MammoScreen

MammoScreen is a browser-based Streamlit application for reviewing and labeling mammography images, including DICOM uploads processed fully in memory.

Features

• Upload JPG, JPEG, PNG, BMP, and DICOM files from the browser
• View one image at a time with previous/next navigation
• Label images as Positive, Negative, Uncertain, or Skip
• Edit earlier labels and add optional notes
• Export labels as CSV
• Keep uploaded data in session memory instead of writing it to disk

Project Layout

MammoScreen/
├── app.py
├── src/
├── tests/
├── sample_data/
├── legacy/
├── .streamlit/
├── Dockerfile
└── docker-compose.yml

The original Tkinter desktop application is preserved in legacy/MammoScreen.py.

Local Development

pip install -r requirements.txt
streamlit run app.py

The app will open on http://localhost:8501.

Testing

pytest

Docker

docker compose up --build

Screenshot

License

MIT