Dear Marc,
You can get a meaningful solution by specifying informative location priors. In VB-ECD the emphasis is on comparing solutions which you cannot do with other methods rather than getting solutions with no prior knowledge.
I don't think it makes sense to fit dipoles to a subset of sensors. You should always use all the sensors as even signal close to the baseline at some sensors is a useful constraint for dipole fitting. Also non-uniform coverage of the head leads to bias in dipole locations.
Best,
Vladimir
On 1 Jun 2010, at 00:30, Marc Recasens <[log in to unmask]> wrote:
> Hi all,
> I have a very naive question regarding dipole fitting with MEG data.
> I have some averaged data converted from fieldtrip which I’d like to analyze in order to localize sources with ECD. I followed all the steps of the source localization (segmentation, co-registration, forward modeling and invert) but all I managed to obtain are weak dipole solutions centered in the middle of the brain -though activation should be located in auditory cortex.
> Now I'm trying to do the same procedure but across hemispheres, analyzing a subset of sensors only. However that seems not to be a good way since I'm not using all the channels and I get an error in the forward modeling step "No good MEG channels were found".
> Does anyone know what is the best way for fitting centered dipoles in a specific subset of sensors?
>
> Thanks a lot!
|