Could the beta rebound in the EEG be suitable to realize a “brain switch”?

Abstract

Objective Performing foot motor imagery is accompanied by a peri-imagery ERD and a post-imagery beta ERS (beta rebound). Our aim was to study whether the post-imagery beta rebound is a suitable feature for a simple “brain switch”. Such a brain switch is a specifically designed brain–computer interface (BCI) with the aim to detect only one predefined brain state (e.g. EEG pattern) in ongoing brain activity. Method One EEG (Laplacian) recorded at the vertex during cue-based brisk foot motor imagery was analysed in 5 healthy subjects. The peri-imagery ERD and the post-imagery beta rebound (ERS) were analysed in detail between 6 and 40 Hz and classified with two support vector machines. Results The ERD was detected in ongoing EEG (simulation of asynchronous BCI) with a true positive rate (TPR) of 28.4% ± 13.5 and the beta rebound with a TPR of 59.2% ± 20.3. In single runs with 30 cues each, the TPR for beta rebound detection was 78.6% ± 12.8. The false positive rate was always kept below 10%. Conclusion The findings suggest that the beta rebound at Cz during foot motor imagery is a relatively stable and reproducible phenomenon detectable in single EEG trials. Significance Our results indicate that the beta rebound is a suitable feature to realize a “brain switch” with one single EEG (Laplacian) channel only.