Position estimation of electrocortical generators

Using subspace techniques we present simulated results from investigating variations in the geometry of the sensor array for EEG measurements. We show that the performance of the subspace techniques degrades as the sources are brought closer to the array. This degradation can be counteracted by changing the curvature of the array. An optimum array curvature exists which exhibits the best detection performance for a given angle of arrival. We also present preliminary results applying subspace techniques to a sample of real EEG data