Gray band in edge region of carpet plot

[lobennett]

What happened?

We ran fMRIPrep v24.1.0rc0 on our dataset and some of our QA outputs have a gray band in the edge voxelwise time series of the carpet plot (see example below). We think this was caused by the brain being positioned near the edge of the field of view, so when the brain mask dilated, the region went outside where there was any data. We don't think this will significantly compromise our analyzes, but thought it was worth opening this issue in the spirit of documentation.

What command did you use?

N/A

What version of fMRIPrep are you running?

24.1.0rc0

How are you running fMRIPrep?

Singularity

Is your data BIDS valid?

Yes

Are you reusing any previously computed results?

FreeSurfer, Work directory

Please copy and paste any relevant log output.

No response

Additional information / screenshots

[tsalo]

Do the coregistration and normalization figures look good to you?

I've noticed issues with the version of nitransforms bundled in fMRIPrep 24.1.0rc0 in fMRIPost-AROMA (see nipreps/fmripost-aroma#58 (comment)), so that might be it.

[effigies]

My instinct is that this is actually just a FoV issue. The "edge" voxels are the brainmask dilated by a few voxels, minus the original mask. When that hits the boundary of the BOLD image, there's just nothing to report.

It would be good to reproduce to figure out exactly why this is appearing and if there are any knock-on effects, but I don't think this is related to MNI resampling or a major problem.