Hybrid-CS-Model-MRI

KO | EN

gitlite — search

#javascript #python #typescript #llm #rust #deep-learning #linux #nodejs #windows #vue #pytorch #security #ios #mcp

Hybrid-CS-Model-MRI

(★ 59)

Hybrid k-space/image domain, W-net, model for compressed sensing magnetic resonance reconstruction.

파일 탐색기

최종 버전 다운로드 (.zip)

- sampling_mask_20perc.npy
- sampling_mask_25perc.npy
- stats_fs_unet_norm_20.npy
- hybrid_5x.gif
- sample_rec.png
- w-net.png
- - w-net-DC-train-checkpoint.ipynb
  - w-net-metrics-checkpoint.ipynb
  - w-net-test-checkpoint.ipynb
  - w-net-train-checkpoint.ipynb
- w-net-DC-train.ipynb
- w-net-metrics.ipynb
- w-net-test.ipynb
- w-net-train.ipynb
- wnet_20.hdf5
- wnet_dc20.hdf5
- frequency_spatial_network.py
- frequency_spatial_network.pyc
.gitignore
LICENSE
README.md

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Hybrid-CS-Model-MRI

Reconstructing magnetic resonance (MR) images from undersampled k-space measurements can potentially decrease MR examination times. This repository contains thesource code for a hybrid k-space/image domain model proposed by our group. Our model is called W-net. If you use this code in your experiments, we ask you to kindly cite our paper:

[Souza, Roberto, and Richard Frayne. "A hybrid frequency-domain/image-domain deep network for magnetic resonance image reconstruction." In 2019 32nd SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), pp. 257-264. IEEE, 2019.](https://ieeexplore.ieee.org/abstract/document/8919674?casa_token=8UZWj9Zs9EoAAAAA:zIhNEBpZ8wkkpksvVNJqV4OQ9HBB-XKJaI5RQ--3GF9xzER00-_LOPEcgMO0vXjbce1utdzuXyI).

## Dataset

The MR raw-data is publicly available as part of the [Calgary-Campinas dataset](https://sites.google.com/view/calgary-campinas-dataset/home) for benhcmarking purposes.

## Code
The code was developed using Python 2.7, NumPy, TensorFlow and Keras. It should be easy to use it, but we appreciate any feedback on how to improve our repository.

## W-net
![w-net architecture](https://raw.githubusercontent.com/rmsouza01/Hybrid-CS-Model-MRI/master/Figs/w-net.png)
W-net architecture. It is composed of a residual U-net on k-space domain connected to an image domain U-net through the magnitude of the inverse discrete fourier transform operator.

![Sample Reconstruction](https://raw.githubusercontent.com/rmsouza01/Hybrid-CS-Model-MRI/master/Figs/sample_rec.png)
Sample reconstruction of our W-net and four other methods published in the literature with special highlight on the cerebellum region. Speed-up factor of 5x.

![w-net reconstruction](https://raw.githubusercontent.com/rmsouza01/Hybrid-CS-Model-MRI/master/Figs/hybrid_5x.gif)

W-net reconstruction gif. From left to right: fully sampled reconstruction, W-net reconstruction from a k-space undersample by 80%, and absolute error differences.

## Contact
Any questions? roberto.medeirosdeso@ucalgary.ca

Hybrid-CS-Model-MRI

파일 탐색기

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Hybrid-CS-Model-MRI

파일 탐색기

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README