This is the official implementation of the paper " ASCON: Anatomy-aware Supervised Contrastive Learning Framework for Low-dose CT Denoising" in 26th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2023). The pre-print version can be found in arxiv; camera-ready version will be soon released.
Sep, 2023: initial commit.
Dec, 2023: update data proprocessing file: /data/data_preprocessing.ipynb.
The 2016 AAPM-Mayo dataset can be downloaded from: CT Clinical Innovation Center (B30 kernel)
The 2020 AAPM-Mayo dataset can be downloaded from: cancer imaging archive
Mayo2016_2d/ |--train/ |--quarter_1mm/ train_quarter_00001.npy train_quarter_00002.npy train_quarter_00003.npy ... |--full_1mm/ train_full_00001.npy train_full_00002.npy train_full_00003.npy ... |--test/ |--quarter_1mm |--full_1mm - Linux Platform - torch==1.12.1+cu113 # depends on the CUDA version of your machine - torchvision==0.13.1+cu113 - Python==3.8.0 - numpy==1.22.3 Training
python train.py --name ASCON(experiment_name) --model ASCON --netG ESAU --dataroot /data/zhchen/Mayo2016_2d(path to images) --nce_layers 1,4 --layer_weight 1,1 --num_patches 32,512 --k_size 3,7 --lr 0.0002 --gpu_ids 6,7 --print_freq 25 --batch_size 8 --lr_policy cosine Inference & testing
python test.py --name ASCON(experiment_name) --model ASCON --netG ESAU --results_dir test_results --result_name ASCON_results(path to save image) --gpu_ids 6 --batch_size 1 --eval Please refer to options files for more setting.
If you find our work and code helpful, please kindly cite the corresponding paper:
@article{chen2023ascon, title={ASCON: Anatomy-aware Supervised Contrastive Learning Framework for Low-dose CT Denoising}, author={Chen, Zhihao and Gao, Qi and Zhang, Yi and Shan, Hongming}, journal={MCCAI 2023}, year={2023} }