Three-Dimensional Variation of the Human Hard Palate Across Populations: A Geometric Morphometrics Study

Background/Objectives: This study evaluated 3D palatal shape variation to distinguish population affinities, aiming to enhance forensic and anthropological assessments of unknown human remains. Methods: Maxillary dental casts (n = 100 per group) from four population groups, Australian Europeans, Malaysian Malays, Malaysian Chinese and Malaysian Indians, were digitised using a 3D scanner. Fourteen gingival and two mid-palatal landmarks and 99 surface points (semi-landmarks) were placed to capture the overall palatal shape. Shape data were aligned using Generalised Procrustes Analysis (GPA) followed by Principal Component Analysis (PCA) to explore shape variation. Canonical Variate Analysis (CVA) and Mahalanobis distances were used to assess group differences. Classification performance was evaluated using accuracy, sensitivity, specificity, precision, and F1 score. Results: The first five principal components (each explaining ≥5% of the observed variation) were retained for CVA. CV1 accounted for 88% of the between-group variance, showing clear separation of Europeans from the clustered Malay-Chinese group, with Indians appearing intermediate. European palates were longer, narrower, deeper, and more anteriorly tapered. Significant shape differences were found between all groups, except between Malays and Chinese. The overall classification accuracy was 45%, with better performance for European and Indian groups. Specificity was higher across all groups (0.70-0.90), while sensitivity, precision, and F1 scores were moderate. Conclusions: Geometric morphometric analysis revealed population-level differences in palatal shape, with moderate accuracy and high specificity. These findings support its role as a complementary tool alongside other data in forensic and anthropological applications. While palatal shape alone cannot definitively classify populations, it may be useful for excluding certain groups.