PyTorch implementation of J-MoDL: Joint Model-Based Deep Learning for Optimized Sampling and Reconstruction (Not official!)
Official code: https://github.com/hkaggarwal/J-MoDL
J-MoDL: Joint Model-Based Deep Learning for Optimized Sampling and Reconstruction by H.K Aggarwal and M. Jacob in IEEE Journal of Selected Topics in Signal Processing, (2020).
Link: https://arxiv.org/abs/1911.02945
IEEE Xplore: https://ieeexplore.ieee.org/document/9122388
Subset of the multi-coil brain dataset used in the original paper is publically available. Test dataset tstdata_jmodl.npz and initial mask initmask6.npz are already included in the data folder. Please download the train dataset from the following link and locate in under the data directory.
Download Link : https://drive.google.com/file/d/1GLqs2A5YpRN8RdDJgdhrspL3zjlG0Qha/view?usp=sharing
The configuration files are in config folder. Every setting is the same as the authors used in their official repo, but not the same as the ones used in the paper.
You can change the configuration file for training by modifying the train.sh file.
scripts/train.sh
You can change the configuration file for testing by modifying the test.sh file.
scripts/test.sh
Saved models are provided.
workspace/base_modl/checkpoints/final.epoch0099-score38.9911.pth
Result image: