CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification

<p>Machine learning technologies have been applied extensively in the last decade to automatically detect and analyze various forms of arrhythmia from electrocardiogram (ECG) signals. Existing deep learning-based models focus on enhancing classification performance by exploring spatial–tempora...

Full description

Saved in:
Bibliographic Details
Main Author: Md Rabiul Islam (6424796) (author)
Other Authors: Marwa Qaraqe (10135172) (author), Khalid Qaraqe (16896504) (author), Erchin Serpedin (3706543) (author)
Published: 2024
Subjects:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1864513507112779776
author Md Rabiul Islam (6424796)
author2 Marwa Qaraqe (10135172)
Khalid Qaraqe (16896504)
Erchin Serpedin (3706543)
author2_role author
author
author
author_facet Md Rabiul Islam (6424796)
Marwa Qaraqe (10135172)
Khalid Qaraqe (16896504)
Erchin Serpedin (3706543)
author_role author
dc.creator.none.fl_str_mv Md Rabiul Islam (6424796)
Marwa Qaraqe (10135172)
Khalid Qaraqe (16896504)
Erchin Serpedin (3706543)
dc.date.none.fl_str_mv 2024-03-13T12:00:00Z
dc.identifier.none.fl_str_mv 10.1016/j.bspc.2024.106211
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/CAT-Net_Convolution_attention_and_transformer_based_network_for_single-lead_ECG_arrhythmia_classification/26403739
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Biological sciences
Bioinformatics and computational biology
Biomedical and clinical sciences
Cardiovascular medicine and haematology
Engineering
Biomedical engineering
Information and computing sciences
Machine learning
ECG
CNN
Transformer
Arrhythmia
Attention
Deep learning
dc.title.none.fl_str_mv CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p>Machine learning technologies have been applied extensively in the last decade to automatically detect and analyze various forms of arrhythmia from electrocardiogram (ECG) signals. Existing deep learning-based models focus on enhancing classification performance by exploring spatial–temporal ECG features or by implementing multi-modal and ensemble classifiers. Such approaches perform well but do not provide comprehensive accessibility for real-life applications due to the multi-lead ECG requirement. To address the issue, a single-lead ECG based network is considered. The proposed convolution, attention, and transformer-based network (CAT-Net) exhibits promising performance on arrhythmia classification by adeptly capturing local and global heartbeats’ morphological characteristics. Along with the local information captured by the convolution layer, the contextual ECG information is extracted by the multi-head attention layer present in the transformer encoder. In addition, most of the existing models suffer from lower predictive performance in minority-class arrhythmias due to the highly imbalanced ECG data. To enhance the predictive performance in minority classes, three balancing techniques – SMOTE-ENN, SMOTE-Tomek, and ADASYN – are systematically evaluated, and SMOTE-Tomek is ultimately integrated. To mitigate potential dataset bias, CAT-Net was assessed across two distinct datasets: MIT-BIH and INCART, respectively, and was shown to achieve state-of-the-art performance. CAT-Net establishes new benchmarks, achieving 99.14% overall accuracy and 94.69% macro F1 score on 5-class arrhythmia classification in the MIT-BIH dataset and 99.58% accuracy with 96.15% macro F1 score for 3-class classification in the INCART dataset.</p> <h2>Other Information</h2> <p>Published in: Biomedical Signal Processing and Control<br> License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br> See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.bspc.2024.106211" target="_blank">https://dx.doi.org/10.1016/j.bspc.2024.106211</a></p> <p>Additional institutions affiliated with: Electrical and Computer Engineering - TAMUQ</p>
eu_rights_str_mv openAccess
id Manara2_99fafd79e28093f376371fcd8b4ccf1d
identifier_str_mv 10.1016/j.bspc.2024.106211
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/26403739
publishDate 2024
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classificationMd Rabiul Islam (6424796)Marwa Qaraqe (10135172)Khalid Qaraqe (16896504)Erchin Serpedin (3706543)Biological sciencesBioinformatics and computational biologyBiomedical and clinical sciencesCardiovascular medicine and haematologyEngineeringBiomedical engineeringInformation and computing sciencesMachine learningECGCNNTransformerArrhythmiaAttentionDeep learning<p>Machine learning technologies have been applied extensively in the last decade to automatically detect and analyze various forms of arrhythmia from electrocardiogram (ECG) signals. Existing deep learning-based models focus on enhancing classification performance by exploring spatial–temporal ECG features or by implementing multi-modal and ensemble classifiers. Such approaches perform well but do not provide comprehensive accessibility for real-life applications due to the multi-lead ECG requirement. To address the issue, a single-lead ECG based network is considered. The proposed convolution, attention, and transformer-based network (CAT-Net) exhibits promising performance on arrhythmia classification by adeptly capturing local and global heartbeats’ morphological characteristics. Along with the local information captured by the convolution layer, the contextual ECG information is extracted by the multi-head attention layer present in the transformer encoder. In addition, most of the existing models suffer from lower predictive performance in minority-class arrhythmias due to the highly imbalanced ECG data. To enhance the predictive performance in minority classes, three balancing techniques – SMOTE-ENN, SMOTE-Tomek, and ADASYN – are systematically evaluated, and SMOTE-Tomek is ultimately integrated. To mitigate potential dataset bias, CAT-Net was assessed across two distinct datasets: MIT-BIH and INCART, respectively, and was shown to achieve state-of-the-art performance. CAT-Net establishes new benchmarks, achieving 99.14% overall accuracy and 94.69% macro F1 score on 5-class arrhythmia classification in the MIT-BIH dataset and 99.58% accuracy with 96.15% macro F1 score for 3-class classification in the INCART dataset.</p> <h2>Other Information</h2> <p>Published in: Biomedical Signal Processing and Control<br> License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br> See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.bspc.2024.106211" target="_blank">https://dx.doi.org/10.1016/j.bspc.2024.106211</a></p> <p>Additional institutions affiliated with: Electrical and Computer Engineering - TAMUQ</p>2024-03-13T12:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1016/j.bspc.2024.106211https://figshare.com/articles/journal_contribution/CAT-Net_Convolution_attention_and_transformer_based_network_for_single-lead_ECG_arrhythmia_classification/26403739CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/264037392024-03-13T12:00:00Z
spellingShingle CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
Md Rabiul Islam (6424796)
Biological sciences
Bioinformatics and computational biology
Biomedical and clinical sciences
Cardiovascular medicine and haematology
Engineering
Biomedical engineering
Information and computing sciences
Machine learning
ECG
CNN
Transformer
Arrhythmia
Attention
Deep learning
status_str publishedVersion
title CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
title_full CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
title_fullStr CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
title_full_unstemmed CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
title_short CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
title_sort CAT-Net: Convolution, attention, and transformer based network for single-lead ECG arrhythmia classification
topic Biological sciences
Bioinformatics and computational biology
Biomedical and clinical sciences
Cardiovascular medicine and haematology
Engineering
Biomedical engineering
Information and computing sciences
Machine learning
ECG
CNN
Transformer
Arrhythmia
Attention
Deep learning