> Since non-stationarity correction is quite important in VBM and
> another popular SPM extension, NS was also not compatible with SPM8,
> we needed some good idea for dealing with such.
Recent work has relativized somewhat the capability of existing
non-stationary corrections to address the problem. You may want to
have a look at
Salimi-Khorshidi et al. 2011, doi:10.1016/j.neuroimage.2010.09.088
I have had the problem with perfusion imaging,
doi:10.1016/j.neuroimage.2010.10.041
but the fact that it is particularly evident in perfusion imaging does
not mean that it does not occur in VBM; it just means that no one has
looked.
Essentially, non-stationary corrections based on random field theory
only correct for unequal smoothness. However, this is only one aspect
of nonstationarity; smoothness is only defined based on the
correlation between adjacent voxels; stationarity in general is
defined over all voxels including those beyond the immediate neighbours.
Best wishes,
Roberto Viviani
Dept. of Psychiatry III
University of Ulm
|