Gender Estimation From Dimensions and Morphologic Types of the Crista Galli in the Pediatric Population Using Regression Tree Model

The present paper aimed to evaluate the morphometric and morphologic features of the crista galli (CG) using computed tomography (CT) images and to investigate the potential role of CG dimensions and morphologic types for estimating gender in the pediatric population. CT images of 523 individuals under the age of 18 (270 males, 253 females) were evaluated. The dimensions of the CG were measured. Based on CG dimensions, 3 different morphologic types were identified. The predictive power of CG dimensions and types in determining gender was evaluated using multivariate binary logistic regression analysis, regression decision tree modeling, and the χ2 test. Males had a higher, wider, and significantly longer CG compared with females (p=0.317, 0.217, and <0.001, respectively). CG length predicted sex with a total accuracy rate of 60.8% and demonstrated a more successful classification rate for females (62.8%) compared with males (54.1%). A child with a CG length value ≥11.95 mm was 1.93-fold more likely to be male than a child with a CG length value <11.95 mm. Using the regression tree model, gender was estimated with an overall accuracy of 69.6% based on the cutoff values obtained for CG dimensions. CG exhibited sexual dimorphism regarding its length in children. The length, height, and width of the CG can be used to determine gender in the pediatric population from CT images using the regression tree model with a relatively sufficient accuracy rate.