- Email: [email protected]
Uniqueness of the anterior dentition three-dimensionally assessed for forensic bitemark analysis
Journal of Forensic and Legal Medicine 46 (2017) 58e65
Contents lists available at ScienceDirect
Journal of Forensic and Legal Medicine j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j fl m
Uniqueness of the anterior dentition three-dimensionally assessed for forensic bitemark analysis A. Franco a, b, *, G. Willems a, PHC Souza b, W. Coucke c, P. Thevissen a a
Department of Oral Health Sciences e Forensic Dentistry, KU Leuven & Dentistry, University Hospitals Leuven, Belgium , Brazil lica do Parana Department of Dentistry e Stomatology, School of Life Sciences, Pontifícia Universidade Cato c Free-lance Statistician, Heverlee, Belgium b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 17 June 2016 Received in revised form 17 November 2016 Accepted 19 January 2017 Available online 21 January 2017
The uniqueness of the human dentition (UHD) is an important concept in the comparative process in bitemark analysis. During this analysis, the incisal edges of the suspects' teeth are matched with the bitemarks collected from the victim's body or crime scenes. Despite playing an essential part to exclude suspects, the UHD contained in the involved incisal tooth edges remains an assumption on bitemark level. The present study was aimed, first, to investigate three-dimensionally (3D) the UHD within different quantities of dental material from the incisal edges; second, to test these outcomes in a bidimensional (2D) simulation. Four-hundred forty-five dental casts were collected to compose 4 study groups: I e randomly-selected subjects, II e orthodontically treated subjects, III e twins and IV e orthodontically treated twins. Additionally, 20 dental casts were included to create threshold groups on subjects from whom the dental impressions were taken at 2 different moments (Group V). All the dental ~o Paulo, casts were digitalized with an automated motion device (XCAD 3D® (XCADCAM Technology®, Sa SP, Brazil). The digital cast files (DCF) were integrated in Geomagic Studio® (3D Systems®, Rock Hill, SC, USA) software package (GS) for cropping, automated superimposition and pair-wise comparisons. All the DCF were cropped remaining 3 mm (part 1), 2 mm (part 2) and 1 mm (part 3) from the incisal edges of the anterior teeth. For a 2D validation, slices of 1 mm, not including incisal edges (part 4), were also cropped. These procedures were repeated in Group V, creating specific thresholds for each of the study parts. The 4 study groups were compared with its respective threshold using ANOVA test with statistical significance of 5%. Groups I, II and III did not differ from the corresponding threshold (Group V) in all study parts (p > 0.05). Scientific evidence to support the UHD was not observed in the current study. Bitemark analysis should not be disregarded but considered carefully when the suspects present similar dental alignment and morphology, such as in orthodontically treated subjects and twins, respectively. © 2017 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved.
Keywords: Bitemark Uniqueness Anterior dentition Forensic dentistry 3D morphometrics
1. Introduction The morphology of the human anterior teeth is associated with the teeth of primates from the Anthropoidea taxonomic group.1,2 Differently from most of the vertebrates that use their teeth exclusively for ingestion,1 humans consciously may bite while attacking or defending under violent circumstances.3 The patterned injury inflicted by the teeth on skin is named bitemark.3 The biomechanics behind the biting phenomenon is complex,4
* Corresponding author. Department of Oral Health Sciences e Forensic Dentistry, KU Leuven and Dentistry, University Hospitals Leuven, Kapucijnenvoer 7, block B, 3000 Leuven, Belgium. E-mail address: [email protected] (A. Franco).
involving not only the action of teeth, but also the position of the tongue,3 the occlusion, the relation between dental arches,4 the bite intention,5 the reaction of the injured person,4 and the type of the bitten material and its underlying structure.6 Bitemarks on skin may be typically presented with two arches facing each other, which often correspond to the anterior maxillary and mandibular dentitions.7 The dental information registered on the bitten tissue may be compared with the suspects' dentitions. The comparative approach is based on the assumption that the human anterior dentition is unique. The assumed uniqueness of the human dentition (UHD) guarantees that the dentition of no other person could have inflicted the bite. However, recent studies pointed out that the incisal edges of dentitions from different subjects may not be unique.8e11 A critical evaluation of these studies,12 indicated that
http://dx.doi.org/10.1016/j.jflm.2017.01.005 1752-928X/© 2017 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved.
