Dear Haoran Li,
> I found I would get different results if I did two-sample t test within a
> selected ROI in these two ways:
> 1. I did two-simple T test on 2 groups of volumetric(2/3D image) data
> within a spceified ROI by applying a explicit mask(such as prefrontal
> cortex, come from wfu_pickaltals) in "specify 2nd level" section, and then
> "Estimate" and "Results";
> 2. I did the same two-simple T test but without applying any explicit
> masks in "specify 2nd level" section, but did a ROI analysis in "Results"
> section.
> I found these two different ways would lead to different results.
> Anybody has any suggestions on this issue?
The voxel-level uncorrected p values should be unchanged. However,
the smoothness estimation used in the correction procedures will be
different. In the first case, the smoothness is based on only the
parts of the image within your mask. In the second case, the
smoothness is based on the whole brain (although you are only
considering, in the end, voxels within a mask). This will change the
thresholds for the corrected values, although I would assume your
results would be somewhat similar (though not identical).
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/
|