pydeface missing when converting NIFTI to 256 conformed MGZ file

[milanmlft]

When running the Converts NIFTI to 256 conformed MGZ file command, as part of the IDEAS workshop (see #44), we run into the following error:

/fastsurfer/nii_to_mgz.sh: line 27: pydeface: command not found
Deface failed

What is confusing is that in the workshop, the defacing is performed after the first run of FastSurfer (after which FastSurfer is re-run to see the impact on the results). So there is no need to have pydeface when we just want to convert the NIFTI files to MGZ, I think?

[milanmlft]

#51 (comment) suggests changing the image for the pydeface command. However, running the Defaces a T1 Nifti scan command on a NIFTI-converted scan works fine when using the healthbioscienceideas/pydeface:minify image.

So the problem seems to be a missing pydeface installation in the fastsurfer container (used for the MGZ conversion).

[p-j-smith]

#51 (comment) suggests changing the image for the pydeface command. However, running the Defaces a T1 Nifti scan command on a NIFTI-converted scan works fine when using the healthbioscienceideas/pydeface:minify image.

So the problem seems to be a missing pydeface installation in the fastsurfer container (used for the MGZ conversio

ah good point!

[milanmlft]

Not sure if there is anything we can do about this? Seems more like a problem on the XNAT side, not with our AWS setup?

I think there's a problem with the healthbioscienceideas/fastsurfer Docker container. I'm assuming https://github.com/HealthBioscienceIDEAS/MedICSS-Project-Repro-Pipelines/blob/6aa7368c6467f29eb5a72ff5a949e0f29e44a4fa/fastsurfer/nii_to_mgz.sh is the nii_to_mgz.sh script that is being run by the container. But I don't see any reference to pydeface there...

Maybe @HChughtai can help with this?

[milanmlft]

I had a look into the healthbioscienceideas/fastsurfer:gpu Docker container, and there the nii_to_mgz.sh actually looks like this:

#!/usr/bin/env bash
# Wrapper script around pydeface, so that it can yank the first
# probably only file from the inoput directory
# and preserve all other arguments

die(){
    echo >&2 "$@"
    exit 1
}

IN_DIR=$1
shift
OUT_DIR=$1
shift

IN_FILE=$(ls ${IN_DIR}/*.nii* | head -1)
[[ -f ${IN_FILE} ]] || die "Did not find nii or nii.gz file in ${IN_DIR}"
IN_BASE=`basename ${IN_FILE} .gz`
IN_BASE=`basename ${IN_BASE} .nii`
OUT_FILE=${OUT_DIR}/${IN_BASE}.mgz

export PATH=/opt/miniconda-latest/envs/neuro/bin:${PATH}
echo IN_FILE=${IN_FILE}
echo OUT_FILE=${OUT_FILE}
echo
echo mri_convert $@ ${IN_FILE} ${OUT_FILE}
pydeface $@ ${IN_FILE} ${OUT_FILE} || die "Deface failed"

So it does indeed call pydeface in the last line and is different from the script mentioned above. Not sure whether this is an error or not?

[HChughtai]

That's an error and I think that Docker image is out of date, and so has the wrong version. You can see in https://github.com/HealthBioscienceIDEAS/MedICSS-Project-Repro-Pipelines/blob/gh-pages/fastsurfer/nii_to_mgz.sh that the line doesn't exist.

And you're right that it should be a separate step to the conversion. I think what happened was that Dave created nii_to_mgz.sh by copying run_pydeface.sh and that was left behind. I'd suggest rebuilding the fastsurfer image

[milanmlft]

Thanks @HChughtai! Can confirm that manually editing the nii_to_mgz.sh script in the Docker container on AWS and just replacing pydeface with mri_convert in that last line gets rid of the error. But now I'm getting this:

--------------------------------------------------------------------------
ERROR: FreeSurfer license file /fs60/license.txt not found.
  If you are outside the NMR-Martinos Center,
  go to http://surfer.nmr.mgh.harvard.edu/registration.html to 
  get a valid license file (it's free).
  If you are inside the NMR-Martinos Center,
  make sure to source the standard environment.
  A path to an alternative license file can also be
  specified with the FS_LICENSE environmental variable.
--------------------------------------------------------------------------
Deface failed

Do we just need to create a new license and add it to the Docker container?

[HChughtai]

Do we just need to create a new license and add it to the Docker container?

Yeah, that would do it!