> again, with regard to the new processing suggested by Catriona Good (et
> al 2001, NeuroImage 14, 21-36), I have a question: I can call
> spm_segment.m on the command line with a normalization matrix which will
> then be applied to the native image I want to segment.
>
> However, this is only applied to the image to achieve a good overlay
> with the a priori maps and is then undone. I wonder if anyone has hacked
> spm_segment.m to prevent this last step (undoing the normalization), so
> that the "optimally normalized", segmented images are written out ?
The idea of Catriona's processing steps is that improved spatial
normalisation can be achieved by estimating the parameters by warping a grey
matter image to match a grey matter template. Because the transforms
estimated by SPM's spatial normalisation are only low frequency (in the order
of about 1000 parameters), higher frequency stuff can't be modelled. Warping
based only on the grey matter helps to improve this by removing the necessity
to match ventricles scalp etc, when it is only the grey matter that really
needs to be matched.
I don't see why the affine transformation estimated from the original MR
image can help spatial normalisation that is based on grey matter.
>
> And also (if this is possible), how can I derive this matrix from an
> individual *_sn3d.mat ? Is it the inverse of the "affine" field ?
I think the matrix you want is derived from:
M = MG*inv(Affine)*inv(MF)
Best regards,
-John
--
Dr John Ashburner.
Wellcome Department of Cognitive Neurology.
12 Queen Square, London WC1N 3BG, UK.
tel: +44 (0)20 78337491 or +44 (0)20 78373611 x4381
fax: +44 (0)20 78131420
http://www.fil.ion.ucl.ac.uk/~john
mail: [log in to unmask]
|