 |
 |
Hi Steve, Thanks for replying. I think I'll separate the analyses. Cheers, Ronan 2008/11/14 Steve Smith <[log in to unmask]> > Hi, > > Are you doing an FSLVBM analysis? In which case I wouldn't expect you to be > using FEAT for the modelling, but using randomise, as per the FSLVBM manual. > In which case you could either run the different modellings as completely > separate analyses, or just use different design.mat and design.con > models/contrasts at the end - I would expect that in this scenario it > shouldn't make too much difference whether you've used individualised > 'study'-specific templates or not. > > Cheers. > > > > > On 13 Nov 2008, at 13:13, Ronan Joseph Kelly wrote: > > Dear FSL experts, >> >> I am conducting a VBM analysis involving the comparison of subjects at >> both between-group and within-subject levels of analysis. I have created my >> study-specific template and design files, and have just finished my >> between-group comparisons. So far it has worked well. >> >> However, my question involves my change of analysis technique. I now need >> to compare across subjects in my repeated measures analysis. For this, do I >> need to set up a new design matrix as if I was conducting a new study (and >> as such copy my necessary files to a new directory), or simply add new EVs >> and contrasts to the old design matrix used for my between-groups contrasts? >> One of my problems is that not all of my subjects have scans at each time >> point, so any longitudinal analysis design matrix that refers to my origdata >> directory would have to take this into account, and ignore those subjects >> that don't have a 'paired scan' as such. Therefore there is more data >> involved in my between-group analysis than my repeated measures. >> >> Intuitively I think I should copy the required images to a new folder, and >> treat it like a new study; create a new study template with the paired scans >> alone (as my old template would be averaged from too much data) and run my >> repeated measures accordingly. Would there be any problems with doing this? >> Is this the correct method in this scenario? As far as I can tell it comes >> down to whether the design.fsf script can handle both methods of >> comparisons. The stat examples on the FEAT webpage don't seem to combine any >> contrast types like this. >> >> Any help would be greatly appreciated! >> >> Cheers, >> Ronan >> >> -- >> Ronan Kelly >> Neuroinflammatory Research Group >> Trinity College Institute of Neuroscience >> University of Dublin, Trinity College >> Dublin 2 >> Ireland >> > > > --------------------------------------------------------------------------- > 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<http://www.fmrib.ox.ac.uk/%7Esteve> > --------------------------------------------------------------------------- > > -- Ronan Kelly Neuroinflammatory Research Group Trinity College Institute of Neuroscience University of Dublin, Trinity College Dublin 2 Ireland