Motor imagery based brain computer interface using transform domain features.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain Computer Interface (BCI) systems facilitate communication via brain electrical activity.
• Improving BCI performance requires effective feature extraction from electroencephalogram (EEG) signals.

Purpose of the Study:

• To enhance Brain Computer Interface (BCI) classification accuracy and mutual information.
• To explore novel features derived from EEG signals for improved BCI performance.

Main Methods:

• Extracted features include discrete Fourier transform magnitude, wavelet coefficients, distance series values, and invariant moments from EEG.
• Utilized various preprocessing, feature selection, and classification schemes.
• Evaluated system performance on dataset III from BCI competition II.

Main Results:

• Achieved a maximum classification accuracy of 90.7%.
• Obtained a maximum mutual information of 0.76 bit.
• Distance series features demonstrated superior performance.

Conclusions:

• Novel features, particularly distance series, significantly boost BCI system performance.
• The proposed feature extraction methods offer a promising approach for advanced BCI applications.
• Further research can explore these features for diverse BCI tasks.