Hi Doug,
I apologize if my last email was a bit confusing/lacking in information. Dr. Roffman gives a more thorough explanation of what we're trying to achieve (see below). Hopefully this gives you a better idea of what we're asking -- thanks!
We are looking at the relationship between age and cortical thickness in two independent groups (each n=22).
We are not interested in mapping the difference in slopes for the age-thickness relationship between group 1 and group 2. Rather, we are interested in mapping the significance of the difference in the age-thickness correlation strength between group 1 and group 2. Two distributions can have a similar slope, but the R2 for one group can be very different than the R2 for the other group, based on how tightly the datapoints conform to the best fit line for each group. That appears to be the case for our data.
If we are using an ROI approach, we see that in group 1, within the left inferior frontal gyrus, when we look at Pearson's r for the relationship between cortical thickness and age, we get r=.717 (p=.0002). In group 2, within the same region, Pearson's r=.169 (p=.452). We can test for a difference in correlation strength between two groups by performing a Fisher r-to-z transformation. With 22 subjects in each group, this yields a z score of 2.25, and p=.0244, indicating that the correlation is significantly stronger in group 1 than in group 2.
The question is whether we can apply this approach in a vertex-wise manner and see where across the brain there are differences in age-thickness correlation strength between the two groups.
Any advice would be greatly appreciated!
Best, Alex
Hi Alex, I'm not sure I understand what you mean. What is the strength of the thickness/age relationship if not the slope between thickness and age? doug
On 04/05/2013 03:56 PM, Alexandra Tanner wrote:
Dear Doug and Freesurfers,
I'm running a group cortical thickness analysis covaried by age, and we're curious if there is a way to correlate thickness and age within groups and then look at whether or not there is a difference in the strength of the thickness/age relationship between groups. Originally the contrast file we used was the group-x-age.mtx (0 0 1 -1) looking at whether or not there was a difference between the group age slopes, but we then realized we're not interested in whether the slopes of the relationships are different, but rather if the strength of the cortical thickness/age relationship differs between groups.
I ran mri_glmfit with an fsgd file that was set up as follows: GroupDescriptorFile 1 Title OSGM Class Group1 Class Group2 Variables Age Input subject1 Group1 30 Input subject2 Group2 40
I've taken a look at the wiki (http://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf2G1V) and can't seem to find a contrast file that does what we're looking for. Any advice on what sort of contrast file we should use, or whether we need to run a different analysis than glmfit, would be greatly appreciated! Thanks!
Best, Alex
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/