Neonatal seizure detection from raw multi-channel EEG using a fully convolutional architecture

Show simple item record

dc.contributor.author O'Shea, Alison
dc.contributor.author Lightbody, Gordon
dc.contributor.author Boylan, Geraldine
dc.contributor.author Temko, Andriy
dc.date.accessioned 2020-01-06T12:37:57Z
dc.date.available 2020-01-06T12:37:57Z
dc.date.issued 2019-11-30
dc.identifier.citation O'Shea, A., Lightbody, G., Boylan, G. and Temko, A. (2019) 'Neonatal seizure detection from raw multi-channel EEG using a fully convolutional architecture', Neural Networks, 123, pp. 12-25. doi: 10.1016/j.neunet.2019.11.023 en
dc.identifier.volume 123 en
dc.identifier.startpage 12 en
dc.identifier.endpage 25 en
dc.identifier.issn 0893-6080
dc.identifier.uri http://hdl.handle.net/10468/9452
dc.identifier.doi 10.1016/j.neunet.2019.11.023 en
dc.description.abstract A deep learning classifier for detecting seizures in neonates is proposed. This architecture is designed to detect seizure events from raw electroencephalogram (EEG) signals as opposed to the state-of-the-art hand engineered feature-based representation employed in traditional machine learning based solutions. The seizure detection system utilises only convolutional layers in order to process the multichannel time domain signal and is designed to exploit the large amount of weakly labelled data in the training stage. The system performance is assessed on a large database of continuous EEG recordings of 834h in duration; this is further validated on a held-out publicly available dataset and compared with two baseline SVM based systems. The developed system achieves a 56% relative improvement with respect to a feature-based state-of-the art baseline, reaching an AUC of 98.5%; this also compares favourably both in terms of performance and run-time. The effect of varying architectural parameters is thoroughly studied. The performance improvement is achieved through novel architecture design which allows more efficient usage of available training data and end-to-end optimisation from the front-end feature extraction to the back-end classification. The proposed architecture opens new avenues for the application of deep learning to neonatal EEG, where the performance becomes a function of the amount of training data with less dependency on the availability of precise clinical labels. en
dc.format.mimetype application/pdf en
dc.language.iso en en
dc.publisher Elsevier Ltd. en
dc.relation.uri http://www.sciencedirect.com/science/article/pii/S0893608019303910
dc.rights © 2019, Elsevier Ltd. All rights reserved. This manuscript version is made available under the CC BY-NC-ND 4.0 license. en
dc.rights.uri https://creativecommons.org/licenses/by-nc-nd/4.0/ en
dc.subject Convolutional neural networks en
dc.subject EEG en
dc.subject Neonatal seizure detection en
dc.subject Weak labels en
dc.title Neonatal seizure detection from raw multi-channel EEG using a fully convolutional architecture en
dc.type Article (peer-reviewed) en
dc.internal.authorcontactother Gordon Lightbody, Electrical & Electronic Engineering, University College Cork, Cork, Ireland. +353-21-490-3000 Email: g.lightbody@ucc.ie en
dc.internal.availability Full text available en
dc.check.info Access to this article is restricted until 24 months after publication by request of the publisher. en
dc.check.date 2021-11-30
dc.date.updated 2020-01-06T12:29:46Z
dc.description.version Accepted Version en
dc.internal.rssid 500172572
dc.contributor.funder Science Foundation Ireland en
dc.description.status Peer reviewed en
dc.identifier.journaltitle Neural Networks en
dc.internal.copyrightchecked Yes
dc.internal.licenseacceptance Yes en
dc.internal.IRISemailaddress g.lightbody@ucc.ie en
dc.relation.project info:eu-repo/grantAgreement/SFI/SFI Research Centres/12/RC/2272/IE/Irish Centre for Fetal and Neonatal Translational Research (INFANT)/ en
dc.relation.project info:eu-repo/grantAgreement/SFI/SFI Research Infrastructure Programme/15/RI/3239/IE/INFANT Discovery Platform/ en
dc.identifier.eissn 1879-2782


Files in this item

Files Size Format View

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record

© 2019, Elsevier Ltd. All rights reserved. This manuscript version is made available under the CC BY-NC-ND 4.0 license. Except where otherwise noted, this item's license is described as © 2019, Elsevier Ltd. All rights reserved. This manuscript version is made available under the CC BY-NC-ND 4.0 license.
This website uses cookies. By using this website, you consent to the use of cookies in accordance with the UCC Privacy and Cookies Statement. For more information about cookies and how you can disable them, visit our Privacy and Cookies statement