Hi There,
I am struggling quite a bit with how to best conduct a longitudinal analysis with my FSL fmri data.
I have data (fMRI data from 2 runs of a block design task) from individuals at 2 timepoints (with a lot of missing datapoints at each timepoint) with variable intervals between the time points. Due to these differing intervals and missing data, I'm interested in obtaining parameter estimates from all subjects for a simple ROI, then using mixed effects models in R to conduct my longitudinal analyses. However, I'm not sure as to how to best get these parameter estimates to then put into my mixed effects models in R.
I believe there are 3 ways to do this but would love input on the most correct way to go about this:
1) Perform 2 group level analyses for all subjects at T1 and all subjects at T2 separately, then used fslmeants to extract PEs from each model and use these PEs in my mixed effects model.
2) Combine all within-subject data into a second level model and use contrasts to model T1 and T2 data, then extract PEs from each of these contrasts for T1 and T2 (however, this method only allows you to use data from subjects who have 2 timepoints).
3) Use featquery to turn COPE/PE values into % signal change. However, when I do this on the second level model (combining run 1 and run 2) the values I get are equivalent to the PEs - so I believe this is not actually calculating % signal change from the second level model.
Thanks in advance for your response!
Natalie
|