Running SPM with fmriprep intermediates

We’re running FMRIPREP preprocessing on a dataset and attempting to import some of the derivatives and regressors into SPM for first-level analysis. The documentation at wasn’t detailed enough to figure out if what the correct intermediate would be, so I started off by substituting the sub-***_task-***_space-MNI152NLin2009cAsym_desc-preproc_bold.nii for the swua (smoothed normalized realigned and STC) spm intermediates.

When we run a first level contrast on this data we get a fairly nonsensical result, resembling white noise. Is this the proper intermediate to use, and if so what would have to be done to the data to register it properly in SPM?

These are not smoothed, but are realigned and normalized and (if you had slice timing information and didn’t disable it) slice-timing corrected.