README.md

metadata

title: Dentimap - Dental X-Ray Segmentation
colorFrom: blue
colorTo: indigo
sdk: docker
pinned: false
license: mit
app_port: 7860

Dentimap 🦷

Dental X-Ray Segmentation API - Automated segmentation and analysis of dental structures from panoramic X-ray images using deep learning.

Features

• Automatic Segmentation: Segments dental X-rays into multiple anatomical classes
• Visual Overlay: Colored segmentation overlaid on original X-ray for easy interpretation
• Fast Inference: CPU-optimized for quick predictions
• Detailed Analytics: Provides class distribution and pixel-level analysis
• REST API: Easy-to-use FastAPI backend

Supported Formats

• JPEG/JPG
• PNG
• BMP

Maximum file size: 10 MB

Model Architecture

• Base Model: U-Net with encoder-decoder architecture
• Input Size: 512x256 pixels
• Output Classes: 4 (background, teeth, dental structures, anomalies)
• Training Dataset: 997 training images, 125 validation, 125 test

API Endpoints

Health Check

GET /health

Predict Segmentation

POST /api/v1/inference/predict
Content-Type: multipart/form-data
Body: file (dental X-ray image)

Returns: Segmented image with overlay (JPEG)

Predict with Metadata

POST /api/v1/inference/predict-with-metadata
Content-Type: multipart/form-data
Body: file (dental X-ray image)

Returns: JSON with base64 encoded image, inference time, class distribution

Usage

Upload a dental X-ray image to get:

1. Segmented overlay: Visual representation of detected dental structures
2. Class distribution: Percentage breakdown of each anatomical class
3. Inference time: Processing time for the analysis

Technical Details

• Framework: FastAPI, PyTorch
• Image Processing: OpenCV, NumPy
• Preprocessing: Histogram equalization for enhanced X-ray contrast
• Visualization: Multi-color overlay with adjustable opacity

Performance

Class	F1 Score	IoU
Background	99.7%	99.4%
Teeth	50.3%	37.8%
Dental Structure	20.4%	15.9%
Anomaly	24.3%	21.1%

Note: This model is a research prototype. Performance varies based on X-ray quality and imaging conditions. Not intended for clinical diagnosis without professional oversight.

Development

Built with:

• Python 3.9
• PyTorch 2.0+
• FastAPI 0.104+
• OpenCV 4.8+

License

MIT License

Disclaimer

This application is for research and educational purposes only. It should not be used as a substitute for professional medical diagnosis or treatment. Always consult qualified healthcare professionals for medical advice.