Map Estimation of EEG Dipole Source Locations for Head Models with Parameter Uncertainty

Multiple, equivalent current dipole localization from the scalprecorded EEG (electroencephalogram) has traditionally been accomplished under the unrealistic assumption of a head model parameterized by known physical parameters, such as tissue layer conductivities. In this paper the MAP estimation method is utilized for the joint estimation of dipole locations and uncertain head (secondary) parameters, which are modeled as random. Simulations are presented for random variations in head layer conductivity and radii for the 4-sphere model to demonstrate the MAP technique as a means of achieving robustness to deviations from an idealized head model.