Blind identification of brain mechanism in MEG

By the decorrelation method of BSS, two components related to a magnetic SEF (somatosensory evoked field) are selected from MEG data. A remixing matrix is applied to the two selected components to retrieve SEF MEG signals. The brain mechanism of the SEF MEG data is identified equivalently by an innovation model from the viewpoint of a statistical inverse problem. If the innovation model has a feedback structure corresponding to brain regions, the transfer functions between two regions are evaluated via the innovation model. Feedback paths of transfer functions are checked by taking advantage of the scaling transformation of SEF MEG data. Since the transfer functions of real paths are invariant for scaling transformation, paths or routes corresponding to a brain mechanism are diagnosed by examination of identified transfer functions.