Dear all,
We really appreciate your advice on the following:
We collected resting state fMRI data on three occasions (within a six month period) in two groups (clinical and control). Now we would like to investigate longitudinal changes in functional connectivity between the two groups. In the cross-sectional analyses on the data of the first measurement we used a seed-based approach using FEAT and found significant differences between the two groups in functional connectivity.
For the longitudinal data we ideally would like to use the same seed region (amygdala) and the same method (FEAT). However we are not sure whether the higher level design we setup, is correct. Particularly, we would like to get advice about the within-subjects effect and on how to model these within FSL (also on the contrast weights in the higher level model setup). For the current design, we used the advice presented on the FSL website under the heading ‘Tripled two-group difference’.
Now, we ran a lower level feat analyses for each participant and each measurement separately. For the higher level analyses we came up with the design (please see attachment). In there, the first part concerns the EV’s (with an additional EV for each participant to model the within-subject effect) and the second part are the contrasts and F-tests. Furthermore, Gr stands for Group and s stands for session. There are contrasts for the different groups and group x session contrasts.
All advice is very welcome and if anything is unclear please let us know!
Best,
Bianca van den Bulk
|