Related Experiment Video
Updated: May 12, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.7K
Sex Estimation Through Orbital Measurements: A Machine Learning Approach for Forensic Science
George Triantafyllou1, George G Botis1,2, Maria Piagkou1
1Department of Anatomy, School of Medicine, Faculty of Health Sciences, National and Kapodistrian University of Athens, 11 527 Goudi, Greece.
Diagnostics (Basel, Switzerland)
|January 8, 2025
Summary
Orbital measurements from skulls can aid in automated sex estimation for forensic science. This study used machine learning to analyze these measurements, achieving 68% accuracy in sex determination.
Area of Science:
- Forensic Anthropology
- Biometrics
- Machine Learning Applications
Background:
- Sex estimation is crucial in forensic science.
- The human skull offers reliable landmarks for sex determination.
- Previous studies have explored various anatomical structures for sex estimation.
Purpose of the Study:
- To develop a machine-learning classifier for sex estimation using orbital measurements.
- To evaluate the efficacy of orbital parameters in forensic identification.
- To explore automated sex estimation techniques.
Main Methods:
- Analysis of 92 dried skulls (35 male, 57 female).
- Measurement of eight orbital parameters: optic foramen height (OFH), optic foramen width (OFW), optic canal height (OCH), optic canal width (OCW), intraorbital distance (IOD), extraorbital distance (EOD), orbit height (OH), and orbit width (OW).
- Development of a Random Forest classifier for sex estimation.
Main Results:
- The Random Forest classifier achieved an overall test accuracy of 0.68.
- Key features identified: orbit width (OW), intraorbital distance (IOD), and extraorbital distance (EOD).
- The model showed a recall of 0.70 and an ROC AUC score of 0.72, indicating good discriminatory ability.
Conclusions:
- Orbital measurements show potential as reliable predictors for automated sex estimation.
- These findings can contribute to advancements in forensic identification techniques.
- Machine learning applied to craniofacial anthropometry offers a promising avenue for forensic analysis.
