Movement imagery classification in EMOTIV cap based system by Naïve Bayes

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Brain-computer interfaces (BCI) offer communication and control solutions for individuals with motor impairments.
• BCI systems translate brain activity into commands, bypassing the need for physical movement.

Purpose of the Study:

• To classify imaginary left and right hand movements using an EMOTIV cap-based system.
• To evaluate the efficacy of Naïve Bayes classification with Common Spatial Pattern filters for EEG signal processing.
• To compare results with existing BCI research and a benchmark dataset.

Main Methods:

• EEG data acquired using an EMOTIV EPOC cap from two subjects.
• Common Spatial Pattern (CSP) filters applied for signal processing.
• Naïve Bayes classifier used for classifying imaginary hand movements (left vs. right).
• Analysis of electrode positions (FC5, FC6, P7, P8) and comparison with (C3, C4, P3, P4).

Main Results:

• Achieved a maximum classification accuracy of 79% for the proposed experiment.
• Attained 85% classification accuracy on Dataset 3 of the BCI Competition II.
• Demonstrated satisfactory classification rates with the selected electrode placements.

Conclusions:

• The chosen electrode positions (FC5, FC6, P7, P8) are suitable for BCI applications.
• The study validates the potential of EEG-based BCI for classifying imagined movements.
• Further research can explore these electrode placements for enhanced BCI performance.