Dear AFNI experts,

good day, hope all's well. I was wondering whether it is possible to directly convert-normalize anatomical and corresponding functional datasets from a subject's original space to MNI space using AFNI. Am I right in assuming that AFNI does that only via talairach-ing as an intermediate stage? If this is so, I was wondering whether this adds noise (original --> Talairach space) on top of noise (Talairach --> MNI space) as compared with a more direct conversion (original --> MNI)? I might of course be utterly wrong, for which I apologize in advance.

Now, re a more methodologically flavored question, I was wondering if one could use values from structural probabilistic atlases to apply a threshold/cut-off point (say, 50% or higher) to the voxels found to be active in group analyses in Talairach/MNI space. Wouldn't this approach somehow enhance the replicability of the results yielded?

Finally, on something quite a bit unrelated, I was wondering whether it is reasonable to still define a minimum cluster size (say, even 2-3 voxels instead of just 1) for the activations yielded by group analyses employing FDR correction (q < .05)?

Many thanks for your time and patience with my rookie questions

best

Georgios