On 06/11/2018 11:07 PM, Hua, Jessica wrote:

Dear Freesurfer Experts:

I analyzed my data following the guidelines on the FSL Tutorial/Group Analysis page . I am trying to plot the adjusted values (covariates: age, sex, and ICV) for the significant clusters for each of my participants.

I assume you mean FreeSurfer, not FSL?

I set up the contrast to test group differences. So testing whether there is a difference between groups regressing out the effects of sex, age, and ICV. Contrasts created were  [1 -1 0 0 0 0 0 0] and [-1 1 0 0 0 0 0 0].

I then extracted values from this file (cache.th20.abs.sig.y.ocn.dat; i.e., the average value of each subject in each cluster) into my dataset. I believe these are the unadjusted values, correct?

These are the average thickness data (assuming you are using thickness as input --y)

I calculated total volume and total surface area by multiplying these values by the total number of vertices in each cluster.

Not sure what you mean here or how it works.

To adjust for covariates, I ran a regression with the ExtractedClusterValues as the DV and Group, Sex, Age, and ICV as predictors. Looking at the output, there was no significant effect of group (p = .769; whereas in FreeSurfer the cluster-wise p < .001).

What were you using to do this test? You really need to have exactly the same design matrix and contrast matrix.

Also when I plotted the predicted values adjusting for covariates, the means between the groups were very similar.

Why am I getting a p = .769 when I adjust for covariates, but FreeSurfer has a cluster-wise p < .001? Or is there a way to calculate adjusted values in FreeSurfer?

Best,

Jessica

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer