This repo contains demo code for the paper, Fidelity imposed network edit (FINE) for solving ill-posed image reconstruction. It includes code and data to run FINE on 2016 QSM challenge, one in-house COSMOS test case and one simulated brain with hemorrhage.

This code requires the following:

• python 3.*
• TensorFlow v1.9+
• Keras v2.2.2+
• CUDA v10.1+

To run the code, simply

python demo_fine_challenge_2016.py

and

python demo_fine_ich_simu.py

and

python demo_fine_cosmos.py

To ask questions, pleasae contact [email protected]