Facial EMG contamination of EEG signals: Characteristics and effects of spatial filtering

Facial electromyography (EMG) contamination of the electroencephalography (EEG) signals is a largely unresolved issue in brain-computer interface (BCI) research. Artifacts can obscure EEG features used in BCI control. Effective artifact detection and minimization require understanding of the artifacts. This study presents the spectral and the topographical properties of the artifacts caused by jaw clenching and eyebrow raising. Measures are introduced to quantify the effects of the artifacts on the EEG signals. We also compare the effectiveness of three spatial filtering methods (monopolar, small Laplacian and bipolar montage) in reducing the artifacts. Experiments on two subjects recorded the EEG signals during weak and moderate muscle contractions. The results show that the weak and the moderate contractions affect all frequencies at all locations (p < 0.01). This clearly demonstrates that EMG artifact detection and minimization are important not only for the BCIs focused on mu and beta rhythms, but also other BCIs that involve low frequency components. ANOVA analysis reveals that the small Laplacian and the bipolar montage are susceptible to these artifacts. Caution has to be exercised when choosing a spatial filtering method as some may be effective in extracting features but do not perform as well in the presence of artifacts.