Subject-specific time-frequency selection for multi-class motor imagery-based BCIs using few Laplacian EEG channels

摘要: The essential task of a motor imagery brain–computer interface (BCI) is to extract the imagery-related features from electroencephalogram (EEG) signals for classifying intentions. However, optimal frequency band and time segment extracting such differ subject subject. In this work, we aim improve multi-class classification reduce required EEG channel in imagery-based BCI by subject-specific time-frequency selection. Our method based on criterion namely Fisher discriminant analysis-type F-score simultaneously select classification. proposed uses only few Laplacian channels (C3, Cz C4) located around sensorimotor area Applied standard dataset (BCI competition III IIIa), our leads better performance smaller deviation across subjects compared state-of-art methods. Moreover , adding artifacts contaminated trials training does not necessarily deteriorate results, indicating that tolerant artifacts.