Abstract

This paper investigates the fusion of ECG with EEG in devising a hybrid brain-computer interface (hBCI). Effortful motor imagery (MI) based BCI experiments were arranged with a twelve seconds of cue-based MI paradigm on six healthy individuals over two sessions of 160 trials, while ECG and EEG signals were simultaneously recorded. The proposed hBCI uses bispectrum based features from EEG and ECG along with an LDA classifier. The off-line analysis shows an improvement in MI task detection accuracy if both ECG and EEG features are considered together. In addition, the time domain analysis of ECG signal shows that the average heart rate increases during MI state, which clearly shows that the cardiac system responds to MI related tasks.