/MRI-Harmonization-via-Disentangled-Latent-Energy-Based-Style-Translation

Project Repository for paper Disentangled Latent Energy-based Style Translation

Primary LanguagePython

Structural MRI Harmonization via Disentangled Latent Energy-Based Style Translation (DLEST)

This is the official repository for the DLEST paper.

Train Site-Invariant Image Generation Module

  1. Modify the YAML file in the /configs directory.
  2. Modify the dataloader.py and dataset_util.py to accommodate your dataset.
  3. Modify train_alae.py, set default_config= to use a specific YAML configuration file.

Train Site-Specific Style Translation Module

  1. Modify train_EBM.py, set default_config= to use a specific YAML configuration file.
  2. Run train_EBM.py to perform training.

Test Site-Specific Style Translation Module

  1. Modify test_EBM.py, set default_config= to use YAML configuration file used during training.
  2. Run test_EBM.py to perform style translation.

Acknowledgement

Our code was inspired by the latenet-energy-transport repository.