Welcome to PARROT

PARROT, which stands for Platform for ARtificial intelligence guided Radiation Oncology Treatment, is a user-friendly, free, and open-source web platform. It allows users to visualize DICOM files, run AI models, display and evaluate predictions easily. The platform includes several trained state-of-the-art dose prediction and contour segmentation models. Users can also add their own models using the embedded code editor.

The app consists of a frontend built in React JavaScript, including the graphical interface with a DICOM viewer, a selection panel of AI models, an editor of contours predictions, and evaluation tools. The backend is built with Flask to handle the predictions of AI models.

PARROT allows you to:

State-of-the-art models

Run state-of-the-art AI models

ct scan

Visualize contours segmented with AI on CT and MR images

ct scan

Compare uploaded and/or predicted dose distributions

ct scan

Correct contours predicted by AI model

ct scan

Full DICOM to DICOM workflow

ct scan

Local web application to ensure patient data privacy

ct scan

Run your own AI models

ct scan

Clinical evaluation tools: NTCP, DVH curves, clinical objectives, dose statistics