Improved SSVEP Classification Through EEG Artifact Reduction Using Auxiliary Sensors.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Steady-state visual evoked potentials (SSVEPs) are crucial for brain-computer interfaces (BCIs).
• Electroencephalography (EEG) signals in BCI are often corrupted by artifacts from muscle, motion, and eye movements.
• These artifacts significantly degrade BCI performance, especially in individuals with high muscle tension or involuntary eye movements.

Purpose of the Study:

• To develop and evaluate an electroencephalography (EEG) artifact reduction method for brain-computer interface (BCI) systems.
• To improve the signal quality and classification accuracy of SSVEPs by mitigating artifacts.
• To identify the most effective auxiliary channels for artifact suppression.

Main Methods:

• Utilized auxiliary channels (central, frontal, electrooculographic, neck, cheek, jaw) to model interference sources.
• Applied linear regression within 1-second windows for EEG signal cleaning.
• Performed frequency-domain analysis and utilized Support Vector Machine (SVM) and Convolutional Neural Network (CNN) algorithms for SSVEP classification.
• Recorded data during controlled artifact generation and visual stimulation at 7, 8, and 9 Hz.

Main Results:

• Achieved a 9% increase in classification accuracy following artifact removal.
• Identified central (Cz) and jaw channels as most significant for artifact suppression.
• Demonstrated substantial improvement in EEG signal quality and BCI reliability.

Conclusions:

• Auxiliary channels effectively reduce EEG artifacts in SSVEP-based BCIs.
• The developed artifact reduction method enhances BCI system performance and reliability.
• This approach shows promise for real-world BCI applications with improved signal integrity.