> according to Good et al., 2001 (NeuroImage) and Mechelli et al., 2005
> (Current Medical Imaging Reviews)
> optimised VBM starts with an initial segmentation
Personally, I wouldn't bother with "optimised VBM". I would do the whole
analysis with SPM5b (not forgetting to install the patches). The whole
optimised VBM thing was just a quick fix that used the existing tools within
SPM.
>
> 1. if the brains have enlarged ventricles (e.g. patients with AD) the
> initial segmentation should
> be done with customised a priori maps, right?
If the customised tissue probability maps have been made from a large sample
of subjects, then they should be more representative of the prior
probabilities of tissue class distribution for your subject population, than
the relatively young subjects that constitute the tissue probability maps
released with SPM.
> 2. the second step is normalisation of GM images to GM template. In case of
> AD patients, the same customised
> GM a priori map as in the initial segmentation should be used, right?
Yes.
> 3. the application of the GM normalisation parameters to the whole raw T1
> image will be done with the normalisation
> function "Write normalised only", right?
Yes, but change the defaults so that the spatially normalised images have
smaller voxels.
>
> a last question is how to open *.mat files (the one from normalisation)
> with MatLab Editor. If I open them with
> the MatLab Editor I can only see hieroglyphs. I want to extract the
> normalisation matrix from the *.mat file.
I'm not sure why you get the hieroglyphs. You could try:
P = spm_get(1,'*_sn.mat','Select the mat file');
load(P)
and then see what has been loaded by looking at the variables in the Matlab
workspace. For example, if you type
whos
then you'll see the variables and their sizes. Note that these may include
some variables that were there before you loaded the mat file.
Best regards,
-John
