Hello, Thanks very much for this response. What is the best way to obtain the <yourClusterInSkeleton> values referred to below? Would this involve noting individual coordinates in FSLview or is there some automated method of creating a table of statistically significant clusters for a given skeleton? Thanks, James On 11/19/08 4:26 AM, "Saad Jbabdi" <[log in to unmask]> wrote: Hi James You can get mean FA values _on the skeleton_ for each subject using fslmeants: e.g.: fslmeants -i all_FA_skeletonised -m <yourClusterInSkeleton> -o <outputTextFile> You need to do that on the _non-filled_ cluster. No need for using tbss_deproject if this is all you want. You may also use the option --showall if you want to see FA values in all the voxels of your cluster. Cheers, Saad. On 18 Nov 2008, at 18:11, James Stone wrote: > I am in the processing of transitioning from data analyses with SPM to > analyses with FSL and was hoping for some guidance on tools to use > for a > task at hand. I am in the process of evaluating a set of military > service > members exposed to repetitive low-level explosive blast to determine > if > their fractional anisotropy values change either before or after > training, > or are different between several groups entering this study with a > previous > history of differing levels of exposure. Here are the steps I have > taken so far: > > 1. Acquire diffusion data (in 30 directions and 2.5 mm slices) > 2. Convert dicom to nifti files > 3. Peform eddy correction > 4. Fit diffusion tensors using DTIFIT > 5. Run the tbss_1_preproc, tbss_2_reg -T, tbss_3_postreg, > tbss_4_prestats, > and suggested randomise steps in the TBSS documentation > instructions. I then > ran the tbss_fill step to fill out the local tracts for display > purposes. > > So I now have a dataset which nicely demonstrates where areas of > statistical > significance exist when comparing study groups. What I really need > at this > point to complete the dataset is to be able to go into the specific > clusters > which are significantly different determine corresponding mean FA > values > within clusters for each subject in the study. > > I see the documentation on "Transforming TBSS results back to native > space", > and can certainly follow these instructions, but I am not entirely > sure how > to get to mean FA values within a given cluster of statistically > significantly different voxels. What I am shooting for is data > similar to > the Giorgio et al. Neuroimage article (NeuroImage 39 (2008) 52-61). > Thanks in > advance for any help you might be able to provide with this. > Saad Jbabdi Oxford University FMRIB Centre JR Hospital, Headington, OX3 9DU, UK +44 (0) 1865 222545 (fax 717) www.fmrib.ox.ac.uk/~saad