Gender Prediction from Angular and Linear Parameters in Cranium Lateral View by using Machine Learning Algorithms: A Computed Tomography Study

The purpose of the study was to predict a gender by using Machine Learning Algorithms (MLA) with variables of the lateral view of the cranium from Computed Tomography (CT) images. A total of 5 parameters (3 linear and 2 angular) of the lateral view of the cranium were evaluated on CT images of 200 female and 200 male adult individuals in the present study. These parameter measurements were analyzed with MLA and Logistic Regression (LR), Random Forest (RF), Linear Discriminant Analysis (LDA), K-Nearest Neighborhood (KNN) and Naive Bayes (NB) models were used. Accuracy (Acc), Sensitivity (Sen), Specificity (Spe) and F1 scores (F1) were used as the evaluation criteria in the study. As a result of MLA, the Acc ratio was found to be 0.77 for the KNN algorithm, 0.84 in the NB algorithm, 0.85 in the LDA algorithm, 0.70 in the RF algorithm and 0.81 in the LR algorithm. As a result of the analysis, 0.85 Acc, 0.85 Sen, Spe 0.85 and 0.85 F1 values were found in the LDA algorithm with the highest accuracy. When the significance level of the variables in the study was examined, it was found that variable A had the best effect. It was found that the MLA used for the variables of the lateral view of the cranium yielded high accuracy regarding gender and the LDA Model was effective in predicting gender.