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
methodological limitations, such as the sample selection (randomly performed13) and imaging techniques applied (bidimensionally),14 potentially biased the results. In the present research, the UHD was investigated using methodological improvements supported by the literature, such as stratifying the sample on orthodontically treated subjects8 and twins,15 and three-dimensionally (3D) registering the dental arches.16 These improvements restrict the search for uniqueness in populations with similar dentitions and register the dental morphology in a realistic manner. Additional methodological improvements applied, were the automated image superimposition, the creation of a threshold combining the possible measurement errors, and the analysis of dental material within different quantities. The last is of special importance in human bitemark analysis, because it enables the systematic assessment of the incisal edges using dental parts possibly involved in bitemarks. Moreover, due to the factors involved in the biomechanics of the biting phenomenon,3e6 the quantity of dental parts involved highly varies. Recent studies with 3D dental registration analyzed the anterior dental crown morphology from 1 up to 3 mm of the incisal edges.17,18 Studies using 2D technology were limited to investigate geometric contours of the incisal edges in occlusal view.8,14 Based on that, the 3D investigation on the UHD within small quantities of the anterior teeth, namely 1e3 mm of the incisal edges, becomes important to bring the research to a specific bitemark level. Moreover, the analysis of slices from the considered small quantities of the anterior teeth mimic the current status of the bitemark analysis, which is mainly based on the 2D superimposition of dental overlays from the anterior dentition in occlusal view.19 Knowing if the UHD is unique both in the 3D and in the 2D simulation set up is essential for bitemark purposes. The present study aims to assess the UHD in the anterior dentition based on systematically reducing the 3D quantity of dental material analyzed; and using slices of the analyzed dental material.
2. Material and methods The current research was conducted after the approval of the local Committee of Ethics in Research, under the protocol number: 19575613.2.0000.0020. Four-hundred forty-five dental casts were digitalized with an automated motion device with angular laser scanning (XCAD 3D® (XCADCAM Technology®, S~ ao Paulo, SP, Brazil) in resolution of <20 mm. The digital cast files (DCFs) were divided in 4 study Groups (I, II, III and IV) (Table 1). The included DCFs presented at least a completely clinically erupted permanent anterior dentition (from canine to canine). The DCFs with restorations, prostheses, and orthodontic retainers in the anterior teeth were excluded. In Group II, a high prevalence of
59
orthodontic retainers was detected in the mandibular arch, therefore the mandibular DCFs were excluded for analysis. All the DCFs were imported in Geomagic Studio® (3D Systems®, Rock Hill, SC, USA) software package (GS) for cropping the clinically visible dental crowns of the anterior teeth as tooth part of interest. Pair-wise DCF comparisons were performed in each group, separately for the maxilla and the mandible, using automated GS superimposition. In total, 2013 comparisons were performed, 110 in Group I; 1711 in Group II, 172 in Group III, 10 in Group IV. The outcomes of each pair-wise DCF comparison resulted in morphological differences quantified by GS as 4 measures, expressed in millimeters as the maximum positive deviation (max.þ), the maximum negative deviation (max.-), the average and the standard deviation (SD). The four quantified measures were converted in a single variable (Euclidean distance from zero), applying the following formula:
Distance ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Maxþ2 þMax2 þAverage2 þStandard deviation2 :
