The only small point I might add is that adding "-radial_correlate_opts -polort 18" should have affected the @radial_correlate script as you hoped. There would be *2* -polort options in the command in the proc script, but yours would be later and therefore override the first (a subtle point).
But still, use the options Paul suggests.
A minor effect of this goes the opposite way though (with @radial_correlate). If you specify "-regress_polort 2", that should change the one used for @radial_correlate to be just 2, which is probably lower than preferable. It does not have a bandpassing option.
So even with Paul's suggestion, you might also add something like "-radial_correlate_opts -polort 9", just to have more reasonable detrending for those QC steps.
- rick