Evaluation of the suitability of cranial measurements obtained from surface-rendered CT scans of living people for estimating sex and ancestry

Abstract

Craniometric data from computed tomography (CT) head scans of 287 living Americans of three descent groups (African, Asian, European) and both sexes were analyzed for measurement precision. Classification accuracy was estimated by a leave-one-out cross validation, and group means were compared to the Forensic Data Bank (FDB). Landmarks were placed on 3D surface models of the skulls to approximate traditional cranial measurements utilized in sex and ancestry estimations. From repeat measurements by one observer on a subset of skulls (n = 14) reflecting 14 different CT protocols, the least precise landmark was euryon (SD ≤ 4.09 mm) and the least precise distances according to the coefficient of reliability (< 0.95) were orbital breadth, nasal height, and frontal and parietal chords. CT group means were generally larger than FDB means. However, successful classifications of FDB group means into CT groups and vice versa (except for Asians) for both sex and ancestry suggest that cranial measurements taken from clinical CT scans from living individuals are comparable to traditional cranial measurements from individuals in osteological databases. Certain cranial measurements need further examination to determine how best to collect data from CT scans in a way that is most compatible with existing dry skull measurements.