Usability of quantitative atlas measurements from computed tomography images for sex estimation: A machine learning approach.

Sex estimation plays a critical role in forensic identification, missing person identification, and forensic investigations. This study aimed to evaluate the usability of quantitative metric measurements obtained from computed tomography (CT) images of the atlas (the first cervical vertebra) for sex identification using machine learning algorithms. The study used CT images from 200 individuals (100 males and 100 females). Eighteen metric parameters of the atlas-comprising 16 lengths and 2 angles-were measured. These parameters were analyzed using 13 different machine learning algorithms. Basic statistical methods were used to compare the effect of each metric parameter on sex estimation. The machine learning models predicted sex with an accuracy ranging from 86% to 89%. The highest accuracy (89%) was achieved by the Gaussian Naive Bayes algorithm using only five selected metric parameters. Additionally, 15 out of the 18 measured parameters showed statistically significant differences between sexes. This study demonstrates that sex can be estimated with high accuracy using only quantitative metric measurements data from the atlas vertebra and machine learning algorithms. Notably, this approach may be especially valuable when only the atlas is available, providing essential preliminary data for forensic medical examinations.