Official implementation of BlindHarmony: Blind harmonization for MR image
H. Jeong, H. Byun, D. Kang, and J. Lee, BlindHarmony: “Blind” Harmonization for MR Images via Flow model, ICCV 2023, [arXiv]
Neural spline flow is used for flow model training. See https://github.com/bayesiains/nsf.git
Use environment.yml
for required packages, or create a Conda environment with all dependencies:
conda env create -f environment.yml
The whole data of OASIS-3 can be accessed by https://www.oasis-brains.org.
The checkpoints can be downloaded from google drive link in https://drive.google.com/drive/folders/1AuCYGiNOZ8hWrqiV_npsjmcodNVfRb6z?usp=share_link
DATAROOT
environment variable needs to be set before running experiments.
Use train_flow.py
.
Use BlindHarmony_simulated_data.py
.
Use BlindHarmony_real_data.py
.