Development and Validation of AI System for Tooth Detection and Diagnosis in Dental Radiographs

Area of Science:

• Artificial Intelligence in Dentistry
• Medical Imaging Analysis
• Deep Learning for Dental Diagnostics

Background:

• Dental radiographic interpretation requires significant clinical expertise.
• Automating dental charting can enhance diagnostic efficiency and consistency.
• Current AI systems may struggle with diverse radiograph types and patient demographics.

Purpose of the Study:

• To develop and validate an AI-automated system for dental charting.
• To ensure the system accounts for multiple radiograph types and younger patients.
• To compare the AI system's performance against human dentists.

Main Methods:

• Collected 3705 dental radiographs (orthopantomograms and intraoral) from Slovakia and Egypt.
• Manually annotated teeth and dental findings by calibrated annotators and clinicians.
• Employed a 3-stage deep learning model for modality classification, tooth detection, and dental finding classification.

Main Results:

• Achieved high accuracy in tooth detection (F1-score: 0.98-0.99) and tooth numbering (0.96-0.98).
• Dental finding classification showed variable effectiveness (F1-score: 0.52-0.99), with lower accuracy for disease-related findings.
• The AI system surpassed dentists in identifying caries, crowns, fillings, primary teeth, root canal treatments, and unerupted teeth.

Conclusions:

• The AI system demonstrates superior performance for common dental findings compared to dentists.
• Further radiograph data is necessary to improve interpretation of rare dental conditions.
• The AI system can significantly support radiographic diagnosis and expedite dental charting.