This repository contains code for the paper 'Learnable Douglas-Rachford Iteration and Its Applications in DOT Imaging'

Dataset

The Jacobian matrix and dataset are available at https://www.dropbox.com/sh/0fijlgtlmlm1y45/AAD9gXtxUUlzzXjWyYUjMJbHa?dl=0

The Jacobian matrix should be put in utility/

How to use the code

Training

python ./train_dot_recon.py --images-dataset dosdata_sim_exp_dataset.mat --measurements-dataset dosdata_sim_exp_dataset.mat --Jacobian-matrix utility/J.mat --batch-size 128 --inversion-regularize-method D --stages 6 --output-name net_configs/prediction_dot_D_stg6_.pth.tar --epochs 300

Testing

python ./predict_dot_recon.py --measurements dosdata_sim_exp_testset.mat --Jacobian-matrix utility/J.mat --batch-size 128 --inversion-regularize-method D --stages 6 --prediction-parameter net_configs/prediction_dot_D_stg6_.pth.tar --output-prefix output/all_D_stg6_