Work

A computational framework for 3D statistical shape modeling of adult Chinese, Indian, and Malay craniofacial structures from CT imaging

Year: 2026

Type: article

Abstract: Background and objective Three-dimensional statistical shape modeling provides a computational method to quantify anatomical variation from medical imaging. This study aimed to construct statistical shape models of adult Asian craniofacial structures using computed tomography data, with the goal of identifying sex-, ethnicity-, and age-related morphological differences and establishing population-specific reference models. Methods A total of 300 high-resolution retrospective CT scans from healthy adults representing Chinese (n = 95), Indian (n = 100), and Malay (n = 105) groups were analyzed. Anatomical segmentation, landmark-guided rigid alignment, and Directed Distance Field-based Internal Structure Modeling (DDF-ISM) were used to generate dense-correspondence meshes for the head surface and craniofacial skeleton (skull and mandible). Principal component analysis was applied to quantify dominant modes of variation, and group differences were evaluated with ANOVA/MANOVA and localized difference mapping. Results Sex showed the largest overall effect size, with the greatest separation observed along PC1 for both the head surface and skeleton. Ethnic background most strongly influenced the nasal bridge, forehead, temporal bone, and alveolar region, while age-related changes were concentrated in vertex height, forehead curvature, nasal projection, and dentoalveolar remodeling. Difference mapping enabled high-resolution visualization of these localized variations across external soft tissue and the underlying bone framework. Conclusions The proposed CT-based computational pipeline demonstrates the feasibility of integrating statistical shape models and statistical analysis to generate population-specific craniofacial references. These models hold potential applications in clinical assessment, surgical planning and simulation, and forensic identification. Distinct from previous surface-only, bone-only, or prediction-oriented studies, the framework characterizes matched external and skeletal morphology across sex, three ethnic groups, and age within the same adult CT sample.

Source: Biomedical Signal Processing and Control

DOI: 10.1016/j.bspc.2026.111501

Authors: Zhuoman Liu, Yan Luximon, Wei Lin Ng, Eric Chung, Pengling Ren

Institutions: Hong Kong Polytechnic University, University Malaya Medical Centre, University of Malaya, Capital Medical University, Beijing Friendship Hospital

Language: English

Topic: Morphological variations and asymmetry

Subfield: Geometry and Topology

Field: Mathematics

Domain: Physical Sciences

SDG: Good health and well-being

Open Access status: hybrid

Estimated APC paid: $2,590