Focusing on the application of linearly-constrained adaptive beamformer techniques to fMRI-MEG integrative analysis, we have examined an approach which determines the center of gravity in each fMRI activated cluster as the location of the linear constraints. First, we simulated MEG data generated from two sources distributed in circular forms placed at visual areas V2 and V5, which are temporally strongly correlated. Two fMRI activated clusters are assumed to cover the two sources. The linear constraints for the voxels in a cluster are defined to suppress the power from the source at the center of gravity of the other cluster. The spatial distributions and the time courses of the two correlated sources at V2 and V5 are successfully reconstructed. We applied the approach to data obtained during an apparent motion perception task and were able to confirm its availability. These results demonstrate that the proposed approach based on linearly-constrained adaptive beamformer techniques is promising as an fMRI-MEG integrative analysis.
