Objectives: This scoping review aimed to map current research on CNN-based imaging using dental radiographs and to examine methodological trends in jaw lesion imaging. The goal is to inform future objectives and methodological frameworks that promote evidence validation, ultimately aiding policymakers and clinicians in deciding whether to implement AI tools.

Materials and methods: The review included studies that classify, detect, and segment abnormalities and diseases using CNNs, yielding 128 eligible studies. Selected studies were broadly mapped under the ICD-10 classification of diseases and the type of radiography used. Twenty-three studies developing CNNs for jaw lesion imaging were analyzed concerning their methodology. 

Results: Our results reveal that CNN-based imaging is rapidly being explored for a wide range of abnormalities and diseases, with the majority utilizing panoramic radiographs. Among these, CNN-based imaging for cysts and tumors is a relatively active field. Findings from this study reveal methodological heterogeneity and incomplete reporting of methodological details, which risks hindering reproducibility and validation of these studies.

Conclusion: Future research should prioritize methodological standardization and reporting frameworks to ease future validation efforts.