Stefano,
> I am trying to use snpm to analyze the results of a VBM script written by
> John Ashburner. Unfortunately, I think the SnPM_ST file is too big (1.8
> gigabyes!) and the program gives a problem in the results phase (after
> chugging through for almost 3 hours in the computation phase). I attach the
> error message.
>
If you don't perform a cluster size test, then the SnPM doesn't create a
massive SnPM_ST and the results phase no longer requires the file. This
can be done by selecting "No" in the "Collect Supra-Threshold Stats?"
option in the set up.
Actually, I am not sure doing a cluster size test in VBM data is such a
good idea. This is because VBM data can be non-stationary (i.e.
smoothness is not uniform), and clusters tend to be large in smooth
regions and small in rough regions. (See Ashburner & Friston (NeuroImage
11:805-821, 2000) for more details)
Hope this helps!
-Satoru
|