Spatial filter approach for comparison of the forward and inverse problems of electroencephalography and magnetoencephalography.

We present an analysis of the relative information content of cortical current source reconstructions from electroencephalogram (EEG) and magnetoencephalogram (MEG) forward calculations by examining the spatial filters that relate the internal sources with the externally measured electric potentials and magnetic fields. The forward spatial filters are seen to be low-pass functions of spatial frequency and spatial resolution degrades in external measurements. Inverse spatial filters may be used to reconstruct cortical sources from external data, but since they are high-pass functions of spatial frequency, they must be regularized to avoid instabilities caused by noise at higher spatial frequencies. The regularization process limits the spatial resolution of source reconstructions. EEG forward spatial filters fall off at lower spatial frequencies than MEG filters; hence, there is less information available in higher spatial frequencies resulting in lower spatial resolution in inverse reconstructions. The tangential component of the magnetic field provides even higher spatial resolution than can be obtained using the radial component. An accompanying article examines the surface Laplacian for both the EEG and the MEG.