Related Experiment Video
Updated: Jun 23, 2026
Reliability of Artificial Intelligence-Based Cone Beam Computed Tomography Integration with Digital Dental Images
Published on: February 23, 2024
Artificial Intelligence Integrated with Intraoral Digital Imaging in Dental Caries Detection, Treatment Planning, and
Sarah Kayali1, Ali Golkari1, Manu Mathur1
1Institute of Dentistry, Queen Mary University of London, London, England, E1 2AH, UK.
Background:
The emergence of AI technologies has revolutionised dentistry, with intraoral imaging being a key area for innovation. Despite advances and growing interest in applying AI algorithms to intraoral x-rays, the methodological quality, diagnostic validity, and clinical applicability of existing studies remain unclear.
Aim:
To synthesise and critically appraise the current evidence on AI integrated with intraoral digital radiographic imaging for detecting dental caries in adults, focusing on diagnostic accuracy compared with gold-standard methods and examining methodological quality, clinical applicability, and implementation challenges.
Methods:
Following the JBI scoping review framework and PRISMA-ScR reporting guidelines, a comprehensive literature search was conducted across the PubMed, Scopus, and IEEE Xplore databases from January 2015 to May 2025. Studies that met the predefined eligibility criteria were included. Thematic analysis, combining inductive and deductive approaches following Braun and Clarke's framework, identified five themes. The CASP quality appraisal was performed to ensure methodological rigour.
Results:
Ten peer-reviewed studies were included in the final data analysis. AI systems detected a greater number of carious lesions than human clinicians, particularly in early-stage caries, with representative metrics including 88% sensitivity, 91% specificity, and 89% accuracy. Other models reported F1-scores up to 89% and AUC ≈95%. Methodological diversity was notable, with histology-validated designs providing the strongest evidence. Implementation challenges included limited external and real-world validation, clinician oversight, ethical/regulatory considerations, and inadequate model interpretability.
Conclusion:
AI exhibits strong potential to enhance early caries detection on intraoral radiographs and support clinical decision-making in adults. Fully realising AI's clinical potential requires overcoming implementation and methodological challenges. Standardised validation methods across diverse populations and settings are crucial to ensure AI diagnostic reliability and generalisability. Current AI applications in dentistry are primarily designed to assist clinicians in detecting caries; however, their greatest potential lies in a future where they can independently guide treatment planning decisions.
