Removing Muscle Artifacts From EEG Data via Underdetermined Joint Blind Source Separation: A Simulation Study

Abstract

Electroencephalography (EEG) recordings are often contaminated by artifacts from electromyogram (EMG). This artifact not only affects the visual analysis but also strongly impedes its various usages in biomedical research. With a sufficient number of EEG recordings, numerous blind source separation (BSS) methods can be applied to suppress or remove such EMG artifacts. However, in many practical applications (e.g., ambulatory health-care monitoring), the number of EEG sensors is often limited, while conventional BSS methods (e.g., independent component analysis) may fail to work in such cases. Considering the increasing need for acquiring EEG signals in ambulatory environments, we propose a novel underdetermined joint BSS method to remove EMG artifacts from EEG data with a limited number of EEG sensors. The performance of the proposed method is evaluated through numerical simulations in which EEG recordings are contaminated with muscle artifacts. The results demonstrate that the proposed method can effectively remove muscle artifacts while preserving EEG signals successfully.