Generative pretrained models on MRI images.
The prior distribution of MRI images learned with generative models has proven to be effective in MRI image reconstruction. Here, we include four PixelCNN models and two diffusion models, one is SMLD and the another one is DDPM. These models are trained with spreco. For more details on how these models were trained, please find them in our paper and the related codes.
How to use
The Berkeley Advanced Reconstruction Toolbox, (BART),
provides many functionalities for MRI image reconstruction.
It introduced the application of the TensorFlow graph as regularization
in this paper. You can try it on colab.
Prior | Model | Phase | Size | Contrast | Subscript |
---|---|---|---|---|---|
PixelCNN | preserved | 1000 | T1, T2, T2-FLAIR, | SC - Small, complex | |
PixelCNN | unknown | 1000 | T1, T2, T2-FLAIR, | SM - Small, magnitude | |
PixelCNN | unknown | ~20000 | MPRAGE | LM - Large, magnitude | |
PixelCNN | generated | ~20000 | MPRAGE | LC - Large, complex | |
Diffusion | generated | ~80000 | MPRAGE | SC - SMLD, complex | |
Diffusion | generated | ~80000 | MPRAGE | PC - DDPM, complex |
Citation
- Luo, G, Blumenthal, M, Heide, M, Uecker, M. Bayesian MRI reconstruction with joint uncertainty estimation using diffusion models. Magn Reson Med. 2023; 1-17
- Blumenthal, M, Luo, G, Schilling, M, Holme, HCM, Uecker, M. Deep, deep learning with BART. Magn Reson Med. 2023; 89: 678- 693.
- Luo, G, Zhao, N, Jiang, W, Hui, ES, Cao, P. MRI reconstruction using deep Bayesian estimation. Magn Reson Med. 2020; 84: 2246-2261.