This procedure was essential to rank the most similar (closest to zero) and the most different (farthest from zero) pairs of DCF in each study group. The 5 most similar pairs of DCF of each group were selected and used in this study. Twenty additional DCFs were obtained from 5 subjects from whom dental impressions and related casts were made twice within a period of 7 days. This Reference Group (Group V) served to establish thresholds, allowing quantification of the errors obtained and included during the casts and DCFs acquisition. In part 1, all the DCFs were cropped leaving maximally 3 mm from the incisal edges of the anterior teeth (calculated from the incisal edge of the highest tooth). In part 2, only the groups with Euclidean distances above the threshold (considered unique) in part 1were used. The DCFs of these groups were re-imported in the software and cropped, retaining a tooth portion of maximally 2 mm from the incisal edges of the anterior teeth. In part 3, the same DCFs were re-cropped keeping maximally 1 mm of tooth material from the incisal edges (Fig. 1). In part 4, all the groups were used and the DCFs were cropped keeping a slice of 1 mm not including the incisal edges (Fig. 2). In each study part, a threshold value was established using Group V. Statistically, the comparison between the studied groups was performed applying One-way ANOVA20 with log-transformed distances, separately for maxilla and mandible. A correction for simultaneous hypothesis testing was applied according to Tukey's range test.21 The human dentition was considered unique when the mean Euclidean distance of any studied Group was statistically significantly higher than the respective reference Group. A Receiver Operator Characteristic (ROC) analysis was performed to assess the threshold potential of Group V e differently cropped in each study
Table 1 Sample distribution of the studied groups per dental arch, sex and zygosity. Group
I II III IV V
Sample
Random subjects Orthodontically treated subjects Twins Orthodontically treated twins Threshold subjectsa
Dental arch
Sex
Zygosity
n
Maxillary
Mandibular
Male
Female
Mono-
Di-
11 59 172 10 10
11 0 172 10 10
7 32 38 2 2
4 27 48 8 3
n/a n/a 39 5 n/a
n/a n/a 47 0 n/a
22 59 344 20 20
DCF: digital cast files; n/a: not applicable; n: sample size; A total of 2.013 pair-wise comparisons were performed: 110 in Group I; 1.711 in Group II, 172 in Group III, 10 in Group IV, and 10 in Group V. In Groups I and II the pair-wise comparisons were performed matching all the DCF in the sample, while in Groups III and IV the comparisons were performed exclusively between twin siblings. Yet in Group V, the pair-wise comparisons were performed with the DCF obtained from the same patients in moment 1 and 2. a In Group V, the threshold was developed performing the dental impression and digitalization of dental casts of five subjects in two different times. The sample distribution for twins (Groups III and IV) based on sex is expressed as pairs of monozygotic (mono-) and dizygotic (di-)siblings.
60
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
Fig. 1. Tooth parts of interest of the DCF used in study parts 1, 2 and 3 (3D analysis). Images A, B and C correspond to the tooth parts of interest of the digital cast files (DCFs) remaining 3 (part 1), 2 (part 2) and 1 mm (part 3) from the incisal edges, respectively.
part, for detecting non-equal (sensitivity) and equal dentitions (specificity). In this context, the Area Under the Curve (AUC) was obtained by means of bootstrapping. The statistical tests were performed with significance rate of 5% using Sþ® 8.0 (Tibco®, Palo Alto, California, USA) software package. 3. Results In part 1 (3 mm at incisal edges), only the mean Euclidean distances obtained for the maxillary (4.35) and the mandibular (4.83) DCFs of Group I (randomly-selected subjects) was higher than, and statistically different from the Euclidean distances from the respective threshold (Group V) (p < 0.05) (Table 2). The ROC analysis revealed sensitivity and specificity of 100% and 70% for maxillary DCFs (87%), and 80% and 73.3% for mandibular DCFs (AUC: 78%), respectively (Fig. 3). In parts 2 and 3 (2 mm or 1 mm at incisal edges, respectively), the mean Euclidean distances of the maxillary (part 2: 3.36, part 3: 2.86) and mandibular (part 2: 3.99; part 3: 3.94) DCFs of Group I remained above the respective thresholds with statistical significant difference (p < 0.05) (Table 2). In both parts the sensitivity and specificity reached 100% for maxillary and mandibular DCFs (AUC: 100%) (Figs. 4 and 5). In part 4 (slice of 1 mm) only Group I was above the respective threshold and statistically significant different from it (p < 0.05) (Table 2). The ROC analysis revealed sensitivity and specificity of 80% and 55% for maxillary DCFs (AUC: 71%), and 80% and 67% for mandibular DCFs (AUC: 76%), respectively (Fig. 6). 4. Discussion The UHD is an important concept in comparative processes of forensic odontology, especially in the fields of dental identification and bitemark analysis.22 Specifically in the last, a comparison is performed between the dentition of a suspect and the bitemark left on the victim or crime scene.3 However, a perfect match between the dentition and the bitemark solely indicates the convergence of evidences. If scientifically confirmed, the UHD would guarantee that only one suspect could inflict the related bitemark. In the
