https://github.com/koflera/xtyt-cnn
Implementation of a 2D spatio-temporal CNN for artefacts-reduction in accelerated 2D cardiac cine MRI.
https://github.com/koflera/xtyt-cnn
deep-learning inverse-problems magnetic-resonance-imaging neural-networks
Last synced: 12 months ago
JSON representation
Implementation of a 2D spatio-temporal CNN for artefacts-reduction in accelerated 2D cardiac cine MRI.
- Host: GitHub
- URL: https://github.com/koflera/xtyt-cnn
- Owner: koflera
- License: apache-2.0
- Created: 2020-10-19T15:13:35.000Z (over 5 years ago)
- Default Branch: main
- Last Pushed: 2020-11-06T16:18:23.000Z (over 5 years ago)
- Last Synced: 2023-08-03T03:10:41.794Z (almost 3 years ago)
- Topics: deep-learning, inverse-problems, magnetic-resonance-imaging, neural-networks
- Language: Python
- Homepage:
- Size: 3.06 MB
- Stars: 5
- Watchers: 1
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# XTYT-CNN
Here, we provide a PyTorch implementation of our method proposed in the paper
"Spatio-Temporal Deep Learning-BasedUndersampling Artefact Reduction for 2D Radial Cine MRI WithLimited Training Data"
by A. Kofler, M. Dewey, T. Schaeffter, W. Wald, and C. Kolbitsch
see also: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8793147 or https://arxiv.org/abs/1904.01574
Unlike in the paper, this implementation allows the processing of complex-valued 2D cine MR images. This is done (as it's mostly the case) by stacking real- and imaginary parts of the images as channels. Further, the method implements the "change of perspective" on the data by switching to xt- and yt-domain within the network and thus allowing it to be ingrated as a building block in iterative or cascaded networks. In the original paper, on the other hand, the xt- and yt-slices are processed offline and the image series are obtained by reassembling the processed slices as a post-processing step.
Further, we provide the model (i.e. the weights) of the CNN E3 C4 K64 following the notation in the paper.
If you find the code useful or if you use for your work, please cite the following:
@article{kofler2019spatio, title={Spatio-temporal deep learning-based undersampling artefact reduction for 2D radial cine MRI with limited training data}, author={Kofler, Andreas and Dewey, Marc and Schaeffter, Tobias and Wald, Christian and Kolbitsch, Christoph}, journal={IEEE transactions on medical imaging},