Dental age assessment: orthopantomogram-based comparison of traditional methods and artificial intelligence in the adult population.

Abstract

Dental age estimation in adults represents an important component of forensic odontology and medicolegal practice. Recent artificial intelligence (AI) approaches have demonstrated improved performance compared to traditional morphometric techniques. This study aimed to compare the reliability and accuracy of three conventional dental age estimation methods with an AI-based approach. A total of 1000 digital orthopantomograms of adults aged 18-79 years with known chronological age were retrieved from the Department of Dental Anthropology database at the University of Zagreb School of Dental Medicine. Dental age was assessed using Kvaal's, Drusini's, and Cameriere's methods in ImageJ, and compared with an AI-based convolutional neural network model developed at the University of Zagreb Faculty of Electrical Engineering and Computing. Intraclass correlation coefficient (ICC) analysis and Bland-Altman plots demonstrated excellent reliability of the AI-based method (ICC = 0.965), with a mean bias of + 4.31 years and superior precision (CV = 4.34%, SEP = 3.64 years). In contrast, traditional methods showed poor reliability: Cameriere (ICC = 0.376; bias of + 9.16 years), Drusini (ICC = 0.331; bias of + 2.21 years), and Kvaal (ICC = 0.310; bias of + 15.7 years). Applicability rates were highest for the AI-based method (91.1%), followed by Cameriere (70.8%), Kvaal (63.1%), and Drusini (39.5%). Although the AI model demonstrated a systematic tendency toward age overestimation, it outperformed conventional methods and shows strong potential for application in forensic dental practice.