CBCT assessment of maxillary sinus sexual dimorphism: morphometry and sex-classification performance in Peruvian adults

Abstract

The maxillary sinus exhibits sexual variation that is useful for forensic identification. Cone-beam computed tomography (CBCT) enables precise quantification of linear and volumetric dimensions; however, evidence regarding the most discriminative parameter remains heterogeneous. This study aimed to determine whether CBCT-derived morphometry of the maxillary sinuses discriminates biological sex in adults and to quantify classification performance. We conducted a retrospective observational study and randomly selected 108 CBCT scans from 150 eligible records (age 20–70 years, permanent dentition). Height, width, length, and volume were measured bilaterally from standardized multiplanar reconstructions; normality was assessed and sexes were compared using Student’s t test. Males showed larger sinus volumes than females (right: 17.41 ± 1.61 vs 15.38 ± 1.79 cm³; left: 17.46 ± 1.37 vs 15.86 ± 2.02 cm³; p < 0.001). In side-specific univariate linear discriminant analysis with 10-fold cross-validation, width was the best single predictor: right width AUC = 0.80 (95% CI: 0.72–0.88) and accuracy = 0.72 (95% CI: 0.63–0.79); left width AUC = 0.74 (0.64–0.83) and accuracy = 0.68 (0.59–0.76). In multivariable models using width + age, linear discriminant analysis achieved AUC = 0.771 (95% CI: 0.676–0.858) and accuracy = 0.722 (95% CI: 0.631–0.798), with sensitivity = 0.729 (95% CI: 0.590–0.834) and specificity = 0.717 (95% CI: 0.592–0.815); logistic regression performed similarly (AUC = 0.771; accuracy = 0.713). Conclusions: CBCT-based morphometry of the maxillary sinus discriminates sex with moderate performance; width is the most informative single metric in this cohort. For forensic application, population-specific external validation and multivariable models integrating shape and volume descriptors are recommended.

Keywords: Maxillary Sinus, Sex Determination, Cone-Beam Computed Tomography, Forensic Anthropology, Discriminant Analysis