gift-surg / SPOT-A-NeonatalRabbit

Segmentation Propagation on Target (...as a Neonatal Rabbit)

SPOT-A-NeonatalRabbit

Segmentations Propagation On Target (...as a Neonatal Rabbit)

This is an open-source repository written in Python 2 and based on LABelsToolkit, NiftyReg, NiftySeg and NifTK to propagate an MRI multi-atlas on a new subject scanned with the same modalities of the multi-atlas (target). Default parameters are tuned to work in particular for the Newborn Rabbit Multi-Atlas developed within the GIFT-surg project in collaboration between University College London (UK) and Katholieke Universiteit Leuven (Belgium).

Tuning the parameters (attributes of the class spotter.Spot_DS) according to the dataset, this software can be a starting point framework to automate general segmentation propagation and label fusion pipelines.

Material related to Neonatal Rabbit Brain Mutli-Atlas:

• MRI Neonatal Rabbit Brain Mutli-Atlas
• Repository for the Rabbit Brain Mutli-Atlas benchmarking
• Documentation with the protocol employed in creating the Rabbit Brain Mutli-Atlas

Getting started

• Requirements

  • Python 2 (Python 2 only for the current release)
  • Libraries in the textfile requirements.txt
  • NiftyReg C++ library
  • NiftySeg C++ library
  • NifTK (only if the differential bias field correction required for the non-rigid registration step)
  • DummyForMRI to create the dummy dataset for running the tests.

• Installation

  • We recommend to install the software in development mode, inside a python virtual-environment with the following commands.

    cd <folder where to clone the code>
    git clone https://github.com/SebastianoF/SPOT-A-NeonatalRabbit.git
    cd SPOT-A-NeonatalRabbit
    source <virual-env with the required libraries>/bin/activate
    pip install -e .
    

• Examples

  • See the examples folder and the wiki page

Code testing

• Unit testing with pytest. Type pytest in a terminal at the cloned repository with the installed libraries.
• Tests are based on a lightweight multi-atlas with analogous structure than the Neonatal Brain Rabbit generated with LABelsToolkit. The first test may take some minutes to create the phantom multi-atlas.

Support and contribution

Please see the contribution guideline for bugs report, feature requests and code re-factoring and re-styling.

Copyright and licence

• Copyright (c) 2018, University College London.
• SPOT-A-NeonatalRabbit is available as free open-source software under 3-Clause BSD.
• To cite the code in your research please cite the related publication.

Authors and Acknowledgments

• The MRI Neonatal Rabbit Multi-Atlas and related code is developed within the GIFT-surg research project in collaboration with KU Leuven (Belgium) and UCL (UK).
• This work was supported by Wellcome / Engineering and Physical Sciences Research Council (EPSRC) [WT101957; NS/A000027/1; 203145Z/16/Z].
• See the related publication for the full list of authors and acknowledgments.