Anthony, Veronica, etc.,
Ziad's advice is very good, as usual. Let me just chime in here and
clarify Veronica's remark: "... because with Caret, unlike with
Freesurfer, you use only 6 core(most stable) landmarks to register
individual brains to the atlas."
As some of you already know, this is based on our "anchor down the anatomy
you can count on, but let the rest flow evenly" approach to surface-based
registration. This works well when looking for population anatomical
differences, as I hope you will soon be reading in a forthcoming paper.
Depending on what you're trying to do, other registration methods (e.g.,
Freesurfer's intensity based approach, where the template choice matters
much more) may be more appropriate. Rutvik Desai's Neuroimage paper also
covers what I call "tailored landmarks" using Caret. Argall et al
describe a very coarse landmark set. All approaches have valid
applications: Different maps for different apps.
Also, it is worth noting that in a yet-to-be-published study here was
wustl.edu (Ilana Rosman, Brad Schlaggar, Steve Petersen, in collaboration
with our lab), we found very few anatomical differences between
nine-year-old kids and adolescents. This was with n1=n2=10 (20 total),
and we're currently working to increase n1 and n2.
Donna L. Dierker
(Formerly Donna Hanlon; no change in marital status -- see
[
home.att.net] for details.)
Van Essen Lab
[
brainvis.wustl.edu]