Hi Markus,
> ... the significant 'blobs' are all far outside the brain ...
This is usually a problem if you haven't used any kind of explicit or
threshold masking. It is briefly mentioned in the SPM-VBM wikibook
http://en.wikibooks.org/wiki/SPM-VBM
in the criticisms section.
To start with, the easiest thing to do is probably to specify an
absolute threshold of 0.05, and to have a look (e.g. with check reg)
at the resulting mask image in the SPM results directory after
estimating the model. You might want to try a threshold of 0.1 or
higher if the mask still looks generous, though this depends a bit on
your data. It's maybe better still to use an explicit mask, if you can
produce one (there is some stuff on this in the mailing list archives)
Best,
Ged.
|