Applications of contemporary artificial intelligence technology in forensic odontology as primary forensic identifier: A scoping review

Background

Forensic odontology may require a visual or clinical method during identification. Sometimes it may require forensic experts to refer to the existing technique to identify individuals, for example, by using the atlas to estimate the dental age. However, the existing technology can be a complicated procedure for a large-scale incident requiring a more significant number of forensic identifications, particularly during mass disasters. This has driven many experts to perform automation in their current practice to improve efficiency.

Objective

This article aims to evaluate current artificial intelligence applications and discuss their performance concerning the algorithm architecture used in forensic odontology.

Methods

This study summarizes the findings of 28 research papers published between 2010 and June 2022 using the Arksey and O'Malley framework, updated by the Joanna Briggs Institute Framework for Scoping Reviews methodology, highlighting the research trend of artificial intelligence technology in forensic odontology. In addition, a literature search was conducted on Web of Science (WoS), Scopus, Google Scholar, and PubMed, and the results were evaluated based on their content and significance.

Results

The potential application of artificial intelligence technology in forensic odontology can be categorized into four: (1) human bite marks, (2) sex determination, (3) age estimation, and (4) dental comparison. This powerful tool can solve humanity's problems by giving an adequate number of datasets, the appropriate implementation of algorithm architecture, and the proper assignment of hyperparameters that enable the model to perform the prediction at a very high level of performance.

Conclusion

The reviewed articles demonstrate that machine learning techniques are reliable for studies involving continuous features such as morphometric parameters. However, machine learning models do not strictly require large training datasets to produce promising results. In contrast, deep learning enables the processing of unstructured data, such as medical images, which require large volumes of data. Occasionally, transfer learning was used to overcome the limitation of data. In the meantime, this method's capacity to automatically learn task-specific feature representations has made it a significant success in forensic odontology.

Identification of a person with the help of bite mark analysis

Forensic dentistry is an essential part of Forensic science, mainly involves the identification of an assailant by comparing a record of their dentition (set of teeth) with a record of a bite mark left on a victim. Other uses in law for dentists include the identification of human remains, medico-legal assessment of trauma to oral tissues, and testimony about dental malpractice. While the practice of human identification is well established, validated and proven to be accurate, the practice of bite mark analysis is less well accepted. The principle of identifying an injury as a bite mark is complex and, depending on severity and anatomical location, highly subjective. Following the identification of an injury as a bite mark, the comparison of the pattern produced to a suspect's dentition is even more contentious and an area of great debate within contemporary odontological practice. Like fingerprints and DNA, bite marks are unique to an individual - such as distance and angles between teeth, missing teeth, fillings and dental work. This type of impression evidence can be left in the skin of a victim and also can be in food, chewing gum and other miscellaneous items such as pens and pencils. The advent of DNA analysis and its recovery from bite marks has offered an objective method of bite mark analysis.

Reality bites--A ten-year retrospective analysis of bitemark casework in Australia

Criticism of forensic science, particularly that of bitemark analysis, has become increasingly common in the last decade. Much of the criticism directed at forensic odontology cites cases where miscarriages of justice have occurred when erroneous, over-confident or even false bitemark evidence has been tendered by odontologists. Despite Australia's own experience with such cases in the past, it is postulated that this does not represent the true nature of bitemark analysis as practiced by odontologists today-at least in this country. A review of 119 cases from the last 10 years confirms that 'identification' of a suspect is rarely, if ever, offered, and that conclusions reached by odontologists with respect to bitemark analysis are generally conservative. However, the results of this study also indicate that in a small but significant proportion of cases, there is still some tendency to reach conclusions that could be considered over-confident when considering the overall quality of the physical evidence offered. It is suggested that odontologists should avoid making conclusive remarks regarding the origin of the mark, or the identification of a perpetrator, when such comments are realistically precluded, given the low evidentiary value of the mark itself.

Validity of a dichotomous expert response in bitemark analysis using 3-D technology

Despite efforts to quantify bitemark evidence, comparison procedures remain subjective and yield different degrees of certainty. Our aim was to study the effectiveness of a comparison procedure requiring a dichotomous response by the expert. We compared overlays from 3-D images of dental casts and bite impressions, obtained using DentalPrint(©) software. Receiver operating characteristic (ROC) analysis was performed on the results of 104 comparisons, finding an area under the ROC curve of 0.955 (standard error=0.029; 95% CI, 0.896-0.986), sensitivity of 92.3% (95% CI, 74.8-98.8) and specificity of 98.7% (95% CI, 93.0-99.8). According to these findings, this bitemark analysis procedure is highly accurate, although study limitations are discussed, placing these results in context. The main advantage of the dichotomous decision model is that it can be more easily understood, facilitating course of justice. Further research is warranted to explore the potential of this approach as an alternative to diagnostic decisions based on certainty levels.

