Hi Yaelan,
Have any trends already been projected, or any tedana regressors?
In general, provide -polort 0 and as many -ort and/or -dsort terms as you wish (where -ort files can have multiple columns). If no drifts have been projected already, then you might consider a -polort option that would be equivalent to that for 3dDeconvolve, i.e., 1+floor(run_length/150).
Detrending should be done in the same 3dTproject command, if it has not already been done. Actually, it should probably be done either way, since the motion terms might have trends in them.
- rick