Automatic Artifact Removal from Electroencephalogram Data Based on A Priori Artifact Information

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Electroencephalogram (EEG) signals are frequently contaminated by nonneural physiological artifacts.
• Accurate artifact removal is crucial for reliable analysis of brain activity.
• Existing methods often struggle with variability in subjects and acquisition environments.

Purpose of the Study:

• To develop an automatic online artifact removal method for EEG data.
• To enhance the extraction of relevant brain activity information.
• To improve classification accuracy in tasks like motor imagery and emotion recognition.

Main Methods:

• Utilized a combination of discrete wavelet transform and independent component analysis (ICA), termed wavelet-ICA, for artifact component separation.
• Employed a priori artifact information for automatic identification of artifact components.
• Reconstructed artifact-free EEG signals by removing identified artifact components.

Main Results:

• Demonstrated statistically significant improvements in classification accuracies.
• The method proved effective in both motor imagery and emotion recognition experiments.
• The automatic online approach adapted well to variable subjects and environments.

Conclusions:

• The proposed automatic online artifact removal method is effective for enhancing EEG data quality.
• This technique offers a robust solution for analyzing brain activities in diverse conditions.
• Improved artifact removal leads to more accurate brain-computer interface applications.