Dear Paul,
On Tue, Feb 1, 2011 at 5:35 AM, Paul Sowman <[log in to unmask]> wrote:
> I am doing some source reconstruction on paediatric MEG. I do not have structural MRIs. I do have polhemus head shapes.
>
> 1. Is it better to use a paediatric template?
What really matters for MEG is the size and position of the head. This
is taken into account just based on the fiducials since the head is
scaled to fit between them. This is of course not very precise and you
can definitely improve things if you have a Polhemus head shape. I
don't think that using a paediatric template will improve things
further. The way you do it it'll even make them worse because SPM
thinks that this is an individual head and does not scale it so the
variance in head size between children is not accounted for. What you
could do is replace the standard template files in spm/canonical with
your paediatric template files but the head will be scaled anyway. The
only advantage is that if you replace both the template image and the
canonical mesh you'll get output corresponding to a smaller head. Look
in spm_eeg_inv_mesh to see where the template file names are
specified.
> 2. I have constructed one using TOM toolbox and used it by selecting MRI from the inversion GUI instead of template. If this is a correct approach that I can use for each subject how do I avoid re-running the processing of the MRI each time that is invoked by spm_eeg_inv_mesh_ui?
As I mentioned this approach is problematic the way you are doing it
now. If you select the same MRI and there is a y_... file in the same
directory, there will be no re-processing and the pre-computed
deformation field will be used.
> 3. Are the hardcoded fiducial locations now in the correct place?
They should be, but there is always some distortion. You can get the
fiducials from D.inv{..}.mesh.fid.fid.pnt and look where they are on
the image. If you want to be really precise I'd look at your
individual image (paediatric template) carefully, record the locations
and use the 'type' option. If you are using Polhemus headshape it
perhaps doesn't matter that much because the final fit is only based
on the head shape and not on the fiducials.
> 4. Should I use the polhemus headshape data?
Yes, I'd just use the standard template with Polhemus.
Vladimir
|