Enhancing Sex Estimation Accuracy with Cranial Angle Measurements and Machine Learning.

Area of Science:

• Forensic anthropology
• Craniometry
• Biometrics

Background:

• Current sex estimation methods often rely on linear measurements, potentially overlooking valuable discriminatory information present in angles.
• Cranial angles have been largely underutilized in sex estimation, despite their potential to reveal significant dimorphic differences.

Purpose of the Study:

• To evaluate the efficacy of cranial angles in sex estimation.
• To identify the most sexually dimorphic cranial angles using machine learning algorithms.
• To assess the potential for enhancing sex estimation accuracy through the integration of cranial angle data.

Main Methods:

• Utilized computed tomography (CT) images from 154 males and 180 females to extract 36 cranial angles.
• Trained machine learning models, including support vector machines, Naïve Bayes, logistic regression, and CN2, for classification.
• Employed an attribute selection scheme to identify optimal subsets of cranial angles for model training.

Main Results:

• Machine learning algorithms achieved high accuracy when trained on specific subsets of cranial angles.
• Angles associated with the lower forehead and upper midface were consistently identified in the best-performing models.
• The study demonstrated significant classification potential inherent in cranial angle measurements.

Conclusions:

• Cranial angles are valuable and underappreciated indicators for sex estimation.
• Incorporating cranial angles into analytical models can substantially improve the accuracy of sex determination.
• This research validates the use of cranial angles as a complementary tool in forensic sex assessment.