Integrating deep learning techniques for analysis of chin morphology among Han Chinese individuals using a large cone-beam computed tomography dataset.
Authors
Affiliations (8)
Affiliations (8)
- Peking University School and Hospital of Stomatology, Beijing, 100191, China.
- Department of Oral and Maxillofacial-Head Neck Oncology, College of Stomatology, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, 200011, China.
- Department of Orthodontics, National Engineering Laboratory for Digital and Material Technology of Stomatology, Beijing Key Laboratory of Digital Stomatology, Peking University School and Hospital of Stomatology, Beijing, 100191, China.
- Key Laboratory of Machine Perception (MOE), Department of Machine Intelligence, School of Intelligence Science and Technology, Peking University, Beijing, 100871, China.
- Shanghai Linkedsmile Orthodontic Information Technology Co., Ltd, Shanghai, 201203, China.
- Department of Orthodontics, National Engineering Laboratory for Digital and Material Technology of Stomatology, Beijing Key Laboratory of Digital Stomatology, Peking University School and Hospital of Stomatology, Beijing, 100191, China. [email protected].
- The First Clinical Division, National Engineering Laboratory for Digital and Material Technology of Stomatology, Beijing Key Laboratory of Digital Stomatology, Peking University School and Hospital of Stomatology, Beijing, China. [email protected].
- Department of Orthodontics, National Engineering Laboratory for Digital and Material Technology of Stomatology, Beijing Key Laboratory of Digital Stomatology, Peking University School and Hospital of Stomatology, Beijing, 100191, China. [email protected].
Abstract
To characterize chin morphology and investigate its associations with sex, as well as various sagittal and vertical skeletal patterns. A total of 743 cone-beam computed tomography (CBCT) images (322 males, 421 females; aged 18-83 years) were analyzed to measure chin height, thickness, and width. An nnU-Net model was trained for automated chin segmentation using a subset of 304 images. The dataset was stratified into 18 categories based on combinations of sex and skeletal patterns (sagittal and vertical). Aligned average chin models were then constructed to facilitate morphological comparisons across these groups. Morphometric analysis revealed that males possessed significantly greater chin height (22.54 ± 3.06 mm vs. 20.87 ± 2.45 mm, p < 0.01) and width (51.25 ± 3.36 mm vs. 50.15 ± 2.87 mm, p < 0.01) than females, whereas no significant difference was observed in thickness. Regarding sagittal patterns, chin thickness emerged as the primary distinguishing feature; Class II individuals exhibited significantly thinner chins (4.74 ± 1.02 mm) compared to both Class I (5.16 ± 1.04 mm) and Class III (5.18 ± 0.98 mm) individuals (p < 0.01), a trend consistent across both sexes. In terms of vertical skeletal patterns, low-angle individuals were characterized by significantly greater chin thickness (5.26 ± 1.10 mm, p < 0.001) and width (51.40 ± 2.99 mm, p < 0.01), but reduced height (21.04 ± 2.74 mm, p < 0.001) relative to high-angle and average-angle groups. Comparative analysis of 3D average chin models corroborated these morphometric findings, providing visual evidence of distinct shape variations. This study systematically characterized chin morphology variations in relation to sex and skeletal patterns, establishing comprehensive average models specifically for the Han Chinese population. These findings provide a high-fidelity anatomical reference that can enhance diagnostic precision and treatment planning in orthodontics and orthognathic surgery. Chin prominence is a decisive factor in orthodontic treatment planning and facial profiling. Furthermore, chin thickness is closely correlated with the alveolar bone volume of the mandibular anterior teeth, acting as a primary anatomical constraint for tooth movement. This morphological insight is particularly critical for planning safe and stable orthodontic outcomes in patients with skeletal Class II or high-angle patterns, where bone support may be limited.