The recovery and persistence of salivary DNA on human skin

Salivary DNA is encountered in many crimes, such as sexual assaults and murders. In this study, saliva from three male donors was deposited on the skin of three female recipients. The amount of male salivary DNA remaining on the female skin was measured over a 96-h period using the Quantifiler™ Y Human Male DNA Quantification Kit. In eight of the nine experiments, a full male DNA profile matching the donor was obtained even after 96 h. In addition, the study showed that the concentration of salivary DNA varied from donor to donor and from day to day. The efficiency of two recovery methods, wet and dry swabbing and minitaping, was compared. The results indicate the tapelift method gave higher DNA recovery. This study also examined the secondary transfer of salivary DNA from skin to fabrics. Cotton and polyester give higher DNA transfer than leather.

Experimental study of bite mark injuries by digital analysis

This study was based on a morphometric analysis of bite marks starting from the quantitative definition of the anterior teeth by the geometric acquisition of the "injuries," using bite marks impressed on pig skin and plastic. Each mark was photographed and acquired. A dedicated program automatically supplied the values of the shape factors and the areas of the pattern geometric figures. The values obtained for the homologous samples were compared to the heterologous values. Statistical comparison was made of the sets by linear regression, determining the correlation coefficient and the determination coefficient for each value. Results showed only 4.8% and 2% of overlap between homologous and heterologous values. This study was carried out in attempting to quantitatively define the anterior teeth of the human dentition and the procedure described and the results obtained support the advantage of morphometric studies and computer-aided programs in this study of bite marks.

Is current bite mark analysis a misnomer?

Four human-to-human bite mark cases in which forensic odontological opinion was found to be in error, or at best deeply divided between experts, are described. These are used to illustrate that there is a growing awareness on the part of the legal profession that bite mark opinions by experts may often be little more than that and that these opinions often cannot be substantiated given the paucity of rigorous scientific evaluation, and will therefore be increasingly challenged. This may not best serve justice and so it is argued that forensic odontology needs to bring more scientific rigour to the evaluation of bite marks. This may threaten to disenfranchise some of the current practitioners and there may be some resistance to change. Forensic odontology is not the only identification science facing such problems, but nevertheless a paradigm shift is predicted in the way bite mark evidence will have to be gathered and evaluated in the future. Some new scientific approaches are described that strive to unravel some of the most basic problems confronting our profession when we attempt to make morphometric comparisons between injuries and the dentition of the biter. A need to capture the actions and consequences of biting in 3D dimensions and simultaneously in real-time is proposed as a path of investigation highly likely to bring some clarity to a confused situation. There is also an urgent need for the ongoing controversy between some of our eminent peers relating to the assumed uniqueness, or otherwise, of the human anterior dentition to be resolved.

Reconstruction of deformed bite marks using affine transformations

Bite marks inflicted on animate and inanimate objects will undergo a certain degree of deformation. This fact remains one of the biggest stumbling blocks when analyzing evidence for court presentation. It has been demonstrated that the reliability of pattern association analysis will not be affected by minimal degrees of warping, shrinkage, and distortion. In this study, affine transformations were applied to bite marks to establish if minimal distortions would affect the mathematically determined relationships of the defined features. In a real case study, it was then tested whether the distorted bite mark matched the dentition of the perpetrator by applying an affine transformation. This was confirmed to be the case. Affine transformations will thus not affect the relationships of the individual features found in bite marks. The numerical calculations validate the reliability of pattern association analysis in the presence of minimal amounts of warping, shrinkage, and distortion.

Comparison of simulated human dermal bitemarks possessing three-dimensional attributes to suspected biters using a proprietary three-dimensional comparison

The infliction of a bite is a four-dimensional space-time event that is ideally analyzed with three-dimensional (3-D) technology. Comparison of 2-D images (photographs) of a bitemark with a 3-D replica of a suspect's dentition is challenging. The authors present a technique that produces 3-D images of indented marks and dentitions for comparisons. Study models and corresponding dental-wax bites were digitized by 3-D scanning, and comparison overlays were generated using DentalPrint software. The effectiveness of the method was analyzed by determining the area under receiver operating characteristic (ROC) curve and the sensitivity, specificity and 95% confidence interval (CI) for each cut-off point. An area under the ROC curve of 0.953 (SE=0.029; 95% CI=0.893-0.985) and high sensitivity and specificity values were obtained for 104 comparisons made by an expert examiner, who correctly identified 92.3% of matching dentitions and 98.7% of non-matching dentitions. This technique can be considered a highly accurate method of bitemark analysis, although indentations must be present in the injury, limiting the cases that can be resolved. A comparative study of the same dentitions using 2-D bitemark photography confirmed the superiority of the new approach.

