Hybrid EEG-fNIRS BCI Fusion Using Multi-Resolution Singular Value Decomposition (MSVD)

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Single-modality Brain-Computer Interfaces (BCIs) have limitations.
• Hybrid BCIs combining electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) offer improved reliability.
• Current hybrid BCIs show modest performance gains due to limited computational fusion methods.

Purpose of the Study:

• To develop and evaluate a novel computational approach for fusing EEG and fNIRS signals in a hybrid BCI.
• To enhance task classification accuracy by effectively integrating multi-modal brain data.
• To overcome the limitations of uni-modal BCIs and modest improvements in existing hybrid systems.

Main Methods:

• A hybrid BCI system was developed using concurrently recorded EEG and fNIRS signals.
• A novel Multi-resolution Singular Value Decomposition (MSVD) approach was proposed for system- and feature-based fusion.
• The proposed MSVD fusion was compared against other methods using KNN and Tree classifiers on multiple datasets.

Main Results:

• The proposed MSVD approach effectively fuses EEG and fNIRS modalities.
• Significant improvements in classification accuracy were observed using the novel fusion technique.
• The MSVD-based fusion demonstrated superior performance compared to existing methods.

Conclusions:

• The novel MSVD-based fusion approach enhances the performance of hybrid EEG-fNIRS BCIs.
• This method offers a promising solution for overcoming the limitations of single-modality BCIs.
• The findings suggest a pathway towards more reliable and accurate multi-command BCIs.