Document Type : Systematic Review
Authors
1 Dept. of Information Technology Management, Faculty of Management and Economics, SR.C., Islamic Azad University, Tehran, Iran.
2 Dept. of Industrial Management, Ka.C., Islamic Azad University, Karaj, Iran.
3 Dental Research Center, Dentistry Research Institute, Tehran University of Medical Sciences, Tehran, Iran.
Abstract
Background: Dental caries remains one of the most prevalent oral diseases worldwide. Although cone beam computed tomography (CBCT) is not routinely indicated for caries detection due to radiation dose considerations, it provides three-dimensional (3D) information that may support caries assessment when CBCT scans are already clinically justified. Artificial intelligence (AI), particularly machine learning (ML) and deep learning approaches, has shown potential in improving diagnostic consistency and efficiency in dental imaging.
Purpose: This systematic review aimed to evaluate current evidence on the application of AI-based methods for dental caries detection and classification using CBCT images.
Materials and Method: A systematic literature search was conducted in PubMed, Scopus, Web of Science, IEEE Xplore, and Google Scholar for studies published between 2018 and 2024. Eligible studies applied AI techniques to CBCT images for caries detection and reported quantitative diagnostic performance metrics. Study selection, data extraction, and quality assessment were performed according to preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines.
Results: Five studies meeting the inclusion criteria were analyzed. Reported AI models, predominantly convolutional neural networks (CNNs), demonstrated sensitivity ranging from 86.67% to 96.5% and specificity from 91.3% to 99.46% in controlled study settings. In limited clinical evaluation studies, AI assistance improved clinician sensitivity while maintaining specificity and reducing assessment time.
Conclusion: AI-based approaches applied to CBCT images show promising preliminary diagnostic performance for caries detection in controlled research settings. However, the limited number of available studies, heterogeneity in study design, predominantly retrospective single-center designs, and lack of external validation restrict the generalizability and clinical applicability of these findings. Further large-scale, multi-center prospective investigations with external validation are required before routine clinical implementation can be recommended.
Keywords
)