Dear FSL experts,
I have a couple of questions hopefully simple enough not to take too much of your time.
can the input of dual regression be spatial maps not derived from RSNs after ICA? I want to look at cortico-striatal connectivity and have created binary structural masks for whole striatum, functional subdivisions and as spherical ROIs in striatal regions. The idea is to investigate whether dopamine agonists and antagonists affect cortico-striatal connectivity. If I understand correctly instead of the melodic_IC file I can create a 4D version of the above maps with fslmerge and use that as input. Would that be correct?
Also, when running ICA on task-related sequences (e.g. n-back), I assume tensor ICA would be more appropriate than concatenation? One thing that I am not clear about is the inclusion of the timeseries model and contrasts in the post-stats tab. If I don’t include them, then the IC maps would be model-free, while inclusion would show loadings of IC maps specific for the task conditions?
Many thanks,
Elias
Many thanks,
Elias
|