MEG analysis with new spatial filter
Shinpei Okawa, Satoshi Honda · Society of Instrument and Control Engineers of Japan · 2003
The combination of ICA and new spatial filtering method for MEG analysis are proposed, and some numerical experiments were conducted to ascertain the efficiency of the method and the limit of the estimation without temporal and any prior information but only with the physical model of MEG. The new spatial filter is defined with Kullback-Leibler divergence, and succeeds to estimate the maximum of the current density on the exactly or almost accurate position. Here ICA was introduced to estimate the source signal and it has received some attention in MEG signal processing for elimination of artifacts and noises or extraction of the signal occurred by brain activity. And the proposed image reconstruction method inhibited the expansion of the estimated current density distribution, and it gave the more reliable estimated result. Thus, the combination of ICA and the new spatial filter has feasibility of being a powerful tool for MEG analysis.