Adapting Artifact Subspace Reconstruction Method for SingleChannel EEG using Signal Decomposition Techniques

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Artifact removal is essential for accurate electroencephalography (EEG) analysis.
• Artifact Subspace Reconstruction (ASR) is effective but limited to multi-channel data.
• Single-channel EEG analysis requires specialized artifact removal techniques.

Purpose of the Study:

• To adapt Artifact Subspace Reconstruction (ASR) for single-channel EEG data.
• To integrate signal decomposition methods with ASR for enhanced artifact removal.
• To evaluate the performance of the proposed single-channel ASR against established methods.

Main Methods:

• Incorporated ensemble empirical mode decomposition (EEMD), wavelet transform (WT), and singular spectrum analysis (SSA) into ASR.
• Decomposed single-channel EEG data into multiple components for ASR application.
• Compared the proposed single-channel ASR with Independent Component Analysis (ICA) on open datasets.

Main Results:

• The proposed single-channel ASR effectively removed artifacts from EEG data.
• The adapted ASR method demonstrated superior performance compared to ICA.
• Validated the technique on two publicly available EEG datasets.

Conclusions:

• The developed single-channel ASR is a powerful tool for artifact removal in EEG analysis.
• This adaptation significantly expands the applicability of ASR to single-channel recordings.
• The findings highlight ASR's potential for improving the quality of neural signal processing.