NOTE: required to install nibabel to save DTI filtering results
(link: https://nipy.org/nibabel/)

NOTE: required to install torch_batch_svd for meanshift filtering
(link: https://github.com/KinglittleQ/torch-batch-svd)

NOTE: please see the notebook files for the essential codes for training GDAE and DTI filtering

NOTE: below codes are used in the notebook files and experiments

[source codes for model]
gae2.py

[source codes for DTI dataset]
data_util.py
tensor_data_util.py
DTI_dataset.py

[source codes for model train]
gae2_trainer.py
util.py

[source codes for score estimation]
gae2_score_estimation.py

[source codes for filtering]
DTI_meanshift2.py
DTI_pointgrouper2.py
DTI_ms_utils.py

NOTE: below codes are used in the experiments

[source codes for N(2) filtering]
N2_filtering.py

[source codes for MVKR method]
MVKR_for_spd_field.py
