Investigation of Photographic and Digital Methods for Odontological Comparison to aid with Human Identification

Abstract

Comparative dental identification may be the most efficient and cost effective means of scientific human identification methods when compared to fingerprint and DNA. Standard forensic odontology techniques for identification are usually unsuitable in cases where the ante-mortem dental records of the victim are not available or inadequate for forensic identification. Photographs of the person smiling may provide valuable information. In such cases, teeth can be used in a successful identification process if the AM photographs show visible anterior dentition that is also present in the post-mortem remains for analysis. There are no reported studies that have investigated the reliability of a superimposition method using 2D photographs of a smile and 3D dental models in dental identification. The application of 3D imaging has not been explored or utilised to its full potential in forensic dental identification. Hence, a study was designed with an aim to explore novel odontological methods by combining 2D photographs with 3D models by simulating a dental identification scenario. The objective was to see if 3D imaging in the field of forensic odontology can assist with human identification as an alternative to multiple PM photographs. The study comprised a total data of 70 digital photographs of the subjects smiling and 62 3D dental models. This study was conducted in two dental identification scenarios: Scenario A and Scenario B. Each scenario was analysed in two phases: Phase I- Visual Comparison of 2D-3D images and Phase II- 2D-3D Superimposition, both methods by the principal investigator (PI). One third of the sample was evaluated by six raters (three experienced forensic odontologists and three MSc. students). This method allowed analysis of the front teeth with emphasis on teeth alignment and morphological features. The inter-rater agreement was assessed using intra-class correlation (ICC 2, 1, absolute). The results of the study suggest that the inter-rater and intra-rater reliability using 3D superimposition was highest (ICC ≈ 1.0). In summary, this study demonstrated that dental comparison was better using 3D PM technology compared to 2D PM comparison. There was an increase in dental match rates and higher certainty among the opinions reached when using the 2D-3D superimposition method. This method also attempted to reduce the limitations of previously reported methods.

Following the results of the above study, the second study focussed on integrating 3D imaging with 2D selfies and to evaluate its feasibility in dental identification using photographs. A pilot study was designed and conducted by the PI in a similar way to the photographic superimposition study using 2D-3D method. The sample consisted of ten 3D scans obtained from the study participants (aged between 25 to 55 years) by the PI. Each participant also provided three selfie images taken from their smartphone within a year where one image with a smile was used for analysis. Two additional non-matching selfie images were included with the ten 2D selfies. The results indicated a change in conclusions by the PI. The lack of inter-rater agreement was a limitation to this study. The role of smartphones can be significant, and a selfie or photographic identification in the area of forensic odontology would be identified as an emerging alternative to conventional methods. Future studies should focus on large sample size with various selfie angulations.

The application of 3D technology in dentistry has widely expanded in recent years and several 3D imaging systems have been developed for scientific and clinical research. A review of various semi-automated 3D systems showed that the applicability of 3D superimposition and evaluation has increased. However, it appeared that most software does not have automated features which facilitate in the automated superimposition process. Hence, an experimental study was conducted to test the performance of the GOM Inspect software for automated identification using multiple 3D dental models. It was found that the software was unable to perform an automated identification with multiple 3D datasets and required manual intervention for comparison. In a forensic context, this software may be useful in single case identifications but not suitable for multiple case scenarios. There is a need for a new automated software that overcomes the limitations of this software.

There are no reported studies which presented a fully automated software that would align 3D dental models and / or scans and identify the correct match from a large dataset. This led to the design and development of a new automated software named AutoIDD which can process large datasets of 3D dental models. This software was tested using full arch post-orthodontic 3D dental models to determine the performance of the software in identifying the correct 3D dental model matching pairs which were obtained from the same patient. The testing was successful as the target scans were accurately identified from a large dataset with similar dental patterns.

The availability of AM 3D data has potential to allow for digital comparison with the PM 3D data which can be applied in dental identification cases. With the encouraging results from the software testing, a study was designed to evaluate the functionality of the AutoIDD software using 3D intra-oral scans (IOS) data from a prospective sample for validity and reliability. The total study sample consisted of 120 3D maxillary and mandibular dental data. To reconstruct a dental identification scenario, 30 maxillary and 30 mandibular dental arches scans were obtained using 3Shape TRIOS Intra-oral Scanner by the PI and were considered as IOS-AM. After one year, another set of IOS (60) were acquired from the same participants and were considered as IOS-PM.

To conclude, the AutoIDD software was able to successfully demonstrate the identification of correct matches with a match percentage that clearly differentiates the matches from non-matches. This software also enabled recognition of the changes in the human dentition, such as restorations and missing teeth. Further research and software development is required in the investigation of pre- and post-orthodontic samples, partial dental remains (jaws) and single tooth PM 3D models.

Date of Award	2020
Original language	English
Supervisor	Peter Mossey (Supervisor), Scheila Manica (Supervisor) & Nathan L. Brown (Supervisor)