Applying correlation analysis to electrode optimization in source domain.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) for neurorehabilitation face challenges with extensive electrode use, complicating signal acquisition and motor imagery EEG (MI-EEG) decoding.
• Traditional electrode optimization methods struggle with the limited spatial resolution of scalp EEG.
• Existing EEG source imaging (ESI) techniques for electrode reduction have limitations in accounting for dipole contribution and electrode-dipole correlations.

Purpose of the Study:

• To develop a novel electrode optimization method (ECCEO) for MI-EEG signal processing in BCIs.
• To enhance decoding accuracy and reduce computational load in neurorehabilitation systems.
• To leverage ESI and correlation analysis for more effective electrode selection.

Main Methods:

• Developed the Electrode Correlation Coefficient-based Electrode Optimization (ECCEO) method.
• Mapped scalp MI-EEG to cortical regions using ESI to identify regions of interest (ROIs) based on dipole amplitudes.
• Calculated and ranked Pearson correlation coefficients between ROI dipoles and electrodes to form optimal electrode sets.

Main Results:

• Achieved high average decoding accuracies of 95.99% and 88.30% on two public datasets.
• Demonstrated significant reductions in computational cost by 65% and 56% respectively.
• Validated the effectiveness and statistical significance of the proposed ECCEO method.

Conclusions:

• The ECCEO method offers an effective strategy for optimizing electrode selection in MI-EEG based BCIs.
• This approach enhances decoding performance while substantially decreasing computational demands.
• ECCEO provides a promising advancement for practical neurorehabilitation systems.