neurite 0.3

Neural networks toolbox focused on medical image analysis.

Neurite

A neural networks toolbox with a focus on medical image analysis in pytorch.

⚠️ Warning: neurite is under active development. We are in the process of finalizing the structure for PyTorch -- interfaces may change.

Install

To use the Neurite library, either clone this repository and install the requirements listed in setup.py or install directly with pip.

pip install neurite

Pre-commit hooks

This repo uses pre-commit to run pycodestyle before commits.

Install once after cloning:

pip install pre-commit
pre-commit install

You can also run the check manually:

pre-commit run pycodestyle --all-files

For users who want to use the stable TensorFlow version, use either pip install neurite, or pull/clone from the dev-tensorflow branch.

Main tools

• nn.functional: tensor ops for smoothing, resampling, interpolation, masking, and spatial math.
• nn.modules: reusable stateful layers, losses, and preprocessing blocks.
• nn.models: prebuilt architectures for arbitrary spatial dimensions (1d, 2d, 3d).
• py.plot: plotting tools for tensor slices, volumes, and flow fields.
• utils.utils: lightweight factory and helper functions and tools.

Papers:

If you use this code, please cite:

Anatomical Priors in Convolutional Networks for Unsupervised Biomedical Segmentation
Adrian V. Dalca, John Guttag, Mert R. Sabuncu
CVPR 2018.
[ PDF | arxiv | bibtex ]

If you are using any of the sparse/imputation functions, please cite:

Unsupervised Data Imputation via Variational Inference of Deep Subspaces
Adrian V. Dalca, John Guttag, Mert R. Sabuncu
Arxiv preprint 2019
[ arxiv | bibtex ]

Development:

We welcome contributions; please make sure your code respects pep8, except for E731,W291,W503,W504, by running:
pycodestyle --ignore E731,W291,W503,W504 --max-line-length 100 /path/to/neurite
Please open an issue [preferred] or contact Adrian Dalca at [e-mail] for question related to neurite.

Use/demos:

Parts of neurite were used in VoxelMorph and brainstorm, which we encourage you to check out!