Spatial Resolution of EEG Source Reconstruction in Assessing Brain Connectivity Analysis.

Brain connectivity analysis has emerged as a tool to associate activity generated in diverse brain areas, making possible the integration of functionally specialized brain regions in networks. However, estimation of the areas with relevant activity is well influenced by the applied brain mapping methods. This paper carries out the comparison of three reconstruction principles that differ in the way the prior covariance is adjusted, including its generalization through multiple and sparse spatial priors. To cluster the locations with significant brain activity (regions of interest), we select the most powerful areas, for which the functional connectivity is measured by the coherence and Kullback-Liebler divergence. From the obtained results on simulated and real-world EEG data, both measures show that the mapping method that includes Multiple Sparse Priors allows improving the connectivity accuracy regardless the used measure for all tested values of added noise.