Development and validation of an automated algorithm for palatal rugae matching in forensic identification

Area of Science:

• Forensic Science
• Biometrics
• Digital Imaging

Background:

• Traditional forensic identification methods (DNA, fingerprints, dental records) are often unavailable for degraded remains.
• Palatal rugae present a unique and stable intraoral anatomical feature for identification.

Purpose of the Study:

• To investigate palatal rugae as an alternative intraoral marker for forensic identification.
• To evaluate a fully automated digital matching process for palatal rugae analysis.
• To compare the accuracy of automated versus semi-automated identification methods.

Main Methods:

• Acquired 3D palatal scans from 345 participants using an intraoral scanner.
• Applied a four-step automated matching process: data preparation, rough alignment (Fast Point Feature Histograms), fine alignment (Iterative Closest Point), and similarity scoring.
• Validated the algorithm on development and unseen datasets, analyzing accuracy and sampling effects.

Main Results:

• The automated system demonstrated 100% accuracy in identifying identical palatal rugae scans and rejecting non-matching pairs.
• Robustness to initial alignment was confirmed through random rotation tests.
• Analysis indicated that 3,000 points are sufficient for perfect accuracy, reducing computational load.

Conclusions:

• Automated 3D alignment of palatal rugae provides highly accurate, reproducible, and objective forensic identification.
• This digital method surpasses semi-automated approaches and can expedite identification in mass casualty incidents.
• Further validation is needed, but the findings support integrating digital palatal data into forensic identification protocols.