| |
Abstract:
While fMRI yields high spatial resolution, brain dynamics are
hardly resolved. The combination with EEG and MEG source
reconstruction adds the desired temporal resolution. However,
several problematic constellations may occur: current sources that
do not show up in the fMRI images, fMRI hotspots without a
corresponding current source, and a slight displacement between
fMRI hotspot and source location. To overcome these problems for a
multiple equivalent current dipole source model, dipoles are
loosely fixed to the fMRI hotspots. Their components, timecourses,
and exact locations are fitted. In addition, an unconstrained
dipole is fitted to account for EEG/MEG data unexplained by the
fMRI activity.
For distributed source models, source locations and normals are
constrained to the cortical sheet segmented from anatomical MRI. A
regularized solution of the bioelectromagnetic inverse problem with
a minimal weighted norm of the source strengths is computed. Source
locations are weighted according to their fMRI-induced
significance.
The performance of the proposed methods is demonstrated.
|
