CACR-Net

Code for JBHI 2021 "Multi-Modality MR Image Synthesis via Confidence-Guided Aggregation and Cross-Modality Refinement"

Usage

The original implementation of CACR-Net is Pytorch. The code has been tested in Linux.

To run the code, you should first install dependencies:

pip install fire

Setup all parameters in config.py

Put your data into ./data (Some samples from BraTs2018 have been stored out in this file)

Train

CUDA_VISIBLE_DEVICES=0,1 python main.py train --batch_size=8 --task_id=2 --gpu_id=[0,1]

The code of the CACR-Net is based on the Hi-Net [https://ieeexplore.ieee.org/abstract/document/9004544].