segmentation_RT: Deep Learning Segmentation Toolbox for radiotherapy

Introduction

Automatic contouring with the help of deep learning is a nontrivial matter and a research area of its own.

The purpose of segmentation_RT is to provide a quick method to use deep learning for contouring. Three modules are provided to do so:

• rs2mask: create a dataset from dicom data.
• dl: deep learning module for training and testing.
• mask2rs: create RT Structure Set from mask.

Installation

segmentation_RT works with Python 3.8.

Dependencies

This project is based on PyTorch 1.7.1 and uses TorchIO 0.18.29.

How to use it

The main segmentation_RT documentation is available online at http://segmentation_RT.readthedocs.io. There are also an main example available at https://github.com/BrouBoni/segmentation_RT/blob/main/main.py showing the interactions between the modules.