DevelopmentalImagingMCRI / MCRIBS

Surface version of the M-CRIB atlases.
12 stars 6 forks source link

M-CRIB-S

Surface version of M-CRIB.

This software uses ANTs/antsJointFusion for voxel labeling and Deformable within MIRTK to perform cortical surface extraction. A customised Freesurfer-like pipeline is provided to perform cortical parcellation on the surfaces with M-CRIB compatible labelling schemes.

The paper is available at https://www.nature.com/articles/s41598-020-61326-2

MIRTK was downloaded from https://github.com/BioMedIA/MIRTK. A customised version is included in this repo.

Installation

Check the source code out by

git checkout https://github.com/DevelopmentalImagingMCRI/MCRIBS.git
cd MCRIBS
./build.sh

The script build.sh will build ITK, VTK and MIRTK. Install the following dependencies prior to running:

This will install all of the dependencies for ubuntu:

apt install zlib1g-dev libboost-dev libglu1-mesa-dev libxt-dev python-dev python3-dev libtbb-dev libflann-dev libeigen3-dev python-contextlib2 python3-contextlib2 python3-imageio python3-numpy python3-scipy python3-pandas python3-numexpr

In centos, do this:

yum install zlib-devel boost-devel mesa-libGLU-devel libXt-devel python-devel python3-devel tbb-devel eigen3-devel python-contextlib2 python3-contextlib2 python3-imageio python3-numpy python3-scipy python3-pandas python3-numexpr

The script build.sh will checkout VTK 9.3, ITK 5.3 and build it.

Docker version

Pull the latest docker image from Docker Hub with the command:

docker pull developmentalimagingmcri/mcribs:latest

Run the software using the following command:

docker run \
  --rm \
  -it \
  --mount type=bind,source=<work dir>,target=/work \
  --mount type=bind,source=<freesurfer license>,target=/opt/freesurfer-license.txt \
  developmentalimagingmcri/mcribs:latest MCRIBReconAll ...

Where

Set up script

Depending on the shell you are using source the configuration files as follows:

The main script to run is MCRIBReconAll. To get usage run MCRIBReconAll --help

Set up your data

Place your raw structural images, for subject subjid into subdirectories as follows:

Running

There is a wrapper script called MCRIBReconAll that runs the pipeline of segmentation, surface extraction. The synopsis of the command is:

MCRIBReconAll [processing directive] [options] <subject id>

Processing directives and options

The following shorthand options may be used:

Other options: