 |
 |
Dear SPM users, I have some longitudinal fMRI scans taken at 2 time points (t0-t1) during a working memory task (three conditions) in two groups (patient-control). To start I want to look for whole-brain changes within the groups across the baseline and follow-up. Then I will locate and compare this change between the groups. I thought that I could answer these questions by specifying and estimating a first-level fMRI model that includes two sessions per participant for the baseline (t0) and the follow-up (t1) scanning, with the same conditions in each as the participants performed the same paradigm in the scanner and head motion parameters as regressors. Then I will create the contrasts comparing the experimental conditions in the two time points, per subject. At the second level, I will create a flexible factorial model with the within factor of experimental condition and the between factor of diagnosis (group). Here is where i get stuck: I want to use accuracy in the working memory task and IQ at t0 and t1 as covariates of no interest at the second level model. However, since each image used in this model comes from the contrast of the two time points I'm not sure which would be the correct way around this. Would it be valid to use the difference between working memory accuracy at t0 and t1, and the difference between IQ at t0 and t1 as my covariates? Many thanks Vasilis