A Novel Artificial Intelligence Approach to Kennedy Classification for Partially Edentulous Patients Using Panoramic...

Area of Science:

• Artificial Intelligence in Dentistry
• Radiographic Analysis
• Dental Prosthodontics

Background:

• Partially edentulous arches require accurate classification for treatment planning.
• The Kennedy classification system is a standard for categorizing these dental conditions.
• Manual classification can be time-consuming and prone to variability.

Purpose of the Study:

• To develop an artificial intelligence (AI) system for automated classification of partially edentulous arches.
• To utilize the Kennedy classification system and Applegate's rules for automated reporting.
• To identify existing teeth within panoramic radiographs.

Main Methods:

• A YOLOv8s deep learning model was trained on 1875 panoramic radiographs.
• Data augmentation increased the dataset size to 2398 images.
• The model detected teeth, classified arches using the Kennedy system, and identified modification areas.

Main Results:

• The YOLOv8s model achieved 98.1% mean average precision for tooth identification.
• Kennedy classification demonstrated high accuracy with 0.962 precision, 0.931 recall, and 0.939 F1-score.
• The system accurately identified existing teeth and classified edentulous areas.

Conclusions:

• The AI system offers a consistent and accurate method for classifying partially edentulous arches.
• This approach reduces diagnostic variability and practitioner workload.
• The technology enables seamless integration into clinical practice and large-scale analysis.