dear JOhn & other SPM experts
sorry for bringing this up again, I know that there has been lots on the
list already - but there are a couple of things about the optimized VBM
protocol in Good et al (2001) that I just can't get straight.
the paper talks about normalising the segemented grey and white matter
images to their respective templates - I assume white.img and gray.img
from the apriori directory. can I assume from the mail below that only
the parameters from the grey matter are applied to the whole-brain
structural files in native space? are the parameters from normalizing
the white matter ever used?
from John Ashburner's recent
mail on the topic...
2) Estimate spatial normalisation parameters from the grey matter image.
The result of the segmentation includes a grey matter image, which can
then
be used to estimate spatial normalisation parameters from by matching it
to a
grey matter template (typically the grey.img in the apriori directory).
You
can estimate better parameter estimates for spatially normalising grey
matter
by estimating them from grey matter, as these images do not include lots
of
other confounding information from other tissues. Disable brain masking
when
you do this, as it would not help.
3) Apply these spatial normalisation parameters to the T1
Use these estimated warps for spatially normalising the original T1 image,
writing the spatially normalised image at a higher resolution (eg.
1x1x1mm).
thanks
chris summerfield
psychology
columbia university
4) Segment the spatially normalised T1.
5) then maybe a few additional steps to clean up the segmented image.
Best regards,
-John
> having read the "new" Good et al NeuroImage (14,21-36,2001)VBM paper,
> and wanting to implement the method myself, I have turned to the group,
> only to be confused by recent posts. If we all pretend for a minute that
> I didn't write spm, then how can I use this method. In particular, as I
> understand it (hmm...) one :
> 1. first affine registerster the T1 to a T1 template (okay, I can manage
> that!)
> 2. segments this into GM and WM volumes (ditto)
> 3. performs a new (non-linear, affine+warp) registration to standard
> space of the GM and WM vols (do I first do an inverse of the first
> transform??how?)
> 4. apply parameters to the native space T1 (okay as long as the GM and
> WM vols were inverse transformed, or can do i apply it to the result of
> 1 above, and if so, doesnt this lead to a double resampling???)
> 5. Segment the wonderfully registered T1 image.
>
> okay, you guys see that I'm floundering here...tips PLEASE!
--
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]
|