krakenleaf / delimit

PyTorch implementation of the paper "DELIMIT PyTorch - An extension for Deep Learning in Diffusion Imaging"

DELIMIT

This repository contains the PyTorch implementation of our Paper "DELIMIT PyTorch - An extension for Deep Learning in Diffusion Imaging".

Contents

• Installation
• Requirements
• Usage
  • By Scripts
  • Data Preprocessing
• Our Paper

Requirements

required packages are: scipy, numpy, torch, pyquaternion

Installation

pip install git+https://github.com/SimonKoppers/DELIMIT

Usage

By Scripts

For simplicity a simple example jupyter notebook is provided within the github repository. It requires the package 'jupyter' to be run correctly.

Data Preprocessing

In order to apply Spherical Harmonic Transformations, or any kind of convolution, the diffusion signal needs to be divided by its b=0 measurement.

Our Paper

If you use our Code for your own research, please cite our paper:

@article{Koppers2018,
title = {DELIMIT PyTorch - An extension for Deep Learning in Diffusion Imaging},
author = {Simon Koppers and Dorit Merhof},
year = {2018},
journal = {arXiV preprint}
}

The Paper is available as PDF on arXiv.