Nose approximation among South African groups from cone-beam computed tomography (CBCT) using a new computer-assisted...

Area of Science:

• Forensic Anthropology
• Medical Imaging
• Geometric Morphometrics

Background:

• High demand for identification of unknown remains in South Africa.
• Need for reliable facial approximation techniques tailored to the South African population.
• Existing methods may not account for specific demographic factors like ancestry, sex, and age.

Purpose of the Study:

• To develop accurate statistical models for predicting nasal soft-tissue shape from underlying skull morphology in a South African sample.
• To account for factors such as ancestry, sex, and age in facial approximation.
• To establish population-specific standards for South African facial reconstruction.

Main Methods:

• Utilized a database of 200 cone-beam computed tomography (CBCT) scans from Black and White South Africans.
• Employed an automated 3D method with dense landmarking for anatomical structure extraction.
• Applied geometric morphometrics and Projection onto Latent Structures Regression (PLSR) for statistical modeling.
• Evaluated prediction accuracy using metric deviations on training and untrained datasets.

Main Results:

• Demonstrated the influence of sex, aging, and allometry on hard- and soft-tissue variability in South African populations.
• Developed accurate statistical models incorporating ancestry, sex, and age.
• Achieved prediction errors between 1.77mm and 2.17mm for trained data and 2.14mm to 2.86mm for untrained data.

Conclusions:

• An automated 3D landmarking method is a valid prerequisite for reliable nose prediction models.
• The developed models meet population-specific standards for South Africans.
• This research enhances facial approximation techniques for forensic identification in South Africa.