Three-Dimensional Analysis of Orbital Sexual Dimorphism in a Romanian Sample: Implications for Sex Estimation and Forensic Facial Reconstruction
Year: 2026
Type: article
Study design: Observational Study
Abstract: Introduction The orbit exhibits sexual dimorphism and may provide useful information for sex estimation in forensic anthropology. However, data comparing orbital measurements with geometric morphometrics are limited for the contemporary Romanian population. This study evaluated the sexual dimorphism of the orbital region using both approaches and assessed their performance for sex estimation, with potential indirect relevance to forensic facial reconstruction. Methods This study included 163 CT head scans from contemporary Romanian adults, including 78 (47.9%) males and 85 (52.1%) females. Ten anatomical landmarks were digitized in three-dimensional space, and their Cartesian coordinates were used for both conventional linear measurements and geometric morphometric analysis. Six conventional orbital measurements were evaluated from three-dimensional anatomical landmarks using univariate and multivariate discriminant function analyses with leave-one-out cross-validation (LOOCV). Geometric morphometric analysis was performed using generalized Procrustes analysis and principal component analysis. Sexual dimorphism in centroid size and Procrustes shape was assessed, and allometry was investigated through multivariate regression of shape on log centroid size, including pooled within-sex and sex-specific analyses. Discriminant analysis was also performed before and after allometric correction. Results Orbital breadth, interorbital breadth, and biorbital breadth were significantly greater in males, whereas orbital height showed limited sexual dimorphism. Biorbital breadth was the most accurate individual conventional measurement, correctly classifying 119/163 (73.0%) individuals after cross-validation, whereas the multivariate linear model correctly classified 114/163 (69.9%) individuals. Significant sex-related differences were also identified in orbital shape and centroid size. Uncorrected Procrustes shape correctly classified 116/163 individuals (71.2%) after cross-validation, increasing to 121/163 (74.2%) after correction for allometry. Centroid size alone correctly classified 119/163 (73.0%) individuals, while the combined uncorrected shape and centroid size model achieved the highest cross-validated accuracy, correctly classifying 125/163 (76.7%) individuals, including 67/85 (78.8%) females and 58/78 (74.4%) males. Conclusion Conventional linear measurements showed lower cross-validated classification accuracy than the combined centroid size and shape model. Integrating three-dimensional orbital size and shape may therefore provide complementary information that may improve sex estimation with potential indirect relevance to forensic facial reconstruction.
Source: Cureus
Institution: Carol Davila University of Medicine and Pharmacy
Language: English
Subfield: Archeology (arts and humanities)
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W7220941396