Tuberculosis-Diagnosis-UDA

An unsupervised domain adaptation method for tuberculosis classification in CXR images.

data

• The model reads the dataset in the format of a JSON file

  - train
    - source
      - NM (normal)
      - TB (tuberculosis)
      - NTN (other lung diseases)
    - target
      - NM
      - TB
      - NTN
  - test
    - target
      - NM
      - TB
      - NTN
  

usage

• Collect the data you need and store it in the /data, and then create a JSON file

• Train a model:

  python main.py --config config/msdan.yaml --json_path data/yourDataJsonFile.json
  

Citation

If you use this code for your research, please cite our papers.