Hi Anne,
just to add my five cents to Markos reply. In SPM99 the default was to
use "brain-masking", meaning that only those parts that were considered
to be brain (in the template) was used when estimating the warps. This
typically means that for structural (e.g. T1) images the results are a
little better because in the compromise that SPM (and any basis-function
normalisation) makes it doesn't need to try and match also the scalp
between the two images.
In SPM2 the defaults is not to use masking (gives slightly better
results when basing normalisation on EPI images).
So, if you are using structurals and haven't yet changed the defaults,
you could try that.
Also, do these subjects actually have bigger heads? If so, you might
want to consider not using the Bayesian priors for the affine step since
these are based on a "normal" population and probably inappropriate for
this.
Good luck Jesper
|