From Panoramic Radiographs to AI-Assisted Radiographic Periodontal Charting: Current Evidence and Future Perspectives

Background: Periodontal charting remains the gold standard for periodontal diagnosis but is time-consuming, operator-dependent, and not always feasible in routine clinical practice. In contrast, panoramic radiographs are widely available and increasingly amenable to automated analysis through artificial intelligence (AI). Recent AI systems have demonstrated high accuracy in the detection and quantification of periodontal bone loss and in radiographic disease staging. Objective: To review current evidence on AI applications in periodontal radiology and explore whether routinely acquired panoramic radiographs may support the development of AI-assisted radiographic periodontal charting. Methods: This narrative review summarizes recent evidence regarding AI-based periodontal and peri-implant radiographic assessment. The literature was identified through searches of PubMed, Scopus, and Google Scholar using combinations of terms related to artificial intelligence, periodontal disease, radiographic diagnosis, panoramic radiography, and peri-implantitis, with emphasis on studies relevant to AI-assisted periodontal assessment. Results: Recent deep learning architectures have demonstrated high diagnostic performance for periodontal bone loss detection, quantification, and disease staging. Current AI systems are capable of identifying radiographic manifestations of periodontal destruction and localising key anatomical landmarks, including the cemento-enamel junction and the alveolar bone crest. However, they remain unable to assess clinical parameters such as probing depth, bleeding on probing, suppuration, and tooth mobility. The review highlights the distinction between Clinical Attachment Level (CAL), Clinical Attachment Loss (CALoss), and radiographic measures of periodontal destruction. In this context, the concept of Radiographic Attachment Loss (RAL) is proposed as the radiographically measurable distance between the cemento-enamel junction and the alveolar bone crest, providing a conceptual estimate of periodontal support loss. Discussion: The principal novelty of this review is the proposal of AI-assisted radiographic periodontal charting as a clinically oriented framework for translating AI-derived radiographic information into meaningful periodontal assessment. Rather than replacing conventional periodontal charting, this approach aims to transform routinely acquired panoramic radiographs into a source of structured periodontal information that may support screening, risk stratification, longitudinal monitoring, and clinical decision-making. Conclusions: Artificial intelligence is unlikely to replace conventional periodontal examination. Nevertheless, AI-assisted radiographic periodontal charting may represent a realistic intermediate step between routine radiographic interpretation and comprehensive digital periodontal assessment. Future research should focus on validating AI-derived radiographic biomarkers, particularly Radiographic Attachment Loss (RAL), determining their clinical utility, and evaluating their potential contribution to periodontal screening, longitudinal monitoring and population-based oral health assessment.