Since you're considering removing the variability due to age, I have two questions for you:
1. Is there significant variability in age across all subjects?
2. Do the two groups have roughly the same or different average age?
> I am interested in the difference in regression with age between my two groups.
Does it mean that you don't care about the group difference for each specific scan? In other words, you want to collapse all the scans for each subject regardless how many scans each subject has? If so, how did you analyze those scans at individual subject level, separately or combined?
Gang