Optimizing Real-Time MI-BCI Performance in Post-Stroke Patients: Impact of Time Window Duration on Classification...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Brain-computer interfaces (BCIs) offer potential for motor neurorehabilitation.
• Balancing classification accuracy and system responsiveness is key for real-time BCI applications.

Purpose of the Study:

• To evaluate the impact of time window duration on classification accuracy and false positive rates in motor imagery BCIs (MI-BCIs).
• To optimize temporal parameters for enhanced MI-BCI system performance and usability.

Main Methods:

• Investigated time window durations on classification accuracy and false positive rates.
• Utilized Linear Discriminant Analysis (LDA), Multilayer Perceptron (MLP), and Support Vector Machine (SVM) classifiers.
• Applied Common Spatial Patterns (CSP) for feature extraction on EEG data from post-stroke patients and the BCI IVa dataset.

Main Results:

• Longer time windows generally improved classification accuracy and reduced false positives across all tested classifiers.
• LDA demonstrated superior performance compared to MLP and SVM.
• An optimal time window of 1-2 seconds was identified as a balance between performance and system delay.

Conclusions:

• Temporal parameter optimization is critical for enhancing the usability of MI-BCI systems in rehabilitation.
• A 1-2 second time window offers a practical trade-off for real-time neurorehabilitation applications.
• Findings support the development of more responsive and accurate BCIs for clinical use.