A novel feature extraction method PSS-CSP for binary motor imagery - based brain-computer interfaces.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) are crucial for assistive technologies.
• Electroencephalography (EEG) is a common modality for BCI signal acquisition.
• Motor imagery (MI) tasks are widely used in binary BCIs.

Purpose of the Study:

• To propose a novel feature extraction method for binary MI-based BCIs.
• To improve the performance of EEG-based BCIs using a combined approach.
• To enhance classification accuracy in motor imagery tasks.

Main Methods:

• A novel method, power spectral subtraction-based common spatial pattern (PSS-CSP), was developed.
• Spectral subtraction was employed for denoising EEG signals.
• The PSS-CSP method calculates power spectrum differences between binary EEG classes for feature extraction.
• Support Vector Machine (SVM) was used for signal classification.

Main Results:

• The proposed PSS-CSP method demonstrated superior performance compared to existing methods.
• Achieved classification accuracy of 76.8% on the BCIIV dataset 2b.
• Attained classification accuracies of 76.25% and 77.38% on the OpenBMI dataset sessions 1 and 2, respectively.

Conclusions:

• The PSS-CSP method offers a significant improvement for binary MI-based BCIs.
• This novel approach enhances EEG signal processing and feature extraction for better BCI performance.
• The findings suggest the potential of PSS-CSP for practical BCI applications.