IMPROVING SOURCE RECONSTRUCTIONS BY COMBINING BIOELECTRIC AND BIOMAGNETIC DATA

Objectives: A framework for combining bioelectric and biomagnetic data is presented. The data are transformed to signal-to-noise ratios and reconstruction algorithms utilizing a new regularization approach are introduced. Methods: Extensive simulations are carried out for 19 diff erent EEG and MEG montages with radial and tangential test dipoles at different eccentricities and noise levels. The methods are verified by real SEP/SEE measurements. A common realistic volume conductor is use d and the less well known in vivo conductivies are matched by calibrat ion to the magnetic data. Single equivalent dipole fits as well as spa tio-temporal source models are presented for single and combined modal ity evaluations and overlaid to anatomic MR images. Results: Normalize d sensitivity and dipole resolution profiles of the different EEG/MEG acquisition systems are derived from the simulated data. The methods a nd simulations are verified by simultaneously measured somatosensory d ata. Conclusions: Superior spatial resolution of the combined data stu dies is revealed, which is due to the complementary nature of both mod alities and the increased number of sensors. A better understanding of the underlying neuronal processes can be achieved, since an improved differentiation between quasi-tangential and quasi-radial sources is p ossible. (C) 1998 Elsevier Science Ireland Ltd. All rights reserved.