Hi,
I'd like to add my vote for the liberal mask policy--
I've run the same analysis using both the original FSL 4.0 and with the FSL
3.0 modifications posted by Steve.
I'm looking for activation in the medial temporal lobe and found that the
missing voxels around the brain stem when using the 4.0 approach impinged on
this area. The number of voxels missing became an issue at the group level.
As an example; using amygdala masks (defined as >50% probability from the
harvard sub cortical atlas) in a gfeat of 26 lower level analyses (from
separate volunteers). 15% of voxels in the left amygdala mask and 20% in
the right were missing from the analysis.
Using the FSL 3.0 approach there were no missing voxels from either side.
Hope this is useful
Mike
|