absence of uniqueness, the perpetrators could claim that other person(s) committed the crime. Despite having an important role in bitemarks, the UHD remains an assumption.12 Criticism on the validity of the bitemark evidence in Court was officially established in 200923, triggering scientific researches in three levels: the existence of the UHD8e11,24; the distortion of bitemarks on skin25e28; and the technical interpretation of the bitemark evidence.29,30 Investigations on the UHD were performed over time, presenting a highly varied methodology,12 showing that the UHD could be assessed using 2D (flatbed scanning and photography) and 3D (laser scanning) image registration; using dental casts or indentations on wax; and using the contour of the incisal edges (extracted through manual hand-tracing or digital landmarking) or the complete morphology of the incisal edges.12 Early studies, dating from 198215 and 198413, were characterized by the use of 2D image registration of bite impressions in wax. At that time, the human dentition was considered unique both on a stratified sample of twins15 and on a random population.13 Especially in a random population, the quantification of the possible combinations of tooth positions was performed indicating that “a general population sample demonstrates the uniqueness of the human dentition beyond any reasonable doubt”.13 However, these studies were mostly limited by the registration of the dental morphology in wax and its analysis in 2D. In 2007, geometric morphometrics was introduced in the field, allowing to study the UHD on flatbed scanner registration of the dental casts14 and suggesting evidence towards the existence of the UHD. The data registration was still
Table 2 Significance level of comparison between the Study Groups I, II, III and IV and the respective Threshold Group (Group V) per study part separate for each dental arch. Part
Dental arch
Comparison
Mean (SD)
1
Maxilla
GI vs. GV GII vs. GV GIII vs. GV GIV vs. GV GI vs. GV GIII vs. GV GIV vs. GV GI vs. GV GI vs. GV GI vs. GV GI vs. GV GI vs. GV GII vs. GV GIII vs. GV GIV vs. GV GI vs. GV GIII vs. GV GIV vs. GV
4.35 1.82 1.90 1.88 4.83 1.86 1.34 3.36 3.99 2.86 3.94 4.46 2.00 2.21 2.20 5.89 1.66 1.45
Mandible
2 3 4
Maxilla Mandible Maxilla Mandible Maxilla
Mandible
Fig. 2. Tooth part of interest of the DCF in study part 4 (simulated 2D analysis). In study part 4, the digital cast files (DCFs) cropped 3 mm from the incisal edges (part 1) were cropped at their gingival portion (A), remaining a slice of 1 mm (B). In axial (occlusal) view (C), these slices mimic the bidimensional registration of forensic bitemark indentations.
(1.55) (0.45) (0.07) (0.35) (0.77) (0.60) (0.15) (1.01) (0.30) (0.86) (0.56) (1.12) (0.48) (0.57) (0.69) (1.50) (0.52) (0.19)
p vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
1.31 1.31 1.31 1.31 1.48 1.48 1.48 1.12 1.41 1.07 1.08 1.78 1.78 1.78 1.78 1.45 1.45 1.45
(0.45) (0.45) (0.45) (0.45) (0.63) (0.63) (0.63) (0.61) (0.24) (0.60) (0.45) (0.82) (0.82) (0.82) (0.82) (0.40) (0.40) (0.40)
0.0001* 0.1836 0.0835 0.1041 0.0001* 0.4959 0.9910 0.0035* 0.0001* 0.0049* 0.0004* 0.0018* 0.8981 0.6663 0.7095 0.0001* 0.8505 0.9990
Part 1: digital cast files (DCF) cropped with remaining tooth part of interest of 3 mm from the incisal edges; Part 2: DCF cropped with remaining tooth part of interest of 2 mm from the incisal edges; Part 3: DCF cropped with remaining tooth part of interest of 1 mm from the incisal edges; Part 4: DCF cropped with remaining tooth part of interest of 1 mm slice excluding the incisal edges. GI: randomly-selected subjects; GII: orthodontically treated subjects; GIII: twins: GIV: orthodontically treated twins; GV: respective threshold (obtained specifically for each study part).Study parts 2 and 3 only compared GI because this was the only study group in which the UHD was observed in study part 1. Mean and standard deviation (SD) of the Euclidean distances obtained in each group. P value calculated from ANOVA test with significance rate of 5%. *Statistically significant difference (p < 0.05).