Dental superimposition: a pilot study for standardising the method

Dental superimposition is becoming more and more important because of the increasing number of illegal immigrants (at least in Italy), with no clinical history, no personal effects or relatives useful for genetic comparison, whose friends and acquaintances can usually only produce photographs. Very few authors have been involved in devising and using this method. The goal of the present study is to establish whether it is possible, and under which conditions, to identify individuals by dental superimposition of teeth visible in an ante-mortem photograph and dental casts of an unidentified body, and to develop a protocol for the spatial orientation analysis of the dentition and qualitative and semi-quantitative analysis of superimpositions. A non-mathematical scoring system has been applied to each superimposition as a first step towards the optimisation of a cheap, quick, semi-quantitative method of identifying individuals when other more used methods are not applicable.

Effectiveness of comparison overlays generated with DentalPrint software in bite mark analysis

Validation studies of the new DentalPrint software were carried out with experimental bite marks in pigskin. The bite marks were digitally photographed according to the ABFO guidelines for evidence collection. Dental casts used in the experiment were scanned in 3D and 2D, and comparison overlays were generated using DentalPrint and Adobe Photoshop software, respectively. Digitized photographs of the experimental bite marks and the biting edges obtained in the overlays were compared by two different examiners to analyze the impact of training and experience with the two methods. Receiver operating characteristic (ROC) analysis, sensitivity, specificity, and 95% confidence intervals for each cutoff point were calculated. The expert examiner using DentalPrint obtained the best results, with an area under the ROC curve of 0.76 (SE=0.057; CI at 95%=0.652-0.876). Fairly high specificity values were found for DentalPrint, and the best results were obtained for the cutoff value that discriminated between the examiner's response "biter" and the rest of the possible (specificity 97.9%, CI at 95%=93.2%-99.6%). Therefore, the results presented here indicate that DentalPrint is a useful, accurate tool for forensic purposes, although further research on the comparison process is needed to enhance the validity of bite mark analysis.

The barriers to achieving an evidence base for bitemark analysis

Forensic dentistry is the union of two scientific disciplines, both of which are undergoing a renewed scientific rigor. In forensic science the advent of the Daubert ruling has required that judges assess the forensic value of 'expert testimony' ensuring that techniques, methodologies and practices are not only commonly accepted (as was the previous hurdle during the Frye era) but that error rates, assessment of reliability and validation studies are published to support their use. This new degree of judicial scrutiny has been mirrored in the field of dentistry itself, where organisations search and summarise randomised controlled trials in order to recommend best practice and devise clinical care pathways that are firmly grounded in proven scientific research. Despite the obvious drive from both of these professions, forensic dentistry, and in particular the sub-discipline of bitemark analysis, has been remarkably slow to address the obvious deficiencies in the evidence base that underpins this element of forensic science. Reviews of the literature reveal that the vast majority of published works are case reports, and very little primary literature exists. This paper reviews those studies that have assessed aspects of bitemark analysis including the crucial issue of the uniqueness of the human dentition; the application of transparent overlays and the application of statistical probabilities in bitemark conclusions. There are numerous barriers to undertaking high quality research in the field of bitemark analysis, the most important of which is the use of a gold-standard that is acceptable both in terms of diagnostic research but is also forensically relevant. If bitemark analysis is to continue to play a role in the judicial process then there is an urgent need for high quality studies that meet the levels of forensic and scientific scrutiny applied to other disciplines within the criminal justice system. Studies are required to determine not that the human dentition is unique, but how this asserted uniqueness is represented on human skin and other substrates. The error rates associated with the analysis of bitemarks are required on a procedural level as well as an individual practitioner basis and scales and interpretative indices of bitemark severity and forensic significance should be validated and introduced into common use.

Forensic dentistry and human bite marks: issues for doctors

The human dentition can be used as a weapon of attack or defence. Bite mark injuries are common in cases of sexual assault, child abuse and homicide. Many bite injuries are first seen in casualty departments where quick and proper recovery of evidence can assist in analysing these injuries. This article describes different bite injuries, collection of evidence and comparative analysis methods.

The scientific basis for human bitemark analyses--a critical review

This article presents a discussion of the scientific basis for human bitemark analyses. Using a review of the literature, the major areas of contention within the field are assessed: including the accuracy of bitemarks on skin, the uniqueness of the human dentition, and analytical techniques. The review revealed a lack of valid evidence to support many of the assumptions made by forensic dentists during bitemark comparisons. The new level of judicial scrutiny of such scientific evidence is likely to emphasise this lack of knowledge upon which bitemark analysis relies. The authors call for a more scientific and evidence-based approach to forensic dental research.