Back

Curvature-based 3D dental comparison to identify trauma-induced surface changes in human teeth: A forensic comparison study

Kofod Petersen, A.; Spin-Neto, R.; Villesen, P.; Staun Larsen, L.

2025-06-20 forensic medicine
10.1101/2025.06.19.25329914 medRxiv
Show abstract

Background/AimThis study evaluates the performance of a curvature-based 3D dental comparison method - the keypoint pipeline - for forensic identification, assessing the effects of standardised blunt force trauma on human dentitions. Material and MethodsThe dental arches in ten human jaw specimens (five maxillae, five mandibulae) were scanned using two intraoral 3D scanners before and after exposure to controlled blunt trauma delivered via a drop tower mechanism applying approximately 3154 Newton of force. Trauma outcomes were documented through high-speed video, digital photography, and 3D scanning. Post-trauma scans were processed using the keypoint pipeline, which quantifies dental surface similarity by comparing curvature signatures. An all-vs-all comparison was conducted between pre- and post-trauma scans, including cross-scanner evaluations. ResultsDespite consistent trauma application, fracture patterns varied by jaw type, with mandibular fractures typically occurring in the frontal plane in the side segments and maxillary fractures in the sagittal plane in the midline suture. The keypoint pipeline successfully scored 92.5% of the true matches to be the best matching comparison, even in the presence of significant structural damage and tooth displacement. Matching pairs yielded lower dissimilarity scores (mean: 0.55) compared to mismatches (mean: 0.78), indicating that curvature features were sufficiently preserved post-trauma. ConclusionsThese findings support the integration of curvature-based 3D dental surface analysis into forensic odontology workflows, particularly in disaster victim identification scenarios involving blunt force trauma.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.