Accurate Brain Segmentation in Malformations of Cortical Development

PyTorch Implementation using V-net variant of Fully Convolutional Neural Networks

Authors: Ravnoor Gill, Benoit Caldairou, Neda Bernasconi and Andrea Bernasconi

---

Implementation based on:
Milletari, F., Navab, N., & Ahmadi, S. A. (2016, October). V-net: Fully convolutional neural networks for volumetric medical image segmentation. In 2016 Fourth International Conference on 3D vision (3DV) (pp. 565-571). IEEE.

Please cite:

@misc{Gill2021,
  author = {Gill RS, et al},
  title = {Accurate and Reliable Brain Extraction in Cortical Malformations},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/NOEL-MNI/deepMask}},
  doi = {10.5281/zenodo.4521716}
}

Pre-requisites

1. Python >= 3.7
2. PyTorch (LTS) <= 1.8.2
3. ANTsPy
4. ANTsPyNet

Installation

conda create -n deepMask python=3.8
conda activate deepMask
pip install -r app/requirements.txt

Usage

TODO: Training routine

Inference using Docker

docker run -it -v /tmp:/tmp docker.pkg.github.com/noel-mni/deepmask/app:latest /app/inference.py \
                                            $PATIENT_ID \
                                            /tmp/T1.nii.gz /tmp/FLAIR.nii.gz \
                                            /tmp

License

Copyright 2021 Neuroimaging of Epilepsy Laboratory, McGill University