Automatic variable extraction from 3D coxal bone models for sex estimation using the DSP2 method

Area of Science:

• Forensic Anthropology
• Bioanthropology
• Computational Anatomy

Background:

• Sex estimation is crucial for biological profiling in forensic anthropology.
• Traditional methods rely on manual measurements, which can be subjective and time-consuming.
• Automated approaches offer potential for increased objectivity and efficiency.

Purpose of the Study:

• To apply an existing algorithm for automatic extraction of 10 variables for the DSP2 sex estimation method.
• To evaluate the robustness of this automated approach on a heterogeneous population sample.

Main Methods:

• Utilized 3D scans of pelvic bones (n=240 for initial validation, n=108 for robustness testing).
• Employed an algorithm (Kuchař et al. 2021) for automated variable extraction.
• Compared automated measurements with manual measurements and assessed sex estimation accuracy.

Main Results:

• High agreement between automated and manual measurements (rTEM < 5% for most dimensions).
• Excellent accuracy for sex estimation using all 10 variables (0.3% error rate).
• Increased undetermined cases noted in Portuguese (25% males) and New Mexican (36.5% females) samples.

Conclusions:

• The automated dimension extraction procedure is effective on different data types and diverse populations.
• Automation significantly accelerates specialist work and reduces subjectivity in sex estimation.
• Further research may be needed to address undetermined cases in specific demographic groups.