An EEG-fMRI Fusion Analysis Based on Symmetric Techniques Using Dempster Shafer Theory

EEG-fMRI data fusion provides a better insight of the brain activity due to its high spatiotemporal resolution. The current paper presents a new framework on EEG-fMRI data using symmetric data fusion based on Dempster Shafer theory. Basically, symmetric methods require the use of a common theoretical model to explore Electroencephalogram (EEG) and functional Magnetic Resonance Imaging (fMRI) data jointly. Dempster Shafer theory has a multivariate use in resolving problems related to uncertainty. Accordingly, Basic Belief Assignment and the combination rule offered by such theory allow fusing multimodel sources such EEG (temporal modality) and fMRI (spatial modality). In particular, mass functions for each modality have been calculated. Then, the combination rule has been computed. Finally, this measure has been used to detect the activated areas in the brain via clustering using the potential-based hierarchical agglomerative clustering method. Both real auditory and artificial data simulation have been employed to evaluate the performance of the proposed approach. Also, true, false activation rates and Receiver Operating Characteristic (ROC curve) have been used to establish a comparison with jointICA method. The obtained results have clearly shown the ability of the introduced approach to outperform a standard method of data analysis to reveal a better activation map.