README.md

Delira - Deep Learning in Radiology

Authors: Justus Schock, Oliver Rippel, Christoph Haarburger

Introduction

Delira was developed as a deep learning framework for medical images such as CT or MRI. Currently, it works on arbitrary data (based on NumPy).

Based on PyTorch, batchgenerators and trixi it provides a framework for

• Dataset loading
• Dataset sampling
• Augmentation (multi-threaded) including 3D images with any number of channels
• A generic trainer class that implements the training process
• Already implemented models used in medical image processing and exemplaric implementations of most used models in general (like Resnet)
• Web-based monitoring using Visdom
• Model save and load functions

Delira supports classification and regression problems as well as generative adversarial networks and segmentation tasks.

Installation

Choose Backend

Currently the only available backend is PyTorch (or no backend at all) but we are working on support for TensorFlow as well. If you want to add another backend, please open an issue (it should not be hard at all) and we will guide you during the process of doing so.

For instructions to install delira with a specific backend, please have a look at the corresponding docs

Installation without a backend (from source)

To install delira without a backend (not all functionalities may be work due to a missing backend) you can simply run:

• pip install git+https://github.com/justusschock/delira.git

Docker

The easiest way to use delira is via docker (with the nvidia-runtime for GPU-support) and using the Dockerfile or the prebuild-images.

Getting Started

The best way to learn how to use is to have a look at the tutorial notebook. Example implementations for classification problems, segmentation approaches and GANs are also provided in the notebooks folder.

Contributing

If you find a bug or have an idea for an improvement, please have a look at our contribution guideline.