Artificial Intelligence Algorithms For Disaster Victim Identification Using Dental Evidence

This study aims to develop a convolutional neural network (CNN) algorithm to enhance the efficiency of forensic odontology in medicolegal identification by comparing antemortem (AM) and postmortem (PM) dental evidence in orthopantomograms (OPGs). The goal is to optimize imaging analysis for the identification of unknown cadavers, especially in mass disaster contexts. This study involved a systematic sampling of patients aged 16 to 30 from the Stomatology Department at the Local Health Unit Santa Maria, with 1235 orthopantomograms (OPGs) analyzed. The study received ethical approval (reference number 271/23). Two convolutional neural network (CNN) algorithms were developed: one for age classification and another for positive identification. The age classification model utilized 1200 OPGs, split into two age groups, while the positive identification model compared pairs of OPGs using a Siamese network architecture. Data preprocessing included resizing, normalization, and augmentation. Performance was evaluated with metrics such as accuracy, precision, recall, and F1 Score, using confusion matrices and heatmaps to visualize decision processes. Statistical analysis was conducted in IBM SPSS® Statistics 29. Both developed models achieved accuracy of 85%, reflecting high overall performance. The age classification model performed better at classifying OPGs from aged between 16 and 23 years, while the positive identification model was significantly better at identifying pairs of OPGs from different individuals. The developed AI model enhances medicolegal identification of unknown cadavers and is particularly effective in mass disaster contexts. By comparing antemortem and postmortem dental evidence in orthopantomograms, it improves the efficiency of forensic odontology.