Browsing by Author "Hasan, Mustafa"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item HYBRID ELECTROENCEPHALOGRAM (EEG) – FUNCTIONAL NEAR INFRARED SPECTROSCOPY (FNIRS) BRAIN-COMPUTER INTERFACE (BCI) CLASSIFICATION OF MOTOR IMAGERY TASKS(2022-01-26) Hasan, Mustafa; Khan, M. UmerHybrid Brain Computer Interface, a combination of two or more neurophysiological signals, is getting attention for its ability to complement a single modality drawbacks and in achieving reliable results by extracting harmonizing features. A hybrid EEG fNIRS BCI system achieved through the fusion of simultaneously recorded functional Near-Infrared Spectroscopy (fNIRS) and Electroencephalography (EEG) signals, is used to overcome the limitations of uni-modality and to achieve high motor tasks classification. Although, hybrid BCI approach enhanced the performance of the system, yet the improvements are still modest because of the lack of computational methods to fuse the two modalities. The contributions of this thesis is twofold: a novel channel selection correlation coefficient approach to select the most representative channels and a novel fusion approach using Multi-resolution singular value decomposition (MSVD). MSVD is utilized to achieve both system-based and feature based fusion for the optimal EEG-fNIRS channels. Another contribution of this thesis is to utilize canonical correlation analysis (CCA) for feature-based fusion. Correlation analysis is used primarily to study the relationship between the two modalities. CCA feature-based fusion improved performance through maximizing the inter-subject covariance across the two modalities. Through simulation results, it can be witnessed that the proposed approaches help to achieve optimal performance while reducing the computational complexity.