VOG-enhanced ICA for removing blink and eye-movement artefacts from EEG

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Authors

Colleges, School and Institutes

Abstract

The steady-state visual evoked potential (SSVEP) is reliable for many paradigms such as clinical neuroscience and brain computer interfaces (BCI), providing high information throughput with minimal between-person variations. However, Electrooculogram (EOG) artefacts in Electroencephalography (EEG) signal limit applications with dynamic SSVEP stimuli due to eye movement and blinks. Independent Component Analysis (ICA) is a successful method for removing EOG artefacts. We propose 'Blink VOG-ICA' (BVOG-ICA) - an enhanced ICA algorithm that uses eye tracker video-oculography (VOG) eye movement and blink detection information. It demonstrates improved performance compared to ICA variants Plochl and our previous VOG-ICA when evaluated on matched VOG and EEG data. SSVEP classification accuracy for the post-ICA clean EEG consequently improves 7% and 4% for static and dynamic SSVEP stimuli respectively, suggesting BVOG-ICA as a potentially reliable automatic EOG artefact removal method for SSVEP paradigms.

Details

Original languageEnglish
Title of host publication3rd IEEE EMBS International Conference on Biomedical and Health Informatics, BHI 2016
Publication statusPublished - 21 Apr 2016
Event3rd IEEE EMBS International Conference on Biomedical and Health Informatics, BHI 2016 - Las Vegas, United States
Duration: 24 Feb 201627 Feb 2016

Conference

Conference3rd IEEE EMBS International Conference on Biomedical and Health Informatics, BHI 2016
CountryUnited States
CityLas Vegas
Period24/02/1627/02/16