Strix

A Medical Deep Learning Platform
Make deep-learning easier for medical problems
Explore the docs »

View Demo · Report Bug · Request Feature

About The Project

Motivation:

We are trying to create a comprehensive framework to easily build medical deep-learning applications.

• Friendly interactive interface
• Good plug and play capacity
• Various tasks support
• Easy debug & Good reproducibility

Design Concept:

We aim to disentangle both Data scientist & Archi Engineer, Networks & Pipelines.
You can easily put your own datasets and networks into Strix and run it!

• Data scientists can focus on data collection, analysis and preprocessing.
• Architecture engineers can focus on exploring network architectures.

Getting Started

Prerequisites

Strix is powered by Pytorch, MONAI and Ignite.
Strix relies heavily on following packages to make it work. Theoretically, these packages will be automatically installed via pip installation. If not, please manually install them.

• pytorch
• tb-nightly
• click
• tqdm
• numpy
• scipy
• scikit-image
• scikit-learn
• nibabel
• pytorch-ignite
• monai_ex
• utils_cw

Installation

For developers, we suggest you to get a local copy up and install.

git clone https://github.com/Project-Strix/Strix.git
pip install -e ./Strix

For users, you can just install via pip.

pip install git+https://github.com/Project-Strix/Strix.git

More details please refer to install.
Notice that Strix is only tested on Linux only, not on Windows. If you find any problem with the deployment on Windows, please submit an issue.

Usage

For more examples, please refer to the Documentation

Strix has 7 different commands:

• strix-train: Main train entry. Use this command for general DL training process.
• strix-train-from-cfg: Begin a training process from specified configure file, usually used for reproduction.
• strix-train-and-test: Begin a full training&testing process automatically.
• strix-test-from-cfg: Begin a testing processing from specified configure file.
• strix-nni-search: Use NNI for automatic hyper-parameter tuning.
• strix-check-data: Visualize preprocessed input dataset.
• strix-gradcam-from-cfg: Gradcam visualization.

Here is a usage example!

How to use my own dataset & network?

• If you want use your own dataset, first you need to create a simple python script of a configuration to generate your dataset. For more details, please refer to this readme
• If you want try your own network, you need to follow this steps to easily register your network to Strix.
• After preparation, just simply put own dataset/network file into custom folder, and run!

Roadmap

See the open issues for a list of proposed features (and known issues).

Contributing

Contributions are what make the open source community such an amazing place to be learn, inspire, and create. Any contributions you make are greatly appreciated.

License

Distributed under the GNU GPL v3.0 License. See LICENSE for more information.

Contact

Chenglong Wang - [email protected]

Project Link: https://github.com/Project-Strix/Strix

Acknowledgements