CNCL_Medical_Image_Denoising

Pytorch implementation of Content-Noise Complementary Learning for Medical Image Denoising.

File Description

1. code

DataLoader_train.py: python script to build pytorch Dataset and DataLoader for training.

DataLoader_test.py: python script to build pytorch Dataset and DataLoader for test.

model.py: the implementation of our proposed CNCL-U-Net achitectures.

train.py: a basic template python file for training the model.

test.py: a basic template python file for testing the model.

2. data

clean: clean medical images (i.e., content images), such as full-dose CT images.

noisy: noisy medical images (i.e., noise corrupted-images), such as low-dose CT images.

dataset_division: contain four .txt files, storing the file names of the images in the training set or test set, respectively.

3. result

saved_models: save the trained models. test: save the predicted content images when testing.