Screening Patient Misidentification Errors Using a Deep Learning Model of Chest Radiography: A Seven Reader Study

This is a PyTorch implementation of the SimChest paper:

@article{kim2025screening,
  title={Screening Patient Misidentification Errors Using a Deep Learning Model of Chest Radiography: A Seven Reader Study},
  author={Kim, Kiduk and Cho, Kyungjin and Eo, Yujeong and Kim, Jeeyoung and Yun, Jihye and Ahn, Yura and Seo, Joon Beom and Hong, Gil-Sun and Kim, Namkug},
  journal={Journal of Imaging Informatics in Medicine},
  volume={38},
  number={2},
  pages={694--702},
  year={2025},
  publisher={Springer}
}

Pretraining task procedure

Model training

CUDA_VISIBLE_DEVICES=0,1,2,3 python main_supcon.py --dataset real --name SupCon_scratch \
--print_freq=5 --save_freq 1 --num_workers 8 --aug True --warm \
--batch_size 8 --model resnet50 --method SupCon --epochs 100

Pretrained model weight

Google Drive or you can use gdown in Python

!gdown https://drive.google.com/uc?id=1I8IgZ6mjPwvCPDk0EutkYH6UYOY865_n

Model inference

If you want to get CXR image similarity logit, run the code below.

python model_inference.py

Result

Contact

Page: https://mi2rl.co

Email: kiduk.amc@gmail.com

Reference

@Article{khosla2020supervised,
    title   = {Supervised Contrastive Learning},
    author  = {Prannay Khosla and Piotr Teterwak and Chen Wang and Aaron Sarna and Yonglong Tian and Phillip Isola and Aaron Maschinot and Ce Liu and Dilip Krishnan},
    journal = {arXiv preprint arXiv:2004.11362},
    year    = {2020},
}