> I am comparing two sets of FLAIR data (patient/control) with each other > (similar to Auer et al., 2001). > > > The details: > The data is normalized (to T1), smoothed, but NOT segmented (into grey, > white matter and CSF) since FLAIR image contrast does not allow for that. > > My question: > I now assume that since I am not dealing with SEGMENTS, but am comparing > FULL BRAIN data, I am NOT " testing for the increased or decreased > probability of a voxel being grey (or white) matter" when comparing the two > groups, > > but which probability would I be assessing with difference maps obtained in > such a fashion? You would be looking for significant differences in the smoothed and spatially normalised FLAIR data. Differences could mean a number of things. These include, differences between FLAIR intensity for homologous tissue types, or differences between the mixture of tissue types within each region (such as a different proportions of GM, WM and CSF). Best regards, -John