Motor imagery task discrimination using wide-band frequency spectra with Slepian tapers

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) enable control through neural signals.
• Motor imagery (MI) tasks are commonly used for BCI control.
• Electroencephalography (EEG) is a primary modality for capturing brain activity.

Purpose of the Study:

• To evaluate the efficacy of wide-band frequency spectra (WBFS) features derived from multi-taper (MT) spectral analysis for motor imagery BCIs.
• To compare WBFS features against traditional Mu-Beta spectral features for EEG signal classification.
• To investigate the impact of electrode coverage (central vs. central+parietal) on discrimination accuracy.

Main Methods:

• Acquisition of human scalp EEG signals during left vs. right hand motor imagery tasks.
• Application of multi-taper spectral analysis to extract WBFS features from EEG data.
• Classification of motor imagery tasks using a Naïve Bayesian classifier with WBFS and Mu-Beta features.
• Analysis of EEG data from central and central+parietal electrode subsets.

Main Results:

• WBFS features utilizing MT spectral analysis demonstrated significantly superior performance compared to conventional Mu-Beta spectral features.
• Classification accuracy was significantly higher when using EEG signals from central+parietal electrodes versus central electrodes only.
• The findings indicate a 95% confidence level for the superiority of WBFS features and the benefit of broader electrode coverage.

Conclusions:

• WBFS features combined with MT spectral analysis offer a more effective approach for motor imagery BCIs.
• Incorporating sensory information from parietal regions alongside central regions significantly improves BCI performance.
• This study highlights the potential of advanced spectral feature extraction and optimized electrode selection for enhanced BCI applications.