| GEJdG:
| Many thanks for the feedback on our structural MRI data in OCD. We are
| mainly interested in looking at regional differences in grey matter
| density in OCD patients compared to controls, so we will use the global
| normalization method you suggest.
|
| We would like to use global normalisation (ANCOVA), but we have not been
| able to do that within the "Compare-populations: 1 scan/subject" option
| within the PET/SPECT models menu. Should we be using a different design?
Try "single subject: conditions and covariates". This should give the
design matrix you are after. The "Compare-populations: 1 scan/subject"
option should give the options you want in the final SPM99 release.
|
| In regard to your suggestion for us to simply use the number of grey
| matter voxels as global, we have not been able to find where we could
| get that measure from for each subject. Could you help us on that?
There is an easier way of doing this. The following script counts the grey
matter voxels and returns the output as a vector that can be entered into
the design matrix. All you need to do is save the function into a file in
your matlab directory (called something like get_globals.m). You can call
the function by:
gl = get_globals;
then you hit the button for setting up your design matrix, and enter the
gl for the user specified globals.
------------------------------------------------
function gl = get_globals
P=spm_get(Inf,'*.img');
V=spm_vol(P);
gl = zeros(length(V),1);
for i=1:length(gl),
for z=1:V(i).dim(3),
img = spm_slice_vol(V(i),spm_matrix([0 0 z]),V(i).dim(1:2),0);
gl(i) = gl(i) + sum(img(:));
fprintf('.');
end;
fprintf('\n');
end;
-------------------------------------------------
| XC:
| I have done an analysis comparing two groups of structural MRI scans.
| However I would like to compare the total amount of grey matter between the
| two groups. I think this used to be possible in previous SPM releases, but
| I'm not sure how to do it (or if it is possible) in SPM99b?
I hope the above script answers this question.
| GEJdG:
| Thanks for the feedback. We have looked at the original image (see
| attached gif file). There is a feature on it which we feel could well
| have been transformed into the square seen in the normalized and
| segmented image we sent before. Would you agree with that?
| MG:
| I tend to agree with your observation though it's hard to be
| precise due to the contrast I get in your gif file. But it certainly
| looks like that area could be segmented into the final image you
| obtained. I wonder what John thinks?
The <Windows>-<Color Editor> option works a treat for improving the
image contrast, but it's difficult to say if the square is artifactual
from the original un-normalised image. If the square is an artifact,
then it is most likely to arise in the segmentation step (and correspond
to one of the relatively high frequency basis functions from the
non-uniformity correction). I think that the equivalent view in the
spatially normalised image (before segmentation) should show if the
segmentation is correctly identifying grey matter.
I hope this has helped,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|