Dental age estimation: a scoping review comparing the manual application of the Demirjian method and artificial intelligence modalities
摘要
The investigation of the viability using Artificial intelligence (AI) compared to the Demirjian method in dental age estimation is growing. This scoping review looked at how AI-assisted dental age estimation models compare to the Demirjian method and their usefulness to general and forensic dentists. Searches were conducted in Medline, Embase, Global Health, Scopus, Social Science Premium Collection, CENTRAL-Cochrane Library Central Register of Controlled Trials, and NZLII-New Zealand Legal Information Institute. Google and Google Scholar searches were also undertaken to identify any grey material. Twenty-two articles were included in this review, organised into three groups to assist comparisons i.e., stage allocation, stage allocation prior to AI model application, and use of the Demirjian method on study samples prior to applying AI models. Comparisons were difficult due to considerable variation in samples, dental features used, metrics, AI modelling, unstated regression methods, and reporting. Of the twenty-two studies included in this review, nine showed similar performance between experts manually applying Demirjian stages with/out scoring and AI models, four reported the AI models were better, two reported AI models were worse, four reported mixed results, one did not recommend either, while two did not report a comparison. Lessons learned from the traditional methods of dental age estimation have been ignored in many respects and important real-world requirements for legal and clinical applications are being overlooked.