Hello fellow SPM:ers,
I am a bit confused with regards to using the time derivative in SPM.
The issue is that I have an event related paradigm where I do post hoc sorting of stimuli to 2 conditions. At the last SPM course I learned that slice time correction should be avoided, so instead of using time slice correction I wanted to include a second basis function to account for timing differences in data acquisition between slices.
To check for regions significantly more activated in cond 1 vs cond 2 I did an F-contrast [1 0 -1 0; 0 1 0 -1] to look for voxels associated with either the HRF or td. I masked this contrast inclusive with a t-contrast [1 0 -1 0] to only look at voxels with activation corresponding to the HRF. The problem here is that I'm telling SPM to look for voxels with different timing depending on condition, while I'm interested in allowing for timing differences between regions regardless of condition.
Another variant I've thought of is to merely do a t-contrast [1 0 -1 0], controlling for any variance associated with the td, but I read that this may give biased results.
The next issue is how to perform the 2nd level analysis. With the latter method it's straightforward, but from what I understand I can't do a random effects on F-contrasts?
Thank you for any suggestions!
Bert regards,
Jonas Persson
|