Remove Artifacts from a Single-Channel EEG Based on VMD and SOBI.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Portable EEG systems increasingly use few-channel or single-channel recordings, necessitating effective artifact removal.
• Current methods often combine decomposition techniques like Empirical Mode Decomposition (EMD) with Independent Component Analysis (ICA), but face limitations.
• EMD suffers from modal mixing and lacks a strong theoretical foundation, while ICA methods based on high-order statistics are less effective for EMG artifacts.

Purpose of the Study:

• To address the limitations of existing methods for single-channel EEG artifact removal.
• To propose and evaluate a novel artifact removal technique for electroencephalography (EEG) signals.
• To improve the accuracy and efficiency of removing electrooculography (EOG) and electromyography (EMG) artifacts from single-channel EEG.

Main Methods:

• Utilized Variational Mode Decomposition (VMD) for signal decomposition, addressing limitations of EMD.
• Employed Second Order Blind Identification (SOBI) for artifact separation, particularly effective for EMG artifacts.
• Combined VMD and SOBI (VMD-SOBI) for a new artifact removal strategy for single-channel EEG.

Main Results:

• The proposed VMD-SOBI method demonstrated superior performance in removing EOG and EMG artifacts compared to the EEMD-SOBI method.
• Semi-simulation experiments confirmed the enhanced artifact removal capabilities of the VMD-SOBI approach.
• The method effectively preserved essential neural information within the single-channel EEG signals.

Conclusions:

• The VMD-SOBI method offers a significant improvement for artifact removal in single-channel EEG.
• This technique provides a more robust and effective solution for cleaning EEG data acquired from portable systems.
• The findings suggest VMD-SOBI as a promising tool for enhancing the quality of single-channel EEG analysis.