The accuracy of localizing equivalent dipoles and the spatio-temporal correlations of background EEG.

Abstract

For the inverse problem of equivalent dipole localization, a new residual function was proposed which is based on spatio-temporal correlation of background electroencephalogram (EEG). This residual has the advantage that it allows the calculation of a confidence region for estimated dipole parameters. This method was applied to two sets of visual evoked potential (VEP) data. The localization was compared by using the volume of the confidence region. The outcome of the equivalent dipole localization was compared for three different residual functions: 1) least square; 2) based on spatial correlations in the background EEG; and 3) the proposed new function which is based on spatial and temporal correlations in the background EEG. It was found that the proposed residual function leads us to the highest accuracy and the fastest convergence in the equivalent dipole localization and that even for two-dipole localization, the present method yields more accurate solutions with less iterations than the conventional methods.