A modified artifact subspace rejection algorithm based on frequency properties for meditation detection application

Electroencephalography (EEG) signals provide insights with human brain states such as brain focusing (BF) states and mind wandering (MF) states in contemplative practice such as meditation. Recently, wearable EEG devices have been developed to detect and facilitate such kind exercises. However, wearable EEG devices with limited computing resources suffer from issues like artifacts and noises even more than traditional ones. Succinct and effective algorithms are critical for online detection of mind states with wearable devices. This study proposes a modified scheme of artifact subspace reconstruction (ASR) in frequency domain suitable for multi-channel EEG meditation detection, allowing a consistent property in the preprocessing calculation, which makes online detection of mind states feasible with wearable EEG devices.