Source localization using a current-density minimization approach

Determining the location of cortical activity from electroencephalographic (EEG) data is important clinically. In this paper, a method is presented which uses the powerful optimization method of simulated annealing in conjunction with a finite-element-based model of the search domain for a single-time slice solution of the EEG-inverse problem. The algorithm highlights a new objective function based on the current-density boundary integral associated with the finite-element formulation as the basis for parameter optimization. In two-dimensional experiments in a shallow tank containing saline, single dipoles are located within 2 mm. Simulations studying the algorithms response to structured noise are also presented. The new objective function is shown to take advantage of the natural framework associated with finite-elements and the results suggest that the approach is capable of resolving dipole locations in simulations and experiments.