A Comparison of Conventional and Tri-Polar EEG Electrodes for Decoding Real and Imaginary Finger Movements from One...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Classifying individual finger movements requires high signal-to-noise ratio (SNR) and spatial resolution due to overlapping sensorimotor cortex representations.
• Standard electroencephalography (EEG) with disc electrodes suffers from low SNR and poor spatial resolution.
• Surface Laplacian estimation enhances EEG spatial resolution and selectivity.

Purpose of the Study:

• To evaluate the effectiveness of Tri-polar Concentric Ring Electrodes (TCREs) for improving the classification accuracy of real and imaginary finger movements using EEG.
• To compare the performance of TCREs against traditional disc electrodes in capturing movement-related potentials (MRPs).

Main Methods:

• Movement-related potentials (MRPs) were recorded from 13 subjects using both TCREs and disc electrodes during real and imaginary finger movements.
• Signal analysis included assessing mutual information, coherence, and classification accuracy using temporal EEG data and spectral power changes.
• TCREs were investigated for their ability to automatically estimate the surface Laplacian.

Main Results:

• MRP signals recorded with TCREs exhibited significantly less mutual information and coherence between neighboring locations compared to disc electrodes.
• TCREs yielded higher classification accuracy for both real ([Formula: see text]%) and imaginary ([Formula: see text]%) finger movements compared to disc electrodes.
• Using temporal EEG data as features with TCREs provided superior accuracy over using spectral power changes in specific frequency bands.

Conclusions:

• TCREs significantly enhance the spatial resolution and signal quality of EEG recordings for motor imagery and execution.
• The improved performance of TCREs facilitates more accurate classification of subtle finger movements, outperforming conventional disc electrodes.
• Temporal EEG data features are more effective with TCREs for finger movement classification than spectral power features.