Dear Simone,
Just a little note to add to Geraint's answer, which I agree with entirely.
If you have a prior hypothesis, it is perfectly reasonable to specify it
(beforehand!) in terms of a sphere, box and / or image, which enclose the area
that you are interested in, and choose the one of these that gives you the most
significant p value. For example, sometimes an image VOI can give a more
conservative correction than a sphere of greater volume.
A common mistake, which can be inadvertently encouraged by the SPM interface,
is to (correctly) decide a sphere radius beforehand, find your peak voxel, and
choose that as your sphere radius centre. Obviously this is biased, and you
have to choose the sphere/box radius _and_ centre a priori - perhaps from a
peak voxel from a previous paper, an orthogonal contrast, or on anatomical
grounds.
Best regards,
Matthew
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|