Deep Learning-Based Measurement of Keratinized Gingiva Width Using Smartphone-Acquired Clinical Images

Periodontal Diseases

Artificial Intelligence, Deep Learning, Keratinized Gingiva Width, KGW, Periodontology, Clinical Photography, Image Segmentation, Automated Measurement, Periodontal Diagnosis, Dental Artificial Intelligence, Computer Vision, Smartphone Imaging

This study aims to develop and validate an artificial intelligence-based system for automated measurement of keratinized gingiva width using smartphone-acquired intraoral clinical photographs. Standardized intraoral images will be collected and analyzed using a deep learning model, and the results will be compared with clinical measurements performed by calibrated expert examiners, which serve as the reference standard. The performance of the proposed system will be evaluated using accuracy metrics including Dice coefficient, Intersection over Union (IoU), precision, recall, and F1-score. This study seeks to support the integration of AI tools into periodontal diagnosis and clinical decision-making to improve measurement consistency and reduce inter-examiner variability.

This observational diagnostic validation study was conducted to develop and evaluate an artificial intelligence-based system for automated assessment of keratinized gingiva width (KGW) using smartphone-acquired intraoral clinical photographs. Standardized intraoral images were collected from eligible participants following predefined inclusion and exclusion criteria. All images were captured using a smartphone under standardized clinical conditions to ensure uniformity in lighting, angulation, and image quality. Clinical measurements of keratinized gingiva width were independently performed by two calibrated expert examiners, serving as the reference (ground truth) standard. A deep learning-based model was trained to segment and measure the keratinized gingival tissue from clinical images. The predicted measurements generated by the AI system were compared against the expert clinical measurements to evaluate model performance. The performance of the system was assessed using multiple evaluation metrics, including accuracy, Dice similarity coefficient, Intersection over Union (IoU), precision, recall, and F1-score. Inter-examiner reliability between experts was also considered to ensure consistency of the reference standard. The study aims to demonstrate the feasibility of integrating artificial intelligence into periodontal diagnostics, specifically for objective and reproducible measurement of keratinized gingiva width. The proposed system may contribute to reducing inter-operator variability and improving clinical efficiency in periodontal assessment.

No related papers found yet.

Egypt