ABSTRACT
The mastoid bone, situated at the base of the skull and characterized by its compact structure in the petrous portion, being among the slowest-growing bones, has gained recognition as a valuable anthropological tool for sex determination. Thus, we have proposed to assess the reliability of the mastoid process in sex determination in a Tunisian population using CT-scan analysis. A cross-sectional study was conducted. CT scans forming the mastoid triangle were analyzed using a General Electric Bright Speed scanner. Nine measurements were taken by a single observer, ensuring reliability through intra- and inter-observer assessments. Normalization and statistical analyses, including logistic regression, were applied to identify sex-discriminating variables. The model's performance was evaluated using learning curves, cross-validation, and various metrics. The resulting logistic regression equation, coefficients, and intercept provided a predictive tool for sex determination. A total of 256 cranial CT scans (126 males, 130 females) were analyzed. Our study revealed that the mastoid region approach achieved an overall accuracy of 80.8 % in sex identification within the examined population. The method demonstrated a sensitivity of 78.9 % and specificity of 81.8 %. All investigated variables (AP, PM, AM, CMH, TMH, OSDmax, OCDmax, MA) exhibited discriminatory capabilities for sex determination, except for AIA. Notably, the most effective discriminators were AP, CMH, and OSDmax. Utilizing an ROC curve analysis to optimize mastoid variables for maximum sensitivity and specificity, we obtained excellent results, with an area under the curve reaching 91 %.
