This repository consists of multiple work packages. The main package is phase_gen, which includes the complex-valued diffusion model for raw data generation.
Each package has its own documentation. You are able to run each package as a standalone.
For more detailed instructions, please find the documentation in each workpackage.
-
Clone repository:
git clone https://github.com/TIO-IKIM/PhaseGen.git -
Create a conda environment with Python version 3.12.4 and install the necessary dependencies:
cd PhaseGen conda create -n my_env python=3.12.4conda activate my_env pip install -r requirements.txt -
Torch is already included in the requirements.txt file. Nevertheless, you should install it separately to ensure that the correct version is installed. You can find the appropriate command for your system at PyTorch.
Model weights of PhaseGen will be uploaded soon. Until then the weights are available upon request.
@misc{rempe2025phasegendiffusionbasedapproachcomplexvalued,
title={PhaseGen: A Diffusion-Based Approach for Complex-Valued MRI Data Generation},
author={Moritz Rempe and Fabian Hörst and Helmut Becker and Marco Schlimbach and Lukas Rotkopf and Kevin Kröninger and Jens Kleesiek},
year={2025},
eprint={2504.07560},
archivePrefix={arXiv},
primaryClass={eess.IV},
url={https://arxiv.org/abs/2504.07560},
}