Fmriprep itk::error during brain extraction



I am preprocessing a dataset of 170 subjects with fmriprep and the processing stops during the brain extraction. Can you help me figuring out what the problem is?

I attached crash log from an example participant.

I am using the version 1.3.2 and I´ve copied the specs for the file and the command from the html:


  • Subject ID: 109
  • Structural images: 1 T1-weighted
  • Functional series: 1
    • Task: rest (1 run)
  • Resampling targets: fsaverage5, MNI152NLin2009cAsym
  • FreeSurfer reconstruction: Run by fMRIPrep


Anatomical Conformation

  • Input T1w images: 1
  • Output orientation: RAS
  • Output dimensions: 240x320x320
  • Output voxel size: 0.85mm x 0.85mm x 0.85mm
  • Discarded images: 0


Note on orientation: qform matrix overwritten

The qform has been copied from sform.


  • FMRIPrep version: 1.3.2
  • FMRIPrep command: /usr/local/miniconda/bin/fmriprep /data/pt_02147/BIDS/ /data/pt_02147/BIDS_OUT/ --template MNI152NLin2009cAsym --template-resampling-grid /data/pt_02147/MNI152_T1_2mm.nii.gz --no-submm-recon --fs-license-file /data/pt_02147/prep/license.txt -w /data/pt_02147/BIDS_WD/ participant --participant-label 109
  • Date preprocessed: 2019-04-09 08:57:14 +0000


  • Node Name: fmriprep_wf.single_subject_109_wf.anat_preproc_wf.brain_extraction_wf.init_aff

crash-20190409-090353-lauckner-init_aff-5f5a8b29-efe6-4d50-9c98-4234b16d9b79.txt (3.1 KB)


Can you check that the original file is actually not empty?



I can confirm, that this file is not empty. All structural files are stopped with the same error.

A little piece of extra information. The dimensionality of these structural images is unusually high and I ran into errors previously because of a high FOV, which I was able to circumvent using the --no-submm-reconoption. But now there is this problem. Could the FOV be the problem here? It would be an easy step to crop the dimensionality.


Could you share that image with me?



that might not be necessary, cropping the images did the job!

Thank you!