Skip to content
/ MRPD Public

Official implementation of "MRPD: Undersampled MRI Reconstruction by Prompting a Large Latent Diffusion Model" by Ziqi Gao and S. K. Zhou.

Notifications You must be signed in to change notification settings

Z7Gao/MRPD

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MRPD

Official implementation of "MRPD: Undersampled MRI Reconstruction by Prompting a Large Latent Diffusion Model" by Ziqi Gao and S. K. Zhou.

[06/28/2024] This repo is still under construction and contains some draft code for the basic reproducibility.

Validation Data

We put the first volume in the official validation set of the single-coil FastMRI knee dataset inside data/example/val folder. Note that the last number denotes the slice number, and that the first five and the last five noisy slices are removed as stated in the paper.

Repository Structure

All the experiments are issued in the form of self-explanatory python codes. To execute each code, we provide python scripts inside stable-diffusion/scripts/ folder. You can run python inverse_fastMRIknee_recon.py --ckpt [LDM_CHECKPOINT_PATH] for the validation set reconstruction. Currently you can download the LDM checkpoints from Latent Diffusion.

Credits

This codebase is modified based on PSLD. Thanks for creating this awesome repo.

We are also thankful for the developers of the following repositories: Latent Diffusion for generative foundation model's pretrained weights; FastMRI and Score-MRI for some MRI helper functions.

Citation

If you find this work helpful, please consider citing

@misc{gao2024mrpd,
      title={U$^2$MRPD: Unsupervised undersampled MRI reconstruction by prompting a large latent diffusion model}, 
      author={Ziqi Gao and S. Kevin Zhou},
      year={2024},
      eprint={2402.10609},
      archivePrefix={arXiv},
      primaryClass={eess.IV},
      url={https://arxiv.org/abs/2402.10609}, 
}

To-do

Clean unecessary files and code inherited from PSLD.

About

Official implementation of "MRPD: Undersampled MRI Reconstruction by Prompting a Large Latent Diffusion Model" by Ziqi Gao and S. K. Zhou.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published