different ECG waveforms, particularly the f-waves. As a result, the performance of such models may be inherently limited. Methods: To address this limitation, we have developed a deep learning model, named RawECGNet, to detect episodes of AF and atrial flutter (AFl) using the raw, single-lead ECG. We compare the generalization performance of RawECGNet on two external data sets that account for distribution shifts in geography, ethnicity, and lead position.... (More)

Introduction Deep learning models for detecting episodes of atrial fibrillation (AF) using rhythm information in long-term ambulatory ECG recordings have shown high performance. However, the rhythm-based approach does not take advantage of the morphological information conveyed by the different ECG waveforms, particularly the f-waves. As a result, the performance of such models may be inherently limited. Methods: To address this limitation, we have developed a deep learning model, named RawECGNet, to detect episodes of AF and atrial flutter (AFl) using the raw, single-lead ECG. We compare the generalization performance of RawECGNet on two external data sets that account for distribution shifts in geography, ethnicity, and lead position. RawECGNet is further benchmarked against a state-of-the-art deep learning model, named ArNet2, which utilizes rhythm information as input. Results: Using RawECGNet, the results for the different leads in the external test sets in terms of the F1 score were 0.91–0.94 in RBDB and 0.93 in SHDB, compared to 0.89–0.91 in RBDB and 0.91 in SHDB for ArNet2. The results highlight RawECGNet as a high-performance, generalizable algorithm for detection of AF and AFl episodes, exploiting information on both rhythm and morphology.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; and
organization
publishing date
type
Contribution to journal
publication status
in press
subject
keywords
Atrial fibrillation, atrial flutter, Data models, Deep learning, deep learning, Detectors, electrocardiogram, Electrocardiography, Recording, Rhythm, Training
in
IEEE Journal of Biomedical and Health Informatics
pages
10 pages
publisher
IEEE - Institute of Electrical and Electronics Engineers Inc.
external identifiers
  • scopus:85194092110
  • pmid:38787663
ISSN
2168-2194
DOI
10.1109/JBHI.2024.3404877
language
English
LU publication?
yes
id
0e6b1c76-5bd3-48d8-96ca-b9904cefb535
date added to LUP
2024-06-19 14:22:25
date last changed
2024-06-20 03:00:04
@article{0e6b1c76-5bd3-48d8-96ca-b9904cefb535,
  abstract     = {{<p>Introduction Deep learning models for detecting episodes of atrial fibrillation (AF) using rhythm information in long-term ambulatory ECG recordings have shown high performance. However, the rhythm-based approach does not take advantage of the morphological information conveyed by the different ECG waveforms, particularly the f-waves. As a result, the performance of such models may be inherently limited. Methods: To address this limitation, we have developed a deep learning model, named RawECGNet, to detect episodes of AF and atrial flutter (AFl) using the raw, single-lead ECG. We compare the generalization performance of RawECGNet on two external data sets that account for distribution shifts in geography, ethnicity, and lead position. RawECGNet is further benchmarked against a state-of-the-art deep learning model, named ArNet2, which utilizes rhythm information as input. Results: Using RawECGNet, the results for the different leads in the external test sets in terms of the F1 score were 0.91&amp;#x2013;0.94&amp;#x00A0;in RBDB and 0.93&amp;#x00A0;in SHDB, compared to 0.89&amp;#x2013;0.91&amp;#x00A0;in RBDB and 0.91&amp;#x00A0;in SHDB for ArNet2. The results highlight RawECGNet as a high-performance, generalizable algorithm for detection of AF and AFl episodes, exploiting information on both rhythm and morphology.</p>}},
  author       = {{Ben-Moshe, Noam and Tsutsui, Kenta and Biton, Shany and Zvuloni, Eran and Sornmo, Leif and Behar, Joachim A.}},
  issn         = {{2168-2194}},
  keywords     = {{Atrial fibrillation; atrial flutter; Data models; Deep learning; deep learning; Detectors; electrocardiogram; Electrocardiography; Recording; Rhythm; Training}},
  language     = {{eng}},
  pages        = {{1--10}},
  publisher    = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}},
  series       = {{IEEE Journal of Biomedical and Health Informatics}},
  title        = {{RawECGNet : Deep Learning Generalization for Atrial Fibrillation Detection From the Raw ECG}},
  url          = {{http://dx.doi.org/10.1109/JBHI.2024.3404877}},
  doi          = {{10.1109/JBHI.2024.3404877}},
  year         = {{2024}},
}