Hi,
It sounds like the simplest solution is to average the data from all
sessions of each subject to give a single image per subject, and feed
this into randomise. You can, for example, use fslroi to extract
chunks of data from all_FA_skeletonised - i.e. extract new 4D files,
one for each subject, and then use fslmaths <4Dinput> -Tmean <output>
to average across those sessions for each subject. Then the randomise
design matrix should be simple and work well.
Cheers.
On 8 Nov 2007, at 08:39, Stefan Rueckriegel wrote:
> Hi FSLer!
>
> A question to tbss within-subject multi-session analysis:
>
> I would like to do a group analysis with multiple sessions per
> subject.
> When setting up a GLM matrix with each subject as an EV (1) I am
> warned to
> set up a matrix that is rank deficient/linear combination.
> Permutating with
> that matrix I dont find clusters of difference, while I do when
> using just
> one session per subject or just treat the multi sessions as
> independent
> images (only groupID connects them - Is that adequate?).
>
> However, I´m not really interested in the timecourse, so I just
> would like
> to use the multi-sessions as a stabilization of intra-subject
> variance, ie.
> compare the mean-skeletons of subjects. Is that possible?
>
> cheers,
> Stefan
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
|