Hi Brittany,

I have a similar dataset that I am using in Conn - two groups of children scanned at two time points. I can't comment on the a recommended approach in Conn, but I think that the same considerations for task-based fMRI will also apply to functional connectivity analysis.

I am associating the functional data to the structural that was acquired at each time point. I wouldn't expect there to be large structural changes over 8 weeks (in my dataset) or 16 weeks (in your dataset), but if there are subtle changes in brain structure then it is surely best that these are accounted for by including the structurals collected at both time points. I imagine that registering a more recently acquired functional image to an older structural image can only be less accurate.

When using Conn 17f, I defined two sessions and the default gave me two sessions to allocate structural images. Perhaps it is just a by-product of the way you are selecting your scans? I did mine by assigning each image to each subject/session one by one.

Hope that helps!

Best wishes,
Joff