Hi Jon,

this has been discussed before, please have a look at the FSL archives with longitudinal vbm as keywords and my last name from the address.
e.g. https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1112&L=FSL&P=R7483&I=-3&d=No+Match%3BMatch%3BMatches
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1210&L=FSL&P=R61091&I=-3&d=No+Match%3BMatch%3BMatches

Bottom line, you're right, it would be fine... If you have 2 groups, 2 timepoints though, don't forget to take the difference between the 2 timepoints in randomise (cf https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1008&L=FSL&P=R8137&I=-3&d=No+Match%3BMatch%3BMatches).

Hope this helps,
GwenaŽlle
--------------------------------------------------------------------
GwenaŽlle Douaud, PhD
FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington
OX3 9DU Oxford UK
Tel: +44 (0) 1865 222 523
Fax: +44 (0) 1865 222 717
www.fmrib.ox.ac.uk/~douaud
--------------------------------------------------------------------

De : Jon Brock <[log in to unmask]>
ņ : [log in to unmask]
Envoyť le : Vendredi 5 avril 2013 14h31
Objet : [FSL] Longitudinal VBM

Hi,

I would like to carry out a longitudinal VBM study (single group measured at two time points) but as I understand it FSL does not currently support this type of analysis - largely due (i think) to the fact that there is no consensus on  how to generate a suitable template.  What I can't fathom is why a template generated from all your subjects would not be suitable? If I had a unpaired design I could create an unbiased template by using an equal number of subjects from each group and that woud be fine. In a paired design, if I use all my data don't I again produce an unbiased template? I could then normalize all my data to this template and run a paired samples t-test? What am I missing? Cheers.