Correction of Eye-Movement Artifacts of DC-EEG Signals

Abstract

For neurofeedback applications EEG frequencies below 0.5 Hz including DC-offset are essential. Since artifacts from eye movement can be found in the same frequency band of the so-called DC-EEG, these artifacts must be removed for proper DC-EEG analysis. To remove artifacts from eye movement, simultaneous electrooculograms (EOG) are recorded as an indicator. The clean EEG is then found by subtracting the EOG from the measured EEG. However, most approaches also remove the DC-part of the EEG in the process. We propose an adaptive algorithm to eliminate eye movement artifacts from an EEG in real-time while maintaining the DC-offset utilizing a recursive least squares approach.