Abstract
Background
The human palatal rugae (folds) comprise a detailed pattern in the hard palate. They are described to be very constant during an individual’s lifetime, comparable with a fingerprint, and, importantly within a forensic setting, can serve as an important supplement in odontological identification of deceased.
Aim
The aim of this study was to test two methods of superimposition in correctly distinguishing between matches (from same individual) and mismatches (from different individuals) of digital scans of rugae.
Methods
We used an existing research database holding digital intraoral scans of 51 individuals from two points in time approximately 6 months apart. The palatal rugae area was cut manually from the digital scans without regards to teeth position. Two superimposition methods (Iterative Closest Point (ICP) and Random Sample Consensus (RANSAC)) were tested in an all-vs-all manner. Similarity was reported as fitness and inlier Root Mean Squared Error (RMSE).
Results
For both ICP and RANSAC none of the similarity measures were able to unambiguously distinguish between matches (rugae from same individual at two time points) and mismatches (rugae from different individuals).
Conclusion
The results imply that the two superimposition methods are not applicable for matching rugae scans. Further exploration of possible methods to distinguish between matches and mismatches are needed to fully exploit the great potential within forensic odontology identification of these digital palatal 3D ‘fingerprints’.
The human palatal rugae (folds) comprise a detailed pattern in the hard palate. They are described to be very constant during an individual’s lifetime, comparable with a fingerprint, and, importantly within a forensic setting, can serve as an important supplement in odontological identification of deceased.
Aim
The aim of this study was to test two methods of superimposition in correctly distinguishing between matches (from same individual) and mismatches (from different individuals) of digital scans of rugae.
Methods
We used an existing research database holding digital intraoral scans of 51 individuals from two points in time approximately 6 months apart. The palatal rugae area was cut manually from the digital scans without regards to teeth position. Two superimposition methods (Iterative Closest Point (ICP) and Random Sample Consensus (RANSAC)) were tested in an all-vs-all manner. Similarity was reported as fitness and inlier Root Mean Squared Error (RMSE).
Results
For both ICP and RANSAC none of the similarity measures were able to unambiguously distinguish between matches (rugae from same individual at two time points) and mismatches (rugae from different individuals).
Conclusion
The results imply that the two superimposition methods are not applicable for matching rugae scans. Further exploration of possible methods to distinguish between matches and mismatches are needed to fully exploit the great potential within forensic odontology identification of these digital palatal 3D ‘fingerprints’.
Originalsprog | Dansk |
---|---|
Publikationsdato | 8 nov. 2024 |
Status | Udgivet - 8 nov. 2024 |
Begivenhed | Dansk Selskab for Retsmedicins Årskursus 2024 - Centralværkstedet, Aarhus, Danmark Varighed: 8 nov. 2024 → 9 nov. 2024 |
Konference
Konference | Dansk Selskab for Retsmedicins Årskursus 2024 |
---|---|
Lokation | Centralværkstedet |
Land/Område | Danmark |
By | Aarhus |
Periode | 08/11/2024 → 09/11/2024 |
Emneord
- Identifikation
- rugae
- 3D skanninger