Brain mask distorted causing registration issues

1

I’m having some issue with the brain mask on 1 of 4 runs - the other runs are all fine. The mask is badly distorted as seen here in red overlay on a boldref:

mask_boldref
mask_boldref1920×1091 123 KB

It appears this is having downstream effects on the preprocessed output. Here’s the T1w space preproc bold:

image
image1920×907 95.5 KB

However, I have multi-echo data and my single echo preproc files are fine:
image
image1920×914 86.5 KB

So while I can still run TEDANA on the single echo files, I can’t apply the MNI registrations from fmriprep to the tedana output as they were based on the distorted files. I followed a previous suggestion to create a mask from the union of CSF, WM, and GM probseg maps, but I still have the problem of not having a good reference image in MNI space to use for antsApplyTransforms.

Does anyone have any suggestions for a workaround that doesn’t involve manually running ants registration?

2

Hi @Joseph_Orr , which fmriprep version are you using? Which coil was used and was bias correction applied to the functional images before fmriprep?

3

Looks like my response failed so I’m trying again.

Hi Julien, I’m using fmriprep 24.0.0. The coil was a Siemens 32 ch. I didn’t apply any offline bias correction and I don’t think there is any online correction.

fmriprep command: fmriprep bids/ fmriprep participant --participant-label sub-1004 --fs-license-file license.txt --me-output-echos --me-t2s-fit-method curvefit --output-spaces anat fsaverage MNI152NLin2009cAsym --cifti-output 91k --nthreads 16 --omp-nthreads 16 --mem 116000