Gülçin Ray; Abdullah Ray; Canan Akünal & Ibrahim Kürtül

This study focuses on sex prediction by machine learning (ML) algorithms using brainstem data obtained from brain magnetic resonance images (MRIs) of a healthy adult Turkish population. In the study, midbrain thickness (TMT), pons thickness (TP), medulla thickness (TM), spinal cord thickness (TSC), mamillopontine distance (MPD), midbrain height (MBH), pontomesencephalic angle (PMA), tectum and tegmentum sagittal diameter, (TTSD) were measured from brain MR images of 200 women and 200 men, tectum and tegmentum transverse diameter (TTTD), right pedunculus cerebri sagittal diameter (RPCSD), left pedunculus cerebri sagittal diameter (LPCSD), right pedunculus cerebri transverse diameter (RPCTD), left pedunculus cerebri transverse diameter (LPCTD) variables were measured. Logistic Regression Classifier (LRC), Gradient Boosting Classifier (GBC), Decision Tree Classifier (DTC), K-Nearest Neighbors (KNN), Linear Discriminant Analysis (LDA), Random Forest Classifier (RFC), Naive Bayes Classifiers (NBC), XGBoost (XGB) ML algorithms were used. RStudio was used for statistical analysis. The RFC algorithm showed the highest ACC ratio of 0.79 and F1 score of 0.79. According to shap analysis, MPD made the greatest contribution to the performance of the classification models. The variables with statistically significant differences between sexes were TM, TSC, TTSD, RPCSD and RPCTD. The study shows that ML algorithms can predict sex with 79 % accuracy out of brainstem data. This result clearly shows the effect of sex on brainstem morphometry.

RAY, G.; RAY, A.; AKÜNAL, C. & KÜRTÜL, I. Sex prediction with machine learning algorithms modeling on brainstem morphometric features: An MRI study. Int. J. Morphol., 44(3):1062-1070, 2026.