Premium
Identification method of independent components related to artifacts in electroencephalograms
Author(s) -
Hiratsuka Shun,
Hayasaka Daisuke,
Kato Kazuo,
Kadokura Hiroyuki
Publication year - 2019
Publication title -
ieej transactions on electrical and electronic engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.254
H-Index - 30
eISSN - 1931-4981
pISSN - 1931-4973
DOI - 10.1002/tee.23010
Subject(s) - artifact (error) , artificial intelligence , independent component analysis , electroencephalography , pattern recognition (psychology) , computer science , eye movement , kurtosis , computer vision , cluster analysis , noise (video) , signal (programming language) , identification (biology) , speech recognition , psychology , mathematics , image (mathematics) , neuroscience , botany , biology , statistics , programming language
When an electroencephalogram (EEG) is measured, artifacts due to eye blinks and eye movements contaminate the EEG signal. One of the solutions applied to this problem is independent component analysis (ICA), which can separate the artifact components from EEG data. However, the identification of independent components (ICs) related to artifacts, particularly in the case of multiple noise signals from eye blinks and movements, has not been investigated fully. The purpose of this work is to identify ICs related to eye‐blink and eye‐movement artifacts in EEG data. In the proposed method, an artificial EEG with 28 channels was first created by superimposing eye‐blink and eye‐movement templates, and then ICA was applied to the artificial EEG to obtain ICs. Afterward, feature amounts of kurtosis for each IC and the cross‐correlation coefficient between each IC and the eye‐blink template and eye‐movement template were extracted, and clustering was performed by K ‐means clustering. As a result, the most reasonable ICs related to multiple artifacts were identified in comparison with the results of round‐robin ICs. © 2019 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.