Hi Joost, you want to do a repeated-measures ANOVA. To do this in the FSGD, set up a class for each subject and use (in your case) 5 continuous variables which will correspond to 1-2, 1-3, 1-4, 1-5, and 1-6. For the first site scan of each subject, assign all variables equal to 1. For the second scan, set the first variable = -1 and 0 for the rest. For the 3rd scan, set the 2nd variable to -1 and 0 for the rest, etc. When you run mri_glmfit, specify that it is DOSS. This means that there will be 11 regressors (6 classes and 5 variables). Your contrast will then be C = [zeros(5,6) eye(5,5)] where zeros(5,6) is a 5x6 array of 0s, and eye(5,5) is the 5x5 identity matrix.

doug

j janssen wrote:

FreeSurfer version: 4.5

Background: we have 6 subjects, each subject has 1 scan per site, there are 6 sites (so each subject has 6 scans).

Aim: to assess whether thickness is different per site: a F-test.

what we did so far:

we used mri_glmfit with the following .fsdg file (here we only show the Input of the first 2 subjects)

GroupDescriptorFile 1 Title scanner MeasurementName thickness Class Site1 Class Site2 Class Site3 Class Site4 Class Site5 Class Site6 Input Subject1 Site1 Input Subject1 Site2 Input Subject1 Site3 Input Subject1 Site4 Input Subject1 Site5 Input Subject1 Site6 Input Subject2 Site1 Input Subject2 Site2 Input Subject2 Site3 Input Subject2 Site4 Input Subject2 Site5 Input Subject2 Site6 . . Input Subject6 Site6

we used the following .mtx file for mri_glmfit:

1 -1 0 0 0 0 0 1 -1 0 0 0 0 0 1 -1 0 0 0 0 0 1 -1 0 0 0 0 0 1 -1

we doubt that this is correct since our contrasts do handle the subjects at each site as an independent group, i.e. it does not take into account that each subject had 6 different scans (one at each site). any advice on how we should set up our contrast matrix? i repeat: our aim is to do a F-test to assess whether thickness is affected by site.

thanks, -joost _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer