Reading Between the Leads: Local Lead-Attention Based Classification of Electrocardiogram Signals

Gouthamaan Manimaran; Sadasivan Puthusserypady; Helena Dominguez; Jakob E. Bardram

doi:10.22489/CinC.2023.221

Reading Between the Leads: Local Lead-Attention Based Classification of Electrocardiogram Signals

Gouthamaan Manimaran^*, Sadasivan Puthusserypady, Helena Dominguez, Jakob E. Bardram

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

1 Downloads (Pure)

Abstract

Self-attention models have emerged as powerful tools in both computer vision and Natural Language Processing (NLP) domains. However, their application in time-domain Electrocardiogram (ECG) signal analysis has been limited, primarily due to the lesser need for global receptive fields. In this study, we present a novel approach utilizing local self-attention to address multi-class classification tasks using the PhysioNet/Computing in Cardiology Challenge 2021 dataset, encompassing 26 distinct classes across six different datasets. We introduce an innovative concept called 'local lead-attention' to capture features within a single lead and across multiple configurable leads. The proposed architecture achieves an F1 score of 0.521 on the challenge's validation set, marking a 5.67% improvement over the winning solution. Remarkably, our model accomplishes this performance boost with only one-third of the total parameter size, amounting to 2.4 million parameters.

Original language	English
Title of host publication	Computing in Cardiology, CinC 2023
Number of pages	4
Publisher	IEEE Computer Society Press
Publication date	2023
ISBN (Electronic)	9798350382525
DOIs	https://doi.org/10.22489/CinC.2023.221
Publication status	Published - 2023
Externally published	Yes
Event	50th Computing in Cardiology, CinC 2023 - Atlanta, United States Duration: 1 Oct 2023 → 4 Oct 2023

Conference

Conference	50th Computing in Cardiology, CinC 2023
Country/Territory	United States
City	Atlanta
Period	01/10/2023 → 04/10/2023

Series	Computing in Cardiology
ISSN	2325-8861

Bibliographical note

Funding Information:
This project is part of the grant I+D+i PLEC2021-007614, funded by MCIN/AEI/10.13039/501100011033 and by the “European Union NextGenerationEU/PRTR” and by the European Union’s Horizon research and Innovation programme under the Marie Skłodowska-Curie grant agreement No. 860974.

Publisher Copyright:
© 2023 CinC.

Access to Document

10.22489/CinC.2023.221

FulltextAccepted author manuscript, 628 KB

Cite this

Reading Between the Leads: Local Lead-Attention Based Classification of Electrocardiogram Signals. / Manimaran, Gouthamaan; Puthusserypady, Sadasivan; Dominguez, Helena et al.
Computing in Cardiology, CinC 2023. IEEE Computer Society Press, 2023. (Computing in Cardiology).

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Manimaran, G, Puthusserypady, S, Dominguez, H & Bardram, JE 2023, Reading Between the Leads: Local Lead-Attention Based Classification of Electrocardiogram Signals. in Computing in Cardiology, CinC 2023. IEEE Computer Society Press, Computing in Cardiology, 50th Computing in Cardiology, CinC 2023, Atlanta, United States, 01/10/2023. https://doi.org/10.22489/CinC.2023.221

@inproceedings{79a15e3322a94230b38f1df058ac65e2,

title = "Reading Between the Leads: Local Lead-Attention Based Classification of Electrocardiogram Signals",

abstract = "Self-attention models have emerged as powerful tools in both computer vision and Natural Language Processing (NLP) domains. However, their application in time-domain Electrocardiogram (ECG) signal analysis has been limited, primarily due to the lesser need for global receptive fields. In this study, we present a novel approach utilizing local self-attention to address multi-class classification tasks using the PhysioNet/Computing in Cardiology Challenge 2021 dataset, encompassing 26 distinct classes across six different datasets. We introduce an innovative concept called 'local lead-attention' to capture features within a single lead and across multiple configurable leads. The proposed architecture achieves an F1 score of 0.521 on the challenge's validation set, marking a 5.67% improvement over the winning solution. Remarkably, our model accomplishes this performance boost with only one-third of the total parameter size, amounting to 2.4 million parameters.",

author = "Gouthamaan Manimaran and Sadasivan Puthusserypady and Helena Dominguez and Bardram, {Jakob E.}",

note = "Funding Information: This project is part of the grant I+D+i PLEC2021-007614, funded by MCIN/AEI/10.13039/501100011033 and by the “European Union NextGenerationEU/PRTR” and by the European Union{\textquoteright}s Horizon research and Innovation programme under the Marie Sk{\l}odowska-Curie grant agreement No. 860974. Publisher Copyright: {\textcopyright} 2023 CinC.; 50th Computing in Cardiology, CinC 2023 ; Conference date: 01-10-2023 Through 04-10-2023",

year = "2023",

doi = "10.22489/CinC.2023.221",

language = "English",

series = "Computing in Cardiology",

publisher = "IEEE Computer Society Press",

booktitle = "Computing in Cardiology, CinC 2023",

address = "United States",

}

TY - GEN

T1 - Reading Between the Leads

T2 - 50th Computing in Cardiology, CinC 2023

AU - Manimaran, Gouthamaan

AU - Puthusserypady, Sadasivan

AU - Dominguez, Helena

AU - Bardram, Jakob E.

N1 - Funding Information: This project is part of the grant I+D+i PLEC2021-007614, funded by MCIN/AEI/10.13039/501100011033 and by the “European Union NextGenerationEU/PRTR” and by the European Union’s Horizon research and Innovation programme under the Marie Skłodowska-Curie grant agreement No. 860974. Publisher Copyright: © 2023 CinC.

PY - 2023

Y1 - 2023

N2 - Self-attention models have emerged as powerful tools in both computer vision and Natural Language Processing (NLP) domains. However, their application in time-domain Electrocardiogram (ECG) signal analysis has been limited, primarily due to the lesser need for global receptive fields. In this study, we present a novel approach utilizing local self-attention to address multi-class classification tasks using the PhysioNet/Computing in Cardiology Challenge 2021 dataset, encompassing 26 distinct classes across six different datasets. We introduce an innovative concept called 'local lead-attention' to capture features within a single lead and across multiple configurable leads. The proposed architecture achieves an F1 score of 0.521 on the challenge's validation set, marking a 5.67% improvement over the winning solution. Remarkably, our model accomplishes this performance boost with only one-third of the total parameter size, amounting to 2.4 million parameters.

AB - Self-attention models have emerged as powerful tools in both computer vision and Natural Language Processing (NLP) domains. However, their application in time-domain Electrocardiogram (ECG) signal analysis has been limited, primarily due to the lesser need for global receptive fields. In this study, we present a novel approach utilizing local self-attention to address multi-class classification tasks using the PhysioNet/Computing in Cardiology Challenge 2021 dataset, encompassing 26 distinct classes across six different datasets. We introduce an innovative concept called 'local lead-attention' to capture features within a single lead and across multiple configurable leads. The proposed architecture achieves an F1 score of 0.521 on the challenge's validation set, marking a 5.67% improvement over the winning solution. Remarkably, our model accomplishes this performance boost with only one-third of the total parameter size, amounting to 2.4 million parameters.

U2 - 10.22489/CinC.2023.221

DO - 10.22489/CinC.2023.221

M3 - Article in proceedings

AN - SCOPUS:85182325715

T3 - Computing in Cardiology

BT - Computing in Cardiology, CinC 2023

PB - IEEE Computer Society Press

Y2 - 1 October 2023 through 4 October 2023

ER -