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
61
Fig. 3. Euclidean distances of the Groups I, II, III, IV and V obtained pair-wise comparing DCFs with remaining tooth part of interest of 3 mm from the incisal edges (study part 1), expressed in Boxplots (A, C) and ROC curves (B, C). A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects), II (orthodontically treated subjects), III (twins), IV (orthodontically treated twins), and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal maxillary (B) digital cast files (DCFs) as nonequal and specificity of 70% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 80% - for classifying equal mandibular (D) DCFs as non-equal and specificity of 73.3% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 87% and 78%, for maxillary (B) and mandibular (D) DCFs, respectively.
performed in 2D, using flatbed scanners. The 3D technology was used four years later, revealing a polemic finding, namely the lack of the UHD.9e11 Despite observing the lack of the UHD with sound statistical approach, these studies presented discussable methodology.12 The comparisons of dentitions relied on manually landmarking the incisal edges of the anterior teeth.9e11 The main restriction in this procedure is founded on operator-dependency, which comprehends the bias inherent to manually positioning landmarks in each examined dental cast. In the literature this bias was improved testing statistically operator reproducibility,8,14 and implementing it in equations as criteria for determining match or mismatch between dentitions.9 Moreover, the comparisons between dentitions were performed using Geometric Morphometrics e a technique that registers manually positioned landmarks as Cartesian coordinates31 and compares their spatial relation31 pairwise between dentitions. According to each study, a various number of landmarks were used, namely 148, 2414, and 3016. The location and number of these landmarks were chosen specially on the incisal edges of the frontal teeth. The use of a higher number of landmarks is observed in more recent studies, in an attempt to register more morphological information from the contour of the incisal edges of the dentitions.9 In the present study GS was used e a 3D modeling software package that enables the automatic morphological match by superimposing DCFs and searching for
their best fitting overlap. Similarly to Geometric Morphometrics, this software considers the 3D DCFs as sets of points that can be compared mutually. Differently from the literature, the comparison is not founded on manually positioned landmarks and the number of points in each 3D DCF sets normally exceeds 1.000 per six anterior teeth. Consequently, more morphological information was assessed in an operator-independent procedure. Class characteristics were used to describe the standard evidences found in human bitemarks on human skin. They are observed as two oval semi-arches facing each other with rectangular and circular contusions or indentations, remaining after the biting contact of incisors and canines with the skin.7 However, the literature has no consensus on the exact quantity of dental material that is involved in bitemarks, but studies with bite impressions on foodstuff6 and wax17 suggest that the incisal edges may be registered in these bitten surfaces from 1 to 3 mm. In the living, the quantity of dental material involved depends not only on the biting force applied, but also on the visco-elastic properties of the skin. A human bite may reach power of 37.72kg32, depending on the size of the muscle, the bone morphology and the articulation between the maxilla and the mandible. On the other hand, the elasticity of the skin decreases considerably25 from a subjected power of 13.5 kg and responds to skin rupture with higher biting forces. However, the human skin responds heterogeneously, depending on the
62
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
Fig. 4. Euclidean distances of the Groups I and V obtained pair-wise comparing DCFs with remaining tooth part of interest of 2 mm from the incisal edges (study part 2), expressed in Boