JSMoCo

This repository is the PyTorch implementation of our MedIA paper, "Joint Coil Sensitivity and Motion Correction in Parallel MRI with a Self-Calibrating Score-Based Diffusion Model".

Fig. 1: The pipeline of JSMoCo.

Qualitative Results

Time-varying CSM estimation

Fig. 2: Qualitative comparisons of jointly estimated static and time-varying coil sensitivity maps (CSMs) by JSMoCo.

Motion Correction on in vivo T2w MRI data

Fig. 3: Quantitative comparisons of reconstruction results for motion-corrupted real-world T2w human brain MRI scans with an acceleration factor $R=2$.

Run a Demo

We provide a demo demo.ipynb to demonstrate how JSMoCo works. The diffusion model checkpoint is available at https://drive.google.com/uc?id=1vAIXf8n67yEAPmH2I9qiDWzmq9fGKPYL

In the demo notebook, you can find that:

• The process of simulating motion-corrupted acquisition,
• The motion-correction procedure performed by JSMoCo
• The motion-corrected results, the corresponding estimated motion parameters and coil sensitivity maps.

File Tree

C:.
│  csm_model.py      # CSM tools
│  demo.ipynb        # demo notebook
│  demo_config.json  # config file
│  jsmoco.py         # main moco function
│  model.py          
│  motion_ops.py     # motion forward and backward operation 
│  README.md         # readme file
│  utils.py          # tools
│
├─Data
│      full_ktraj.npy       # sampling trajectory of k-space
│      gt_coil_imgs.nii.gz  # GT multi-coil MRI complex image
│      gt_img.nii.gz        # GT coil-combined MRI complex image
│      ref_csm.nii.gz       # CSM estimated by ESPIRiT
│
├─diffusion_weight
├─Fig
│
├─ncsnv2              # diffusion model 
│
└─Nufft_Torch         # Nufft operator 

Main Requirements

To run this project, you will need the following packages:

• PyTorch
• ants
• dotmap
• SimpleITK
• tqdm
• numpy
• other dependencies

License

This code is available for non-commercial research and education purposes only. It is not allowed to be reproduced, exchanged, sold, or used for profit.

Citations

If you find our work useful in your research, please site:

@article{chen2025joint,
  title={Joint coil sensitivity and motion correction in parallel MRI with a self-calibrating score-based diffusion model},
  author={Chen, Lixuan and Tian, Xuanyu and Wu, Jiangjie and Feng, Ruimin and Lao, Guoyan and Zhang, Yuyao and Liao, Hongen and Wei, Hongjiang},
  journal={Medical Image Analysis},
  pages={103502},
  year={2025},
  publisher={Elsevier}
}

Our code uses the prior work from the following papars, which must be cited

@inproceedings{song2019generative,
  title={Generative modeling by estimating gradients of the data distribution},
  author={Song, Yang and Ermon, Stefano},
  booktitle={Advances in Neural Information Processing Systems},
  pages={11918--11930},
  year={2019}
}

@article{song2020improved,
  title={Improved Techniques for Training Score-Based Generative Models},
  author={Song, Yang and Ermon, Stefano},
  booktitle={Advances in Neural Information Processing Systems},
  year={2020}
}

@article{levac2022motion,
  title={Accelerated Motion Correction for MRI using Score-Based Generative Models},
  author={Levac, Brett and Jalal, Ajil and Tamir, Jonathan I},
  journal={arXiv preprint arXiv:2211.00199},
  year={2022}
}

@article{jalal2021robust,
  title={Robust Compressed Sensing MRI with Deep Generative Priors},
  author={Jalal, Ajil and Arvinte, Marius and Daras, Giannis and Price, Eric and Dimakis, Alexandros G and Tamir, Jonathan I},
  booktitle={Advances in Neural Information Processing Systems},
  year={2021}
}

We use data from the NYU fastMRI dataset, which must also be cited:

@inproceedings{zbontar2018fastMRI,
    title={{fastMRI}: An Open Dataset and Benchmarks for Accelerated {MRI}},
    author={Jure Zbontar and Florian Knoll and Anuroop Sriram and Tullie Murrell and Zhengnan Huang and Matthew J. Muckley and Aaron Defazio and Ruben Stern and Patricia Johnson and Mary Bruno and Marc Parente and Krzysztof J. Geras and Joe Katsnelson and Hersh Chandarana and Zizhao Zhang and Michal Drozdzal and Adriana Romero and Michael Rabbat and Pascal Vincent and Nafissa Yakubova and James Pinkerton and Duo Wang and Erich Owens and C. Lawrence Zitnick and Michael P. Recht and Daniel K. Sodickson and Yvonne W. Lui},
    journal = {ArXiv e-prints},
    archivePrefix = "arXiv",
    eprint = {1811.08839},
    year={2018}
}

@article{knoll2020fastmri,
  title={fastMRI: A publicly available raw k-space and DICOM dataset of knee images for accelerated MR image reconstruction using machine learning},
  author={Knoll, Florian and Zbontar, Jure and Sriram, Anuroop and Muckley, Matthew J and Bruno, Mary and Defazio, Aaron and Parente, Marc and Geras, Krzysztof J and Katsnelson, Joe and Chandarana, Hersh and others},
  journal={Radiology: Artificial Intelligence},
  volume={2},
  number={1},
  pages={e190007},
  year={2020},
  publisher={Radiological Society of North America}
}