AI-based detection of gingivitis using intraoral photographs

1. Title:

Accuracy of Artificial Intelligence-Based Photographic Detection of Gingivitis

2. Authors:

Reinhard Chun Wang Chau; Guan-Hua Li; In Meei Tew; Khaing Myat Thu; Colman McGrath; Wai-Lun Lo; Wing-Kuen Ling; Richard Tai-Chiu Hsung; Walter Yu Hang Lam

3. Summary:

This study aimed to develop and validate a novel artificial intelligence (AI) system for the automated detection of gingivitis using frontal-view intraoral photographs. The researchers collected 567 photographs, which were labeled by a calibrated dentist into three categories: healthy, diseased, or questionable. The AI system, built using the DeepLabv3+ architecture, achieved a sensitivity of 0.92 and a specificity of 0.94, demonstrating performance nearly on par with visual examinations conducted by human dentists. The results suggest that this technology can be a valuable tool for monitoring patient plaque control and supporting community-based periodontal disease prevention

4. Key Words: Gingivitis; Periodontal diseases; Community dentistry; Deep learning; Neural networks; Artificial intelligence

5. Extracted data

5.1. Year: 2023

5.2. Modality: Intraoral photography (frontal-view clinical photographs)

5.3. Dataset: 567

5.4. Dataset Split: Training: 80% (453); Validation: 20% (114)

5.5. Network Architecture: DeepLabv3+ with Xception and MobileNetV2

5.6. Metrics: Sensitivity, Specificity, Mean Intersection-over-Union (mIoU), Accuracy

5.7. AP - Professional Qty: 1

5.8. AP - Supervisor Presence: No inforamtion

5.9. AP - Experience Level: Not specified

5.10. AP - Expertise Area: General Dentist

5.11. AP - Tool or System: No inforamtion

5.12. ML Task: Semantic Segmentation + Multi-class Classification

5.13. Project Objective: To validate an AI system for gingivitis detection

6. Clinical Relevance: Supports automated plaque control monitoring and early periodontal disease detection.