Detecting the spatiotemporal dynamics of neural activity on the cortical surface

The neurophysiological signals that are recorded in EEG (electroencephalography) and MEG (magnetoencephalography) originate from current flow perpendicular to the cortical surface due to the columnar organization of pyramidal cells in the cortical gray matter. These locations and directions have been used as anatomical constraints for dipolar sources in estimations of neural activity from MEG recordings. Here we extend anatomically constrained beamforming to EEG, which requires a more sophisticated forward model than MEG due to the blurring of the electric potential at tissue boundaries, but in contrast to MEG, EEG can account for both tangential and radial sources. Using computed tomography (CT) scans we create a realistic three-layer head model consisting of tessellated surfaces representing the tissue boundaries cerebrospinal fluid-skull, skull-scalp and scalp-air. The cortical gray matter surface, the anatomical constraint for the source dipoles, is extracted from magnetic resonance imaging (MRI) scans. EEG beamforming is implemented in a set of simulated data and compared for three different head models: single sphere, multi-shell sphere and realistic geometry multi-shell model that employs a boundary element method. Beamformer performance is also analyzed and evaluated for multiple dipoles and extended sources (patches). We show that using anatomical constraints with the beamforming algorithm greatly reduces computation time while increasing the spatial accuracy of the reconstructed sources of neural activity. Using the spatial Laplacian instead of the electric potential in combination with beamforming further improves the spatial resolution and allows for the detection of highly correlated sources.