Dear Christine,
> Another minor issue. As you probably know, the template creation takes a long time to run. DO you think it is absolutely necessary to make the template based on all subjects? I am just imaging the senario, where an ongoing study collected data on 20 subjects (10 in each group), and plans to collected some more. Is it justifiable to make the template based on these 20 and them applied to the last 10, since they are collected based on the same sequence.
I would include all of your subjects in the template, even though it
will take some time. Your results will be more accurate, so the extra
time seems a small price to pay.
The only caveat occurs in cases where you have unbalanced groups—for
example, if you have 30 controls and 20 patients. In this case I
would suggest making a template of all 20 patients, and a
randomly-selected 20 controls, so that the template is not biased
towards one group or the other.
> Another related problem I have was on "New segment". When I used it with brains with sever atrophy, it inflate gray matter in the c1* images (filled in gray matter where it should be atrophied), whereas "segment" produce reasonable segmentation. I wonder whether anyone here had similar experience.
I have definitely observed that the "new segment" produces
qualitatively different results than the standard segmentation.
However, in theory it is more accurate. That being said, I don't know
if it has been evaluated in cases of severe atrophy, so it could be
that there are ways to optimize it a bit better. I would be very
interested to hear others' experiences. Perhaps using an approach
that doesn't rely on tissue probability maps derived from healthy
brains would be more accurate? It may be worth looking at Christian
Gaser's VBM toolbox, which takes this approach:
http://dbm.neuro.uni-jena.de/vbm/
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|