Deep Learning for Dental Imaging
We are a small independent research team focused on medical imaging analysis using deep learning. Our work applies state-of-the-art neural network architectures to dental imaging analysis.
Our research spans two complementary domains: Endodontics (3D CBCT tooth segmentation and root canal measurement) and Prosthodontics (2D panoramic X-ray analysis for edentulous region segmentation).
Explore our research through interactive notebooks. Each demonstration allows you to upload your own dental images and observe the segmentation pipeline in action.
Prosthodontics
Multi-stage pipeline combining YOLO and nnU-Net for comprehensive panoramic dental X-ray analysis. Identifies edentulous regions and segments present teeth with physical measurements.
Endodontics
Dual-branch nnU-Net V2 architecture for tooth instance segmentation in cone-beam CT (CBCT) scans. Provides individual tooth identification with FDI numbering system and 3D visualization.
Self-configuring deep learning framework that automatically adapts to dataset characteristics for optimal segmentation performance.
Dual-branch network combining semantic segmentation with instance discrimination for individual tooth identification.
Integrates YOLO detection with nnU-Net segmentation for comprehensive analysis of 2D panoramic radiographs.
Post-processing techniques for instance separation and anatomical structure identification.