Dear Matthew,
>I was wondering if anyone could help me with a problem we've been
>having. I do research on an UCLA OCD/Depression research study.
>
>We have PET data collected on both the Siemens neuro ECAT 831 and the
>Siemens ECAT EXACT HR+961. We have run SPM anaysis using both SPM96
>and SPM99. SPM analysis on the data group that was run on the 961
>seems to be fine.
>However, when running the 831 group, we encounter marked activation
>along the entire ventrical and white matter tract, regardless of which
>831 group we use. The pattern of activation (increase) shown looks the
>same each time; and this pattern does not appear in the decrease
>(deactivation) analysis. I thought SPM was supposed to exclude these
>regions?
>We have also increased our height threshold to about 1.4 to rule out
>this ventricular and white matter areas, but few other regions show
>up.
>I was wondering if anyone knew what this problem might be? Does anyone
>have suggestions for correcting this pattern of activation that always
>appears?
This pattern is usually associated with an inappropriate correction for
global confounds. If you are using AnCova normalization I would switch
to scaling. Have a look at the globals and see if there are any
outliers or systematic group/condition differences. The other
explanation may be systematic anatomical differences that are not
removed by spatial normalization. One approach to this would be to use
a secondary template based on 831 data.
I hope this helps - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|