Using an sbref with Susceptibility Distortion Correction, Should 'IntendedFor' field include sbref?

Should the ‘IntendedFor’ field in field mapping images include the single-band reference? The functional data are going through fmriprep with SDC turned on, but I’m not quite following the BOLD workflow. It seems like the BOLD reference image, when it’s calculated without an sbref, undergoes distortion correction prior to alignment with the anatomical. But when the reference image is the sbref, is distortion correction still applied? Or, does fmriprep need the sbref mentioned in the json sidecar of the fmap image?

"IntendedFor": ["ses-pre/func/sub-01_ses-pre_task-motor_run-1_bold.nii.gz",
                "ses-pre/func/sub-01_ses-pre_task-motor_run-1_sbref.nii.gz"]

vs. just

"IntendedFor": ["ses-pre/func/sub-01_ses-pre_task-motor_run-1_bold.nii.gz"]

Thanks!

Hi @psadil,

While it may be more BIDS-correct to specify the sbref in IntendedFor, fMRIPrep will apply SDC to whatever reference image is used, whether it’s an sbref or estimated from the BOLD series.

Got it. Thanks for the clarification.

Jus to clarify, if both a bold and sbref image are provided in the bids func/ directory, will fmriprep automatically use the sbref as the reference, or does that require an explicit call? Currently my IntendedFor field only contains the bold file, not the sbref, so I’m not entirely sure what would need to be done to use the sbref at the reference.

Thanks

@dlevitas

The docs at least say that the sbref will be applied, unless fmriprep is told to ignore it (e.g., using --ignore sbref in the call to fmriprep). So, I think the setup is “use sbref unless explicitly told to ignore.”

But, I suppose the presence of this thread points out that I’m not the best at following the workflows. I can at least report that fmriprep is using my sbref images, despite that they’re not listed in the IntendedFor field of the fmap images (and that sounds like your situation).

Correct. The reason to acquire them is typically to improve the quality of registration, so that’s what we assume users want unless they specify otherwise.

1 Like

Got it, thanks! This is a reminder that I need to take a deep dive into fmriprep’s workflows at some point.