mxliu/Unpaired-Volumetric-MRI-Harmonization-with-Conditional-Latent-Diffusion

Official Repository for paper Unpaired Volumetric Harmonization of Brain MRI with Conditional Latent Diffusion

Unpaired Volumetric Harmonization of Brain MRI with Conditional Latent Diffusion (HCLD)

Official Repository for the paper: "Unpaired Volumetric Harmonization of Brain MRI with Conditional Latent Diffusion"

Authors: Mengqi Wu, Minhui Yu, Shuaiming Jing, Pew-Thian Yap, Zhengwu Zhang, Mingxia Liu*

Train 3D Autoencoder

1. Modify the data/MRIdata.py file to configure it with custom MRI datasets.
2. Run python train_aekl.py to train the 3D Autoencoder.

Train the Conditional Latent Diffusion (cLDM) Module

1. Load the checkpoint from the previously trained 3D autoencoder in train_HCLD.py.
2. Run python train_HCLD.py to train the cLDM.

Model Inference

1. Load the checkpoint from the previously trained 3D autoencoder in infer_HCLD_DDIM.py.
2. Load the checkpoint from the previously trained cLDM in infer_HCLD_DDIM.py.
3. Run python infer_HCLD_DDIM.py to perform model inference.

Acknowledgement

Our code was implemented using the MONAI framework