We are looking for anatomical changes in young women at two points of the menstrual
cycle. We have T1 structural images from each woman (n=14) at both time points (each
woman was scanned twice approximately 2-weeks apart). Although our design is similar
to a longitudinal within-subjects design, I haven't found a posting specific to our situation
where the time points are commutative. I'm wondering if that matters, and what the best
approach would be.
We are using optimized VBM in SPM2 for processing, and using a paired-t statisical design
to compare gray matter volumes between the two points of the cycle.
Initially we performed the analysis using two customized templates (one for each group),
and found several regions with highly significant changes in gray matter volume (as well
as gray matter density, but are unsure of how to interpret those findings?). When we
complete the analysis with a single template comprised of average images made from
the two scans of each woman, our results are not nearly as significant. Which approach
(if either) is valid?
Additionally, if we repeat the analysis in SPM5, would it be better to use the standard
template or create our own, considering our sample is only women?
Also, should we co-register each woman's structual images together?
Thanks in advance for any advice and guidance,
Ashley
|