There are several ways to proceed:
1. The most rigorous is two use a random effects model. The random effects
package may already be installed in your version of SPM, and contain
documentation regarding performing a group comparison. Essentially you would
create a contrast image for each condition (is this an 15O-water study?) You
would then study the contrast between groups, using a single contrast image for
each subject. However, you likely do not have enough subjects for this to have
any likelihood of success. 12 is an accepted minimum per group (see Woods work
on this subject:
http://www.ncbi.nlm.nih.gov/htbin-post/Entrez/query?db=m&form=6&dopt=r&uid=0009345532
and Friston, Holmes and Worley
http://www.ncbi.nlm.nih.gov:80/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=10385576&dopt=Abstract)
2. Study the activations within each group, and compare qualitatively, or mask
the activations of one group with that of the other.
3. Study activations individually, and compare activation in a region of
interest. Again, it may be hard to find a statistically significant group
difference.
The upshot is that you are likely only to find a trend, not a rigorously
defensible difference between groups, unless of course, you're very lucky.
Best of luck.
Jennifer Thompson wrote:
> We are analyzing a PET study using SPM96 in which we have two groups: (1) a
> control group of ten subjects and (2) a patient group of ten subjects. For
> each of the controls we have one scan. For each of the patients we have 5
> scans. We want to compare these groups to see if there are differences in
> activation between them and are not sure how to proceed. Thank you.
>
> Lara
|