Introduction: The purpose of this study was to analyze the respective contribution of the skull and mandible to sex estimation in an entire cranium using metric and geometric morphometric methods and multislice computed tomography (MSCT) analysis.
Materials: The study sample comprised 120 individuals aged 23 to 84 years and divided into three groups (whole sample and over or under 40 years of age).
Methods: Forty-eight osteometric landmarks were positioned using Osirix®, 34 for the cranium and 14 for the mandible. The data were analyzed using univariate analyses and logistic regression using backward stepwise selection with cross-validation of the classification results. Generalized procrustes analysis (GPA) was used, and Goodall's F test and Mahalanobis D2 matrices allowed an assessment of statistical significance.
Results: The classification accuracy of cranium models ranged from 87% to 88.3%, and from 68% to 81.4% for the mandibular models. With geometric morphometry, accuracy was of 94-100% for the cranium and 84.2% for the mandible.
Discussion: We conclude that the mandible had a higher accuracy rate for individuals over 40 years than individuals under 40 years of age, although the accuracy of geometric morphometry did not differ significantly between the age groups. The cranium, however, presented greater predictability for all the parameters assessed.
Keywords: Forensic anthropology; Forensic science; Geometric morphometry; Mandible; Sex estimation; Skull.