If you want to look for an interaction between two groups, then use the contrast [1 -1 -1 1]

We don't have a way to fit nonlinear models, but you can apply a nonlinearity to the blood data and then include it as a linear covariate

doug

On 06/28/2014 06:20 PM, Jacobs H (NP) wrote:

Dear FreeSurfer team,

We have processed our T1 data of 4 groups of subjects: controls and patients of varying degrees illness. Now we want to perform :

• a F-test to compare the 4 groups (we have one covariate that we want to take into account)
• a F test to investigate a differential effect of a blood value across the 4 groups. However, this effect might be non-linear (e.g. inverted U-shape).

We have made our fsgd-files, but now we are struggling with the contrast files. Could you help us out on how to create these F-test contrasts?

Thanks! Heidi

---

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