Sex Estimation Using Reproducible Craniometric Traits in a Modern Greek Population: A Pilot Craniofacial Study

BACKGROUND: Sex estimation is a fundamental component of skeletal identification in forensic and medicolegal contexts. Although cranial metrics are widely used, their accuracy and reproducibility are influenced by population specificity and landmark reliability. Within the modern Greek population, craniometric reference data remain limited. METHODS: This pilot study analyzed 100 dried adult crania (51 males and 49 females) from Greece using 13 craniometric measurements obtained by 2 independent raters. Measurements were recorded manually using digital calipers, with a computed tomography (CT) subset (n=30) used for intermethod comparison. A novel cranial base metric, the Prosthion-Opisthion chord (PO chord), was introduced as a reproducible alternative to conventional cranial length. Landmarks were classified using Bookstein typology, and a Composite Bookstein Classification (CBC) was applied to estimate expected measurement reliability. Sex estimation models were developed using logistic regression with internal validation via 10-fold cross-validation. RESULTS: Inter-rater reliability was high (ICC=0.864), with greater reproducibility observed for CBC Type I and II measurements. Significant sexual dimorphism was identified for the PO chord, facial height, nasal height, and foramen magnum length. The final multivariable model retained a quadratic term for the PO chord, along with cranial breadth, facial height, and nasal breadth. Cross-validated model performance was moderate, with a mean area under the receiver operating characteristic curve (AUROC) of 0.78 and an overall accuracy of 68.0%. Apparent (in-sample) accuracy was higher (77.7%), underscoring the importance of internal validation. Computed tomography-based measurements demonstrated acceptable concordance with manual measurements, particularly for CBC Type I and II traits. CONCLUSIONS: This study provides population-specific pilot data on cranial sex estimation in a modern Greek sample. The PO chord represents a reproducible alternative to conventional cranial length, and the CBC offers a transparent framework for interpreting measurement reliability. Although classification accuracy was moderate, the findings emphasize the importance of reproducibility, internal validation, and population specificity in cranial sex estimation. Larger multicenter studies incorporating expanded data sets and 3-dimensional morphometric approaches are warranted.