Validity and reliability of intraoral photographs for tooth-level caries classification using ICCMS™ in preschool children

Objectives: The International Caries Classification and Management System (ICCMS™) provides a structured framework for comprehensive caries diagnosis and management. However, its clinical application in young children is often limited by logistical challenges. This study aimed to evaluate the diagnostic validity and inter-method reliability of intraoral photographic assessment for classifying tooth-level caries using ICCMS™ merged categories, compared with conventional clinical visual-tactile examination in preschool children. Methods: A cross-sectional study was conducted among 280 preschool children aged 3–4 years from public kindergartens in Ho Chi Minh City. Each child underwent a full-mouth clinical examination using the ICDAS™ protocol, followed by intraoral photographs taken under standardized lighting and positioning conditions. Caries lesions were reclassified into ICCMS™ merged categories: 0 (sound), A+/− (initial), B+/− (moderate), and C+/− (extensive). Blinded evaluators assessed all photographs independently. Diagnostic validity (sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV]) and reliability (Cohen’s Kappa, intraclass correlation coefficient [ICC]) were analyzed at the tooth level using confusion matrices and F1-scores. Results: A total of 5,523 teeth were evaluated. Overall agreement between photographic and clinical examinations was substantial (Cohen’s Kappa = 0.772). Diagnostic accuracy was highest for sound teeth (category 0: sensitivity 91.6%, specificity 89.9%) and extensive lesions (C+/−: sensitivity 93.4%, specificity 97.4%). Moderate agreement was observed for early-stage lesions (A+/−: Kappa = 0.684; sensitivity 74.6%) and moderate lesions (B+/−: Kappa = 0.640; sensitivity 60.5%), indicating underdetection of subtle enamel changes in photographs. F1-scores were highest for categories 0 (0.849) and C+/− (0.937), confirming strong diagnostic performance for healthy and advanced lesions. Conclusions: Intraoral photography demonstrates substantial validity and reliability for tooth-level caries classification using ICCMS™ in preschool children, particularly for identifying sound and extensive lesions. While this approach is promising for teledentistry, epidemiological surveillance, and preventive dentistry programs, its reduced sensitivity for early lesions suggests the need for higher-resolution imaging or adjunctive detection technologies.