Hello,
A couple of quick questions:
1) I unable to define a T contrast when using the single subject
covariates only PET model. We are looking at GM volume correlations with
Age and other variables with TIV as a nuisance variable. In the case of
GM and age, I have tried to define a new T contrast for decreasing GM
with age (-1 0 0 - 0's for the block and TIV). Is this correct? The F
contrasts work just fine but are two tailed and we would like to see
only the areas where there is either an increase or a decrease of GM.
Are we doing something wrong here?
2) When we run the analysis and inspect the p values, the cluster level
stats are not provided although under the voxel level heading, all
numbers appear. I understand that this was not the case in spm99. Is the
idea that we have to inspect each cluster seperately by moving the
cursor and then clicking for the cluster p values or is theira bug in
our local version of spm2?
3) When using the corrclusth script (written by Tom Nichols and taken
from the spm mail list) for clusterwise analysis, our extent threshold
is always 1000 voxels when we enter a cluster defining threshold p value
of .001. Why is this occurring and is this ok?
4) Finally, when thresholding, is it acceptable to use a Z score
regardless of the corrected p values reported by spm?
Thanks,
Luke
-------------------------------
Luke Spencer-Gardner
Research Assistant
Hoglund Brain Imaging Center /
Department of Neurology
Mail Stop 1052
The University of Kansas Medical Center
3901 Rainbow Blvd
Kansas City, KS 66160
Phone: (913) 588-0073
Fax: (913) 588-9071
Email: [log in to unmask]
|