AUTHOR=Khan Muhammad Umer , Hasan Mustafa A. H. 

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

JOURNAL=Frontiers in Human Neuroscience

VOLUME=Volume 14 - 2020

YEAR=2020

URL=https://www.frontiersin.org/journals/human-neuroscience/articles/10.3389/fnhum.2020.599802

DOI=10.3389/fnhum.2020.599802

ISSN=1662-5161

ABSTRACT=Brain-computer interface (BCI) multimodal fusion, has the tendency to generate multiple commands in a much more reliable manner by alleviating single modality’s drawbacks. A hybrid EEG-fNIRS BCI system—achieved through a fusion of concurrently recorded functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG) signals—is used to overcome the limitations of unimodality and achieve high motor tasks classification. Although, hybrid approach enhances the performance of the system, yet the improvements are still modest due to the lack of availability of computational approaches to fuse the two modalities.  Here, we propose a novel approach using Multi-resolution singular value decomposition (MSVD) to achieve system- and feature-based fusion. Finally, we compare both fusion approaches based upon Tree and KNN classifiers. The results show that the proposed approach effectively fuses both modalities while reducing the computational load and improving the classification accuracy.