Image-based deep learning for emergency electrocardiogram classification
- 1Heart Institute
Abstract
Automated electrocardiogram analysis has advanced largely through digital waveforms, yet many emergency-care workflows rely on ECGs available only as printed tracings, scanned reports, PDFs or mobile photographs. We developed an image-based deep learning system for emergency ECG classification and evaluated it in InCor-EMG, an expert-adjudicated dataset of 18,519 emergency ECGs spanning 12 ECG categories, with labels from 19 cardiologists. On the held-out test set, the final ConvNeXt ensemble achieved a macro F1-score of 0.807 (95% CI, 0.788-0.825), compared with 0.820 (95% CI, 0.805-0.832) for annotating cardiologists, and higher F1-scores than Mortara Veritas in most evaluated categories. Performance was associated more strongly with inter-reader agreement than with training sample size and remained informative across scanned and photographed ECGs, with supportive performance in model-enriched temporal and heterogeneous public-image evaluations. These findings support ECG-image classification when digital waveforms are unavailable.
Related
- Automated brain MRI image classification through machine learning and deep learning approaches
- A hybrid deep feature extraction and ensemble learning framework for robust multi-class ECG image classification
- Classification Performance of a Deep Learning Algorithm for Common Skin Lesions in the Emergency Department: A Multicenter Preliminary Validation Study
- Deep Learning for Real-Time Image Interpretation in Arthroscopy: A Systematic Review of Classification, Segmentation, and Detection Tasks
- Deep Learning-Based Versus Threshold-Based Bone Removal for CT Angiography of Head and Neck Arteriovenous Fistulas: A Retrospective Comparison of Image Quality, Key-Structure Display, and Post-Processing Time
BibTeX
@misc{dias2026imagebased,
title = {{Image-based deep learning for emergency electrocardiogram classification}},
author = {F. Meneguitti Dias and E. Ribeiro and N. Olivetti and O. Carvalho and J. E. Krieger and M. Gutierrez},
howpublished = {medRxiv},
year = {2026},
month = jun,
doi = {10.64898/2026.06.18.26355968},
url = {https://www.medrxiv.org/content/10.64898/2026.06.18.26355968v2},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - Image-based deep learning for emergency electrocardiogram classification
AU - F. Meneguitti Dias
AU - E. Ribeiro
AU - N. Olivetti
AU - O. Carvalho
AU - J. E. Krieger
AU - M. Gutierrez
PY - 2026
DA - 2026/06/18
PB - medRxiv
DO - 10.64898/2026.06.18.26355968
UR - https://www.medrxiv.org/content/10.64898/2026.06.18.26355968v2
AB - Automated electrocardiogram analysis has advanced largely through digital waveforms, yet many emergency-care workflows rely on ECGs available only as printed tracings, scanned reports, PDFs or mobile photographs. We developed an image-based deep learning system for emergency ECG classification and evaluated it in InCor-EMG, an expert-adjudicated dataset of 18,519 emergency ECGs spanning 12 ECG categories, with labels from 19 cardiologists. On the held-out test set, the final ConvNeXt ensemble achieved a macro F1-score of 0.807 (95% CI, 0.788-0.825), compared with 0.820 (95% CI, 0.805-0.832) for annotating cardiologists, and higher F1-scores than Mortara Veritas in most evaluated categories. Performance was associated more strongly with inter-reader agreement than with training sample size and remained informative across scanned and photographed ECGs, with supportive performance in model-enriched temporal and heterogeneous public-image evaluations. These findings support ECG-image classification when digital waveforms are unavailable.
ER -