fMRIPrep 25.2.3 SDC distorts image in multi-echo data

Software Support
1

Summary of what happened:

Hi everyone! I’m currently trying to use fMRIPrep to preprocess a dataset from openneuro (ds006072). fMRIPrep is running smoothly in general, but there seems to be a problem with SDC, which leads to a distorted and elongated corrected image. This is one example, but the data for all my sessions is similarly distorted:

Screenshot 2026-01-19 at 10.23.52
Screenshot 2026-01-19 at 10.23.521920×932 186 KB

The JSON sidecards for the fmap data were missing the B0FieldIdentifier or IntendedFor fields. I checked other threads here for multi-echo data, so I only linked to the first echo of their respectives files. I also checked that the dataset_description.json has set “DatasetType”: “raw”. Here is how I added the IntendedFor fields once for AP:

    "IntendedFor": [
    "ses-1/func/sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-1_part-mag_bold.nii",
    "ses-1/func/sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-1_part-mag_bold.nii"
    ]
and once for PA:

"IntendedFor": [
"ses-1/func/sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-1_part-mag_bold.nii",
"ses-1/func/sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-1_part-mag_bold.nii"
]

and here is the file tree:

sub-P3

    ses-1
        anat
              sub-P3_ses-1_acq-t1mpr0p9mm_rec-NDNORM_run-1_T1w.json
              sub-P3_ses-1_acq-t1mpr0p9mm_rec-NDNORM_run-1_T1w.nii.gz
              sub-P3_ses-1_acq-t1mpr0p9mm_rec-NDNORM_run-2_T1w.json
              sub-P3_ses-1_acq-t1mpr0p9mm_rec-NDNORM_run-2_T1w.nii.gz
              sub-P3_ses-1_acq-t2spc0p9mmiso_rec-NDNORM_run-1_T2w.json
              sub-P3_ses-1_acq-t2spc0p9mmiso_rec-NDNORM_run-1_T2w.nii.gz
              sub-P3_ses-1_acq-t2spc0p9mmiso_rec-NDNORM_run-2_T2w.json
              sub-P3_ses-1_acq-t2spc0p9mmiso_rec-NDNORM_run-2_T2w.nii.gz
        dwi
        fmap
              sub-P3_ses-1_acq-SpinEchoFieldMapAP_dir-AP_run-1_epi.json
              sub-P3_ses-1_acq-SpinEchoFieldMapAP_dir-AP_run-1_epi.nii
              sub-P3_ses-1_acq-SpinEchoFieldMapAP_dir-AP_run-2_epi.json
              sub-P3_ses-1_acq-SpinEchoFieldMapAP_dir-AP_run-2_epi.nii
              sub-P3_ses-1_acq-SpinEchoFieldMapPA_dir-PA_run-1_epi.json
              sub-P3_ses-1_acq-SpinEchoFieldMapPA_dir-PA_run-1_epi.nii
              sub-P3_ses-1_acq-SpinEchoFieldMapPA_dir-PA_run-2_epi.json
              sub-P3_ses-1_acq-SpinEchoFieldMapPA_dir-PA_run-2_epi.nii
        func
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-1_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-1_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-2_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-2_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-3_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-3_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-4_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-4_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-5_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-1_echo-5_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-1_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-1_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-2_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-2_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-3_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-3_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-4_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-4_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-5_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST1_dir-AP_run-2_echo-5_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-1_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-1_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-2_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-2_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-3_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-3_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-4_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-4_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-5_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-1_echo-5_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-1_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-1_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-2_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-2_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-3_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-3_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-4_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-4_part-mag_bold.nii
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-5_part-mag_bold.json
              sub-P3_ses-1_task-BOLDREST2_dir-AP_run-2_echo-5_part-mag_bold.nii

Command used (and if a helper script was used, a link to the helper script or the command generated):

BIDS_DIR="/mnt/d/Tim_MA/rawdata"
OUT_DIR="/mnt/d/Tim_MA/derivatives"
FILTER_FILE="/mnt/d/Tim_MA/bids_filters.json"
SUBJECT="$1"  # pass subject as argument

# Create a unique work directory for each subject
WORK_DIR="/mnt/d/Tim_MA/work/${SUBJECT}"
mkdir -p "${WORK_DIR}"

DOCKER_IMAGE="nipreps/fmriprep:latest"
NCPUS=8
MEM_GB=32
MEM_MB=$((MEM_GB*1000-500))

export TEMPLATEFLOW_HOME=$HOME/.cache/templateflow

echo "Processing ${SUBJECT}..."

docker run -ti --rm \
    -v ${BIDS_DIR}:/data:ro \
    -v ${OUT_DIR}:/out \
    -v ${WORK_DIR}:/scratch \
    -v ${FILTER_FILE}:/bids_filters.json:ro \
    ${DOCKER_IMAGE} /data /out participant \
    --participant-label ${SUBJECT} \
    --fs-license-file /out/license.txt \
    --bids-filter-file /bids_filters.json \
    --work-dir /scratch \
    --output-spaces MNI152NLin2009cAsym \
    --nprocs ${NCPUS} \
    --mem ${MEM_MB} \
    --skip_bids_validation \
    --stop-on-first-crash

Version:

25.2.3

Environment (Docker, Singularity / Apptainer, custom installation):

Docker

Data formatted according to a validatable standard? Please provide the output of the validator:

On openneuro.org it says that the data is BIDS valid, but missing some optional flags, like B0FieldIdentifier.

Relevant log outputs (up to 20 lines):

fMRIPrep is running without any error messages.

Screenshots / relevant information:

This is what the distorted version looks like, before SDC:

Screenshot 2026-01-19 at 11.32.38
Screenshot 2026-01-19 at 11.32.381920×932 188 KB

And these are the fmapCoreg_bold images:
Screenshot 2026-01-19 at 11.36.39
Screenshot 2026-01-19 at 11.36.391920×932 144 KB

Screenshot 2026-01-19 at 11.37.01
Screenshot 2026-01-19 at 11.37.011920×932 147 KB

I also only want the resting-state scans preprocessed to save some time, so I used this bids_filter:

{
  "bold": {
    "datatype": "func",
    "suffix": "bold",
    "task": [
      "BOLDREST1",
      "BOLDREST2",
      "BOLDREST3",
      "BOLDREST4",
      "BOLDREST5",
      "BOLDREST6"
    ]
  }
}

Thanks for the help!

2

Hi @tblascht and welcome to neurostars!

I wonder if the phase encoding direction in all the images are correctly labeled. It looks like your BOLD is being elongated where it should be compressed and vice-versa. Can you run a test subject where you switch the sign of your BOLD PE-dir?

Besides that, the only things I can think of right now are to try rolling back fMRIPrep versions or trying SynBOLD-DisCo for SDC instead. Or, just not using SDC at all; those images before SDC didn’t have that bad of distortions to begin with.

Best,
Steven

3

Hi Steven,
thank you so much for your quick reply! I’ll try it out as soon as possible and see whether that changes anything. If it doesn’t work, I will probably fall back to just not using SDC at all.

Best
Tim