Hi,
On 15 Mar 2007, at 22:29, Rob Roach wrote:
> Hi,
>
> We have spgr images taken 5 minutes after gadolinium infusion. They
> are high
> res, 1x1x1, from 3T magnet. We have 24 subjects, each with a pair
> of these
> images, the pair being before and after an intervention. We have no
> a priori
> hypotheses about where there might be gadolinium leakage, and to
> compare all
> 24 subjects pre to post is the approach we would like to take. We have
> experience with Randomise for SIENA voxelwise and in TBSS analyses.
> But even
> though the images are very high resolution, and map nicely to MNI
> space, we
> don't have anything like the TBSS skeleton nor the SIENA edge mask to
> constrain the individual data for input into Randomise.
>
> We have FLIRTED the pairs to each other, and both to MNI. Is there
> any other
> step you would recommend to prepare the images to get the best
> Randomise
> analysis? Like -dil in SIENA voxelwise?
>
> Maybe we're missing something, but it seems that if each image in
> the 4-D
> input image isn't well-aligned to the same space, that true
> differences in a
> specific region will be obscured by comparing the "wrong" voxels
> across
> images. Is that correct?
Indeed that is correct and is crucial question here. If you only use
FLIRT affine normalisation into standard space then the easiest way
to help with this is then to smooth by an appropriate amount, to
"correct" for the cross-subject spatial variability that you want to
discount. If this looks insufficient then you could try using the
IRTK nonlinear reg programs as used in the TBSS scripts - but you'll
have to work out the IRTK syntax yourselves!
Hope this helps - Cheers, Steve.
ps note that you can use a paired t-test model here which is more
sensitive than an unpaired test.
>
> Thanks in advance,
>
> Rob
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|