Sex determination through analysis of mandibular indices using lateral cephalogram: An Artificial intelligence diagnostics

Forensic odontology involves identification and evaluation of dental evidence to determine the sex of unknown skeletal remains. This is crucial in times of mass disasters and natural calamities. Anatomical features of the skull, particularly the mandible, help in sex determination, and lateral cephalograms provide valuable insights. Artificial intelligence (AI), especially machine learning (ML) and deep learning (DL), has proven effective in medical and dental diagnostics. However, AI integration in forensic odontology is still in early stages. This study aims to assess the accuracy of sex determination using linear and angular measurements of the mandible from lateral cephalograms, by comparing machine learning algorithms. 712 lateral cephalometric radiographs were collected from our College. Cephalometric analysis was performed using Webceph software. Key measurements (ramus height, mandibular body length, articular angle, gonion angle, mandibular plane angle) were evaluated using Random Forest and XGBoost machine learning models. Both Random Forest and XGBoost demonstrated high performance in sex determination. The Random Forest model achieved an accuracy of 97.20% and precision of 97.65%, slightly outperforming XGBoost. Machine learning models are reliable and accurate for sex determination based on mandibular features. These models hold significant potential for practical applications in forensic and clinical settings.