Diagonal loading common spatial patterns with Pearson correlation coefficient based feature selection for efficient...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) offer communication solutions for individuals with nervous system disorders.
• Common Spatial Pattern (CSP) is a standard BCI feature extraction technique but suffers from noise susceptibility and overfitting.
• High-dimensional and irrelevant features hinder effective classifier learning in BCIs.

Purpose of the Study:

• To develop an improved feature extraction method for motor imagery electroencephalography (MI-EEG) signals.
• To enhance the accuracy and robustness of BCIs for individuals with neurological conditions.
• To overcome the limitations of traditional CSP methods in BCI applications.

Main Methods:

• Introduced Regularized CSP with diagonal loading (DL-CSP) for feature extraction.
• Employed Pearson correlation coefficient (PCC) for discriminative MI-EEG feature selection.
• Utilized an ensemble of classifiers: bidirectional long short-term memory (Bi-LSTM), K-nearest neighbors (KNN), and Naïve Bayes (NB).
• Implemented decision-level fusion via majority voting for improved system robustness.

Main Results:

• Achieved classification accuracies of 86.96% (data-1), 91.70% (data-2), and 85.75% (data-3) across three public MI datasets.
• Demonstrated superior performance compared to existing state-of-the-art techniques.
• Validated the effectiveness of DL-CSP and PCC feature selection in MI-EEG analysis.

Conclusions:

• The proposed DL-CSP and PCC feature selection method significantly improves MI-EEG classification accuracy.
• Ensemble classification with decision-level fusion enhances BCI system robustness and performance.
• This approach presents a promising advancement for BCI applications, particularly for individuals requiring alternative communication methods.