Meta-analysis finds AI algorithms show high diagnostic accuracy in dental imaging tasks

OBJECTIVES: This systematic review and meta-analysis aimed to synthesize the available evidence on the use of AI in dental diagnostic decision-making and treatment planning, evaluating both diagnostic accuracy and its influence on clinical decision-making across different dental specialties and imaging modalities. METHODS: A comprehensive search of MEDLINE, Embase, Cochrane CENTRAL, Web of Science, and Scopus was conducted from database inception to December 2025. Eligible studies evaluated AI algorithms used for dental diagnostic tasks or treatment planning and reported quantitative performance metrics or measurable decision-making outcomes. Random-effects meta-analyses were conducted to pool diagnostic performance measures. RESULTS: Twenty-seven studies involving 60,857 radiographic images were included. AI systems demonstrated a pooled sensitivity of 0.85 (95% CI: 0.76-0.91) and specificity of 0.94 (95% CI: 0.86-0.97). The pooled F1-score was 0.90 (95% CI: 0.77-0.96), and pooled precision was 0.88 (95% CI: 0.71-0.96). For segmentation tasks, the pooled Dice Similarity Coefficient was 0.89 (95% CI: 0.13-1.00). Substantial heterogeneity was observed across studies (I² > 95%). YOLO-based architectures achieved the highest performance for tooth detection and segmentation, with sensitivities approaching 99% and mean average precision exceeding 0.96. AI assistance also improved diagnostic efficiency and interobserver agreement while reducing diagnostic interpretation time. CONCLUSIONS: AI systems demonstrate strong diagnostic performance in dental imaging and decision support, particularly for tooth detection, segmentation, and pathology identification. However, substantial heterogeneity, retrospective study designs, and limited external validation highlight the need for rigorous prospective evaluation before widespread clinical implementation.