Virtual determination of sex: metric and nonmetric traits of the adult pelvis from 3D computed tomography models.

Examination of the adult os coxae and sacrum is one of the most common methods of sex estimation from bone. Medical imaging, such as computed tomography (CT), provides the opportunity for three-dimensional (3D) imaging of the skeleton from clinical scans of known individuals in situ. In this study, a randomly selected subset of abdominopelvic CT-derived models were used to evaluate simple, repeatable metric methods of sex estimation based on a combination of obstetric measurements and the traditionally nonmetric Phenice-derived traits. A four-variable discriminant function for sex estimation was developed based on statistical analyses. Overall, the cross-validated accuracy of this method was 100%, with inter-observer error showing an average of only 2.2%. Comparative analysis was run on the data set using FORDISC 3.0. This study shows that current sex determination standards from the pelvis should be updated to include more in vivo data to increase the accuracy of identification.