This repository contains the code for the demo of the paper GANcMRI: Cardiac magnetic resonance video generation and physiologic guidance using latent space prompting by Milos Vukadinovic, Alan Kwan, Debiao Li and David Ouyang.
To run the demo you need to have a gpu, NVIDIA drivers, docker and NVIDIA container toolkit. If you don't see here how to install them. To install NVIDIA container toolkit on linux, run
apt -y install nvidia-container-toolkit
Step 1: Download weights (best_net.pkl) from https://drive.google.com/file/d/1_BqIPf5RfRCkukwFECftq7-Kz0dR0ok9/view?usp=drive_link and place it in this repo
Step 2:
docker build -t custom-tf-jupyter-image .
Step 3:
docker run --gpus device=0 -it --name gancmri -p 8888:8888 -v $(pwd):/data --rm custom-tf-jupyter-image
Step 4:
Click on Jupyter notebook link that was displayed by docker, and inside jupyter select a notebook Demo.ipynb.
Step 5:
Run All cells, the demo will be displayed in the last one
For inquiries about the repository, please create a GitHub issue.