Discussion of nonlinear local optimization methods for EEG inverse problem solution

The EEG inverse problem solution is an important issue in EEG research. Nonlinear local optimization methods, such as Simplex and Levenberg-Marquart algorithms, were used to solve EEG dipole source localization problems. The relationship between localizing errors and noise level was discussed on the condition that the source number was known. If the source number was unknown, the selected number in model might not equal to the actual one, then a computation was carried out and a corresponding discrimination criteria was proposed. Approximation accuracy between the simulation measurements and dipole-generated potentials was given. The computation speed and the localizing results between the two algorithms were compared. Computer simulation demonstrates that nonlinear local optimization methods are effective for EEG inverse solution if the source number is one or two and the initial iterative values are reasonable.