Discriminating hand gesture motor imagery tasks using cortical current density estimation.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Current electroencephalography (EEG) based brain-computer interface (BCI) systems offer limited control dimensionality.
• Existing BCI paradigms may not align with naturalistic rehabilitative or recreational use cases.
• There is a demand for realistic motor imagination (MI) tasks for effective BCI control.

Purpose of the Study:

• To classify distinct hand gesture motor imagination (MI) tasks.
• To evaluate a novel EEG inverse imaging approach for enhanced MI classification.
• To improve the naturalness and applicability of BCI systems.

Main Methods:

• Utilized a novel EEG inverse imaging technique.
• Focused on classifying four right-hand MI tasks: flexion, extension, supination, and pronation.
• Employed both temporal and spatial specificity within the source domain for analysis.

Main Results:

• Achieved up to 95% accuracy in binary classification of any two MI tasks.
• Demonstrated superior performance compared to sensor-domain classification, which reached a maximum of 79% accuracy.
• The novel approach successfully separated different hand gesture MI tasks.

Conclusions:

• The novel EEG inverse imaging approach enhances the accuracy of classifying hand gesture MI tasks.
• This method offers a more effective paradigm for BCI control in rehabilitative and recreational applications.
• Improved classification accuracy supports the development of more intuitive and functional BCI systems.