AUTHOR=Saglietto Andrea , Baccega Daniele , Esposito Roberto , Anselmino Matteo , Dusi Veronica , Fiandrotti Attilio , De Ferrari Gaetano Maria 

TITLE=Convolutional neural network (CNN)-enabled electrocardiogram (ECG) analysis: a comparison between standard twelve-lead and single-lead setups

JOURNAL=Frontiers in Cardiovascular Medicine

VOLUME=Volume 11 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2024.1327179

DOI=10.3389/fcvm.2024.1327179

ISSN=2297-055X

ABSTRACT=Background: Artificial Intelligence (AI) has shown promise in early detecting various cardiac conditions from standard 12-lead electrocardiograms (ECGs). No study has systematically investigated AI's ability to identify abnormalities from single-lead recordings across a range of pathological conditions. This study aimed to assess the performance of a Convolutional Neural Network (CNN) using a single lead (D1) instead of the standard 12-lead setup, in correctly identifying electrocardiographic abnormalities.We designed and trained a lightweight Convolutional Neural Network (CNN) to identify 20 different cardiac abnormalities on ECGs, based on data from PTB-XL dataset. The network was designed to accommodate different combinations of leads as input, with a relatively simple architecture (less than 100,000 learnable parameters). We compared various setups of leads including the standard 12-lead, D1 alone and D1 paired with an additional lead.The CNN based on single-lead ECG (D1) achieved satisfactory performance compared to the standard 12-lead framework (average percentage AUC difference: -8.7%). Notably, for certain diagnostic classes, the diagnostic AUC did not differ between the single-lead and the classical 12lead setup. When a second lead was available to the CNN in addition to D1, the AUC gap was further reduced to an average percentage difference of -2.8% compared to the standard 12-lead setup.Conclusions: A relatively lightweight CNN can predict different classes of cardiac abnormalities from lead D1 alone from the standard 12-lead ECG. Considering the growing availability of wearable devices capable of recording a D1-like single-lead ECG we discuss the perspective of our findings to contribute the foundations for large-scale screening of cardiac abnormalities.