Sex estimation based on frontal sinus computed tomography images using machine learning and artificial neural networks
Küçük Resim Yok
Tarih
2025
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Taylor & Francis Ltd
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
Due to its anatomical uniqueness, the frontal sinus (FS) shows significant inter-individual differences by ancestry, age, and sex, making it useful for preliminary identification processes. This study aims to estimate sex using machine learning (ML) algorithms and artificial neural networks (ANN) applied to morphometric data from FS computed tomography (CT) images. This retrospective study analysed CT scans of 338 females and 338 males aged 18-65. FS measurements comprised sinus floor anteroposterior length, volume, area, height, depth, width, and anterior wall thickness (AWT). Sex estimation was performed using several ML algorithms, including Linear Discriminant Analysis, Quadratic Discriminant Analysis, Logistic Regression, Extra Trees Classifier, Decision Tree, Random Forest, k-Nearest Neighbours, and Gaussian Naive Bayes. Additionally, a multilayer perceptron classifier, representing ANN models, was utilized. The highest classification accuracy (94%) was achieved by the Logistic Regression model. According to the SHapley Additive exPlanations analysis, the two most influential parameters were identified as the right and left AWT, respectively. This study, with a comparatively large sample size, found that all morphometric FS parameters - especially AWT - hold significant potential in forensic identification. ML- and ANN-based models showed high classification accuracy, surpassing previous studies. These findings may guide future research involving diverse populations and regions.
Açıklama
Anahtar Kelimeler
Frontal sinus, artificial neural networks, machine learning, sex estimation, paranasal sinuses
Kaynak
Australian Journal of Forensic Sciences
WoS Q Değeri
Q4
Scopus Q Değeri
Q2
