Dear Tianhao
> I am sorry for posting my question again, but I really need your help.
> Otherwise my professor won't give me salary.
Oh dear!
> My question is about ROI analysis. (1) Can I do RFX within ROI without
> enough subjects? I am not familiar with those fomula and I just guess
> the sensitivity maybe improve within smaller volume. (2) I understand
> one method is S.V.C for analyzing ROI now, but what is standard to
> define the size of ROI? How to get the t-map with corrected p-value
> within ROI? (3) Do anybody know how to get the special 'mask.img'. such
> as visual cortex, auditory cortex etc?
1) Yes.
2) The standard size for the VOI comprising a search volume is a sphere
of between 8 and 16mm radius (depending on the size of the
area/structure you have a hypothesis about). I think Keith Worlsey has
a table of typical volumes in his unified theory paper. Remember the
SPM{T} is exactly does not change. The only thing that changes with the
search volume is the corrected p value.
3) You do not need a mask.img to do a S.V.C. However, I am sure masks
do exist - Mathew Brett in Cambridge has expertise in these masked
corrections (among other things).
I hope this helps - Karl
|