Performance of Artificial Intelligence in the Diagnosis of Periapical Periodontitis Based on Cone-Beam Computed Tomography: A Systematic Review and Meta-Analysis

This systematic review assessed the diagnostic performance of artificial intelligence (AI) for detecting periapical periodontitis on cone-beam computed tomography (CBCT). A systematic literature search of Scopus, PubMed, Web of Science, and Embase was conducted to identify peer-reviewed diagnostic studies involving human participants and published up to August 30, 2026. Studies evaluating the application of AI for the detection of periapical periodontitis on CBCT were deemed eligible for inclusion. Conversely, studies focusing exclusively on periapical cysts, other odontogenic lesions, or root resorption were excluded. The risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies–Artificial Intelligence (QUADAS-AI) tool, while the certainty of the evidence for the pooled outcomes was evaluated according to the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach. Of 659 records identified through database searches, 18 studies were included in the systematic review and 6 in the meta-analysis. The exploratory HSROC synthesis yielded a pooled sensitivity of 0.91 (95% CI, 0.84-0.94) and a pooled specificity of 0.94 (95% CI, 0.80-0.98). Substantial between-study heterogeneity was observed, with I² values exceeding 98% for virtually all reported metrics, and the corresponding prediction regions were wide. Consequently, the pooled estimates should be interpreted solely as exploratory descriptive summaries rather than as clinically generalisable estimates of diagnostic performance. The GRADE certainty of evidence was rated as very low for each of the four pooled outcomes. The diagnostic performance and clinical utility of AI for the detection of periapical periodontitis on CBCT remain uncertain, owing to the extreme heterogeneity, wide prediction regions, and very low certainty of evidence identified in this review. These findings support further investigation rather than routine clinical implementation at present. AI warrants further evaluation as a potential adjunct to clinician-led assessment of periapical periodontitis. The current evidence does not establish clinical benefit or justify recommending routine use or replacement of professional judgement.