Improving the performance of an EEG-based motor imagery brain computer interface using task evoked changes in pupil...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) offer a communication pathway for individuals with severe motor impairments.
• Current EEG-based motor imagery BCIs lack the accuracy of traditional control methods.
• There is a critical need to enhance the reliability and accuracy of BCIs for practical application.

Purpose of the Study:

• To investigate the efficacy of incorporating pupil diameter as a psycho-physiological parameter to improve EEG-based motor imagery BCI accuracy.
• To assess the potential of multi-modal sensing in overcoming limitations of traditional EEG BCIs.

Main Methods:

• A user study with 30 participants was conducted using a standard EEG-based motor imagery BCI setup.
• Common Spatial Patterns (CSP) were employed to differentiate motor imagery from a resting state.
• Pupil diameter, measured via video oculography, was integrated as an additional feature alongside EEG data.

Main Results:

• The classification accuracy of discriminating motor imagery from a control condition was significantly improved by including pupil diameter as a feature.
• The multi-modal approach demonstrated superior performance compared to using EEG-derived features alone.
• Pupil diameter changes were observed to correlate with motor imagery tasks.

Conclusions:

• Monitoring pupil diameter alongside EEG presents a promising strategy for enhancing BCI accuracy and robustness.
• Multi-modal approaches integrating psycho-physiological parameters can mitigate challenges like long training times and poor signal quality in BCIs.
• This research paves the way for more reliable and accessible BCI systems for individuals with motor disabilities.