Analyzing 3D facial morphology: Insights from a comparative European and South African study on population affinity, sex, age, and allometry

Variable growth patterns and multifactorial mechanisms cause variation in facial shape. These differences in facial morphology pose challenges for craniofacial reconstruction. Three-dimensional (3D) imaging modalities are a valuable resource for examining these variations. In this study, we used geometric morphometric methods to evaluate the effects of population affinity, sex, age, and allometry on the variation and covariation of hard and soft tissue facial morphology matrices in a sample of French and white South African individuals. Seventy-six and 108 cone-beam computed tomography scans of white South African and French nationals, respectively, were retrospectively acquired. Three-dimensional anatomical structures (hard and soft tissue matrices) were extracted using MeVisLab© v. 2.7.1 software for dense landmarking of 43 craniometric, 50 capulometric, and 559 sliding landmarks. Geometric morphometric analyses were used to quantify shape differences attributed to population affinity, sex, age, and allometry and assess the covariation between hard tissue structures and soft facial tissues. Hard and soft tissue facial matrices were influenced by population differences, sexual dimorphism, and aging. Compared to sex and age, population affinity had the strongest influence on variation. In French individuals, all hard and soft tissue matrices were sexually dimorphic, except for the eyes and left external auditory meatus (EAM). In white South Africans, sexual dimorphism was observed for the mouth, midface, and left EAM. Significant shape differences were also observed for specific age groups. The underlying skull and overlying soft tissues were strongly associated with the nose and anterior nasal aperture (correlation, r²-PLS = 0.976), followed by the right ear and right EAM (r²-PLS = 0.875) and the left ear and left EAM (r²-PLS = 0.871) in white South Africans. For French individuals, relatively weak to moderate correlations were observed, and the covariation between matrices was nonsignificant, except for the association between the right ear and right EAM (r²-PLS = 0.499). The smallest covariation was observed between the mouth and midfacial matrix in both populations (South African: r²-PLS = 0.464; French: r²-PLS = 0.367), which was also nonsignificant. This study revealed that 3D imaging technology and geometric morphometric methods can accurately quantify and visualize facial morphology differences. These methods can also evaluate the association between skull structure and soft facial features.