A computed tomography (CT) study of the eyeball position and estimation models for craniofacial identification.

Correctly setting the eyes within the orbits is an important first step in craniofacial identification. As several estimators often exist for each eyeball attribute, clarification of their accuracies in out-of-group samples is critical. In this study, we use CT scans from 58 individuals to assess standard errors (SEE) for published estimators, cross-validate methods and examine the linear regression utility of orbit volume for estimating eyeball protrusion. Our results confirm the eyeball to be positioned closer to the superior and lateral orbital walls (2.5 & 5.6 mm, respectively). Horizontal placement was best estimated by the pupil center at 42.4% of the orbit width and medial to the lateral orbital margin (SEE = 1.6 mm). Anterior eyeball position was best set by the grand mean exophthalmos (16.2 mm) measured in living subjects (SEE = 2.9 mm). Vertical eyeball height could be estimated, with equivalent error, by either: (a) setting the pupil center 22.9 mm inferior to the superior orbital margin (SEE = 1.7 mm); or (b) inferior to the superior orbital margin by 44.1% of the orbit height (SEE = 1.6 mm). Orbit volume weakly correlated with anterior eyeball protrusion (r = 0.33), but the relationship was not statistically significant (p = 0.14). The correlation direction contradicted Gerasimov’s reports (positive, not negative). Linear regression using the volume of an elliptical cone, calculated from three orbit-associated measurements, reduced protrusion SEE to 2.4 mm. Near equivalent accuracy by mean exophthalmos and its far greater simplicity, however, awards the mean value more practicality for casework.