Evaluation of 3D palatal surface registration for human identification: The impact of developmental dentition stages

Abstract Background Palatal morphology shows considerable inter-individual variability; however, its utility in forensic identification remains uncertain. It is also unclear whether developmental changes in palatal shape affect matching performance when comparing against earlier antemortem records from primary and mixed dentitions. Our study aimed to evaluate the discriminative potential of 3D palatal surface scans for antemortem-postmortem (AM-PM) matching across developmental stages. Methods Maxillary dental casts from a longitudinal twin cohort ( n = 81 pairs) were digitised to generate palate-only models simulating PM datasets, while full scans served as AM references across permanent, mixed, and primary dentitions. Comparison was performed using a two-stage registration framework incorporating feature-based RANSAC (Random Sample Consensus) and geometry-based Iterative Closest Point (ICP) algorithms. The matching performance was assessed using global alignment score (RANSAC inlier ratio), fitness (final correspondence ratio after refinement), root mean square error (RMSE; mean surface distance after alignment), and rank-based identification analysis. Results True matches demonstrated significantly higher fitness values than true mismatches across all comparisons, whereas RMSE was significantly lower only in permanent-to-permanent comparisons. Identification performance decreased with increasing developmental differences, with the lowest accuracy in permanent-to-primary comparisons. Fitness demonstrated greater robustness to cross-stage morphological variation than RMSE. Conclusion 3D palatal surface scans show discriminatory potential for AM-PM matching, but performance reduced across developmental transitions. Palatal comparison could still potentially act as a supplementary tool for screening and candidate prioritisation in forensic identification.