Charred or fragmented, yet comparable: Quantifying dental surface dissimilarity across teeth, jaws, and heat exposure

TY - JOUR

T1 - Charred or fragmented, yet comparable

T2 - Quantifying dental surface dissimilarity across teeth, jaws, and heat exposure

AU - Kofod Petersen, Anika

AU - Manica, Scheila

AU - Forgie, Andrew

AU - Boyle, Richard

AU - Pandey Gala, Hemlata

AU - Villesen, Palle

AU - Staun Larsen, Line

N1 - © 2025 The Author(s). Published by Elsevier B.V.

PY - 2025/10

Y1 - 2025/10

N2 - Accurate dental matching is essential for forensic identification, particularly in challenging cases involving dentitions with no dental work, incomplete dentitions or damaged remains. This study evaluates similarity scoring schemes for 3D dental data using three datasets: full jaws versus single teeth (DATA-A), and two collections of heat-traumatized teeth (DATA-B and DATA-C). The similarity scores are assessed for their ability to quantify tooth curvature (dis)similarity and distinguish matching from mismatching dental comparisons. The results demonstrate the effectiveness of the methods in handling dental fragmentation (ROC-AUCDATA-A = 0.899 (95 % CI 0.840–0.948) and heat trauma (ROC-AUC DATA-B = 0.996 (95 % CI 0.98–1.00); ROC-AUC DATA-C = 0.993 (95 % CI 0.980–1.00), and that they offer a robust tool for forensic applications.

AB - Accurate dental matching is essential for forensic identification, particularly in challenging cases involving dentitions with no dental work, incomplete dentitions or damaged remains. This study evaluates similarity scoring schemes for 3D dental data using three datasets: full jaws versus single teeth (DATA-A), and two collections of heat-traumatized teeth (DATA-B and DATA-C). The similarity scores are assessed for their ability to quantify tooth curvature (dis)similarity and distinguish matching from mismatching dental comparisons. The results demonstrate the effectiveness of the methods in handling dental fragmentation (ROC-AUCDATA-A = 0.899 (95 % CI 0.840–0.948) and heat trauma (ROC-AUC DATA-B = 0.996 (95 % CI 0.98–1.00); ROC-AUC DATA-C = 0.993 (95 % CI 0.980–1.00), and that they offer a robust tool for forensic applications.

KW - Forensic odontology identification

KW - Disaster victim identification

KW - 3D dental comparison

KW - Biometric identification

U2 - 10.1016/j.forsciint.2025.112577

DO - 10.1016/j.forsciint.2025.112577

M3 - Article

C2 - 40675091

SN - 0379-0738

VL - 375

JO - Forensic Science International

JF - Forensic Science International

M1 - 112577

ER -