Oralytics

Oralytics is a low cost oral health scanner that uses 405 nm near-UV light to make dental plaque fluoresce, then analyzes photos inside a Flutter mobile app. The app highlights problem areas and returns a severity score in seconds.

This repository contains the mobile app (Flutter). The app connects to inference services or local model runners depending on your setup.

What it does

1. Capture or upload an oral photo (taken under near-UV illumination)
2. Run a pipeline of computer vision models
3. Overlay results (masks and highlights) and compute a severity score
4. Save results for tracking over time

Models in the pipeline

The app is built around four ML tasks:

• Teeth area-of-interest segmentation
  Isolates teeth from lips, tongue, and background.

• Plaque / biofilm segmentation
  Detects fluorescent regions associated with plaque and produces a pixel mask.

• Gingivitis detection
  Estimates inflammation risk from visual cues near the gumline.

• Calculus (tartar) classification
  Classifies tartar presence and severity based on texture and color patterns.

Setup

Prerequisites

• Flutter SDK installed
• Xcode (for iOS) and/or Android Studio (for Android)
• A device or simulator

Check your environment:

flutter doctor

Install dependencies

flutter pub get

Run the app

iOS simulator:

flutter run

Specific device:

flutter devices
flutter run -d <device_id>

How results are displayed

• Teeth and plaque are shown as overlays (segmentation masks)
• Gingivitis and calculus are shown as predictions with confidence and a severity score
• The app can store scan history to show progress over time (depending on your storage setup)

Notes and limitations

• Results depend heavily on lighting conditions and image quality
• Near-UV capture should follow consistent distance and exposure for best repeatability
• This tool is for research and educational use and is not a medical device

Contact

If you have questions about the app or setup, feel free to email [email protected]