Inverse solution for time-correlated multiple sources using Beamformer method

The Adaptive Beamformer draws attention as an MEG inverse solution for multiple sources. The method has a drawback that we cannot apply it to the temporally correlated sources. We had developed a method to overcome this drawback. Signal-subspace eigenvectors of an original data-covariance matrix was decomposed by the GA-SA method (the hybrid method of Genetic Algorithm and Simulated Annealing) and a new decorrelated data-covariance matrix was reconstructed. The GA-SA method, however, requires the number of components of the eigenvector as a foreknowledge. In this study, we introduced the minimum-norm method, that requires no foreknowledge, instead of the GA-SA method. We carried out a three-dimensional simulation study using the proposed method, and applied to the auditory evoked response field measured by using the Vectorview system. As a result, multiple sources with time-correlations were successfully estimated.