Automatic Muscle Artifacts Identification and Removal from Single-Channel EEG Using Wavelet Transform with...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Muscle artifacts contaminate electroencephalogram (EEG) signals, compromising accuracy in brain-computer interface (BCI) systems and medical diagnoses.
• Effective artifact removal is crucial for reliable EEG interpretation.

Purpose of the Study:

• To develop a novel, multi-stage method for removing muscle artifacts from EEG signals without losing essential information.
• To enhance the quality of EEG data for improved BCI performance and diagnostic accuracy.

Main Methods:

• A hybrid approach combining wavelet packet decomposition (WPD) with a modified non-local means (NLM) algorithm for EEG signal denoising.
• Utilized a pre-trained classifier for artifact identification, followed by WPD, NLM filtering, and inverse WPD for reconstruction.
• Employed two meta-heuristic algorithms for optimizing filter parameters, ensuring an automated and efficient denoising process.

Main Results:

• The proposed method demonstrated superior performance in removing muscle artifacts from EEG signals.
• Achieved an average mutual information (MI) of 2.9684 ± 0.7045 on real EEG data, indicating high-quality signal reconstruction.
• Outperformed existing denoising techniques in terms of reconstruction quality and automation.

Conclusions:

• The novel multi-stage EEG denoising method effectively removes muscle artifacts while preserving crucial signal information.
• The system's fully automatic nature and superior performance make it a valuable tool for BCI and clinical applications.
• This approach offers a significant advancement in EEG signal processing for both research and diagnostic purposes.