We implemented stationary wavelet transform with the help of pytorch wavelets. You need to first install pytorch_wavelets to run wavelet transform.

git clone https://github.com/fbcotter/pytorch_wavelets
cd pytorch_wavelets
pip install .

You need save Mayo Clinic dataset properly. By default, we located our project repository and mayo clinic dataset as follows:

data
├── denoising
│   ├── train
│   │   └── mayo
│ │ ├── full_1mm
│ │ ├── full_3mm
│ │ ├── quarter_1mm
│ │ └── quarter_3mm
│ └── test
│      └── mayo
│ ├── full_1mm
│ ├── full_3mm
│ ├── quarter_1mm
│ └── quarter_3mm
works ── wavelet-ldct-denoising

• Training the model

python train.py --model <model> --datasets <list of data>

• Test the mode

python test.py --model <model> --test_datasets <list of data>

waveletdl is our proposed model trained with $L_{wo}$ and waveletganp is our proposed model trained with $L_{wp}$