Hi,
> I have been using spm_smooth to prepare artificial random fields for testing
> purposes. I use spm_smooth (SPM99) on an artificial 3-D volume directly,
> i.e. without saving it on disk as an analyze image. I noticed that the
> results of my testing sort of conform to my theoretical expectations when I
> add extra zero voxels to all sides of my volume (about 3-4 times the FWHM).
That's right; you will get edge effects with the SPM smoothing routines;
see my random fields tutorial script for an example of the same kind of
padding that you were doing:
http://www.mrc-cbu.cam.ac.uk/Imaging/Common/randomtalk.m
Best,
Matthew
|