Three-dimensional EEG source imaging via maximum entropy method

Electroencephalographic imaging or the estimation of three-dimensional dipole current sources on the cortical surface from scalp measured potential distribution is a highly underdetermined inverse problem as there are many `feasible´ images which are consistent with the EEG data. Here, the authors explore the maximum entropy approach to obtain better solutions to this problem. This estimation technique selects that image from the possible set of feasible images which is maximally smooth and most likely to give rise to the measured potential field. The authors have also incorporated a noise-rejection mechanism in their scheme that allows them to estimate the noise component of the recorded potentials. Computer simulation demonstrates that the proposed method can reconstruct three-dimensional current distribution where the conventional least-squares minimum-norm method fails. A preliminary testing of this method on evoked potential data measured during visual checkerboard stimulation at 3 Hz yielded sources largely concentrated in the posterior areas of the brain including a region roughly corresponding to the primary visual cortex