Systematic Review Of Artificial Intelligence Applications In Dental Radiology Diagnostic Accuracy For Caries, Periodontal Disease, And Orthodontic Assessment

Abstract

Background: The rapid advancement of artificial intelligence (AI) has significantly transformed diagnostic practices in dental radiology. AI systems—particularly those based on deep learning—have demonstrated promising capabilities in automating image interpretation, reducing diagnostic variability, and enhancing the early detection of dental conditions. However, evidence regarding their diagnostic accuracy across major dental pathologies, including dental caries, periodontal disease, and orthodontic assessment, remains scattered and inconsistent. Objective: This systematic review aims to critically evaluate and synthesize current evidence on the diagnostic performance and clinical utility of AI applications in dental radiology, specifically focusing on the detection of caries, assessment of periodontal disease, and support for orthodontic evaluation. Methods: A comprehensive search of major databases (PubMed, Scopus, Web of Science, and IEEE Xplore) was conducted for studies published from inception to 2025. Studies evaluating AI algorithms—including convolutional neural networks, machine learning classifiers, and hybrid models—used for diagnostic tasks in dental radiographic imaging were included. Data extraction followed PRISMA 2020 guidelines, and study quality was appraised using QUADAS-2. Diagnostic accuracy metrics such as sensitivity, specificity, AUC, and F1-scores were synthesized narratively due to heterogeneity across AI models and imaging modalities. Results: Across the included studies, AI systems demonstrated high diagnostic accuracy for caries detection, with several deep learning models achieving sensitivity and specificity values exceeding 0.85 in bitewing and periapical radiographs. For periodontal disease, AI showed strong performance in detecting bone loss and periodontal defects, although accuracy varied according to annotation quality and imaging type. Orthodontic applications—such as cephalometric landmark detection, skeletal classification, and treatment planning support—achieved high precision, with landmark localization errors frequently below 2 mm. Despite these strengths, inconsistencies in dataset size, external validation, and reporting standards were common limitations. Conclusion: AI applications in dental radiology exhibit substantial potential to enhance diagnostic accuracy for caries, periodontal disease, and orthodontic assessment. While current evidence supports integration into clinical workflows, the lack of standardized validation, limited generalizability across populations, and variability in imaging quality highlight the need for robust, multicenter prospective studies. Strengthening methodological consistency will be critical for translating AI-driven diagnostic tools into routine dental practice.