Multi-source localization by genetic algorithm using MEG

We developed a method to estimate multi-dipole sources in human brain by using Genetic Algorithm for source localization of the olfactory magnetoencephalography (MEG). Two current dipoles in each hemisphere of the brain are supposed as a source model. To investigate the ability of this method, source localization in simulation was performed by the presented method. As a result, if the noise level is restricted under 2 percent (S/N/spl ap/20), the source localization of the simulation was obtained with the error less than 1 cm.