OptEF-BCI: An Optimization-Based Hybrid EEG and fNIRS-Brain Computer Interface.

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Machine Learning

Background:

• Individual neuroimaging modalities like electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) have limitations.
• Multimodal data fusion combines complementary information from different sources to overcome these limitations.
• Investigating the synergy between EEG and fNIRS is crucial for advancing neuroimaging research.

Purpose of the Study:

• To systematically investigate the complementary nature of fused EEG and fNIRS features.
• To develop and evaluate an optimization-based feature selection algorithm for multimodal data.
• To enhance classification performance in neuroimaging through efficient feature subset selection.

Main Methods:

• Acquired and preprocessed EEG and fNIRS data from 29 healthy individuals.
• Computed temporal statistical features for each modality separately.
• Fused features and employed a wrapper-based binary enhanced whale optimization algorithm (E-WOA) for optimal subset selection.
• Utilized a support-vector-machine-based cost function for feature selection.

Main Results:

• The proposed E-WOA feature selection method achieved a high classification rate of 94.22 ± 5.39%.
• Classification performance increased by 3.85% compared to the conventional whale optimization algorithm.
• The hybrid classification framework significantly outperformed individual modalities and traditional feature selection methods (p < 0.01).

Conclusions:

• The proposed approach effectively enhances classification performance by selecting the most efficient fused feature subset.
• The study demonstrates the potential efficacy of multimodal EEG-fNIRS fusion with advanced feature selection for neuroclinical applications.
• This methodology offers a promising direction for improving brain activity analysis and diagnostics.