SDC introduces distortions in 25.X.X but not 25.0.0

[amirhusseinab]

What happened?

Hi all,

I'm having an issue with SDC using the PEPOLAR method in fMRIprep 25.X.X (PE-dir = LR/RL), which results in more distortion rather than correcting it. Interestingly, SDC performs completely normally in 25.0.0 (see figures below). This reminds me of an older issue with 22.X series (#3013), which was resolved with version 23.2.0 by implementing Jacobian weighting during unwarping. I also noticed a similar issue was reported a while ago for 25.2.3 (#3596).

Same subject, fMRIprep ran twice:
With 25.0.0:

With 25.2.5:

Any help is appreciated.

Best,
Amir

What command did you use?

fmriprep-v25.2.5.simg /input/ /derivatives/fmriprep_25.2.5 \
--fs-license-file /toolboxes/license.txt \
participant --participant-label ${subjid} -t rest \
--output-spaces MNI152NLin6Asym:res-2 MNI152NLin2009cAsym \
-w /work/ --fs-subjects-dir /derivatives/fastsurfer_2.4.2 \
--stop-on-first-crash --nthreads 8

What version of fMRIPrep are you running?

25.2.5

How are you running fMRIPrep?

Singularity

Is your data BIDS valid?

Yes

Are you reusing any previously computed results?

Anatomical derivatives, FreeSurfer

Please copy and paste any relevant log output.

Additional information / screenshots

No response

[ZeneWang]

I’ve been experiencing a similar issue with version 25.2.5, particularly in the temporal lobe, where that region seems to go missing after preprocessing. I also noticed the same problem with version 23.2.3.