Groupwise image registration with edge-based loss for low-SNR cardiac MRI

This repo contains the official implementation for the paper Groupwise image registration with edge-based loss for low-SNR cardiac MRI.

Overview

AiM-ED is designed to perform registration for single-shot cardiac images with low signal-to-noise ratio (SNR), especially at low field strengths. Our method:

• Jointly registers multiple noisy source images to a noisy target image
• Utilizes a noise-robust pre-trained edge detector to define the training loss
• Produces high-quality cardiac MR images from free-breathing acquisitions

Setup

Clone the repository: git clone https://github.com/OSU-MR/aimed.git cd aimed

Create and activate a conda environment: conda env create -f environment.yml -n aimed

Usage

To get started:

1. Navigate to the notebooks directory

2. Launch Jupyter: jupyter notebook

3. Open the demo.iypnb to explore the implementation

Structure

• demo.ipynb: Jupyter notebooks demonstration of the AiM-ED framework
• data/: Example data
• utils/: Utility functions and helper scripts
• pre_trained_weights/: Pre-trained weights for digital phantom and healthy subjects from scanner

Method Validation

AiM-ED has been validated using:

• Synthetic late gadolinium enhanced (LGE) images from the MRXCAT phantom
• Free-breathing single-shot LGE images from healthy subjects and patients on 3T/1.5T scanners
• Clinical data from patients scanned on a 0.55T scanner

Model Training

1. Preparing your own training data. (intensity correction with SCC then normalization)
2. Save your datasets into .nii.gz files and add the name to predefined_dataset_idx.py file.
3. Train your own model using: python trainer.py --device_number 0 (don't forget to adjust the device number if you have more than one device)

Tip

Check the example datasets in niidata_c folder for more information (dimension, maximum magnitude, etc)

Contact

For questions or issues regarding this repository, please contact: Xuan Lei Email: lei.337{at}osu.edu

Citation

@article{lei2024image, title={Image Registration with Averaging Network and Edge-Based Loss for Low-SNR Cardiac MRI}, author={Lei, Xuan and Schniter, Philip and Chen, Chong and Ahmad, Rizwan}, journal={arXiv preprint arXiv:2409.02348}, year={2024} }