Hybrid optimization between iterative and network fine-tuning reconstructions for fast quantitative susceptibility mapping
This repo contains demo code for the papers, Hybrid optimization between iterative and network fine-tuning reconstructions for fast quantitative susceptibility mapping, and Fidelity imposed network edit (FINE) for solving ill-posed image reconstruction.
This code requires the following:
- python 3.*
- pytorch v1.7.1+
- CUDA v10.1+
COSMOS dataset pre-training:
main_COSMOS_2nets.py
Amortized FINE domain adaptation:
main_FINE_2nets_all.py
Hybrid optimization between iterative and network fine-tuning:
main_HOBIT_resnet.py
To ask questions, pleasae contact jz853@cornell.edu