Abstract Background: Establishing the biological sex of individuals is a key component of forensic identification, especially in the analysis of degraded or partial remains. Cone beam computed tomography (CBCT) enables precise, non-invasive for evaluating craniofacial structures. Objective: To assess the effectiveness of maxillary and frontal sinus volumes, along with facial soft tissue thickness (FSTT), obtained from CBCT scans, for estimating sex in an Indian population using statistical and machine learning techniques. Methodology: A dataset comprising 450 CBCT scans was evaluated. FSTT was recorded at 28 standardised craniofacial points following the Steyn and Cavanagh protocol. Volumetric assessment of the maxillary and frontal sinuses was measured. Discriminant function analysis (DFA) and two artificial neural network (ANN) models were used to classify sex based on the collected data. Results: Both sinus volumes and FSTT showed statistically significant differences between males and females ( P < 0.005). DFA achieved an overall accuracy of 80%, correctly identifying 73% of males and 87% of females. Cross-validation resulted in 75.6% accuracy. FSTT varied significantly between sexes at all landmarks. ANN Model 1 achieved 72% accuracy, while Model 2 exhibited potential accuracy up to 99%. Conclusion: Morphometric data obtained from CBCT scans—such as sinus volumes and FSST—serve as reliable indicators for estimating sex. When combined with artificial intelligence models, these measurements provide an effective and practical approach to forensic identification within the Indian population.
Jayakrishnan et al. (2025) studied this question.