Hi there,
would it be possible to add in an option to 3dTcorrelate to include some more 'nuisance' terms in the regression model, e.g., the nuisance caused by subjects who wouldn't stay completely still? I get lovely results for subs who are completely immobile, but there's a serious diminuendo in quality for people who scoot around just a little. (There's a good difference in results between with/without motion regressors with these subs using a standard model with 3dDeconvolve).
Perhaps more importantly, does it make sense to do this? I could too easily imagine that a big overdetermined mess could result... I had briefly thought about running 3dTcorrelate on the time-series residuals (after running with polorts and motion regressors), but this approach clearly has its pitfalls (e.g., one shouldn't really do it).
Any thoughts welcomed,
thanks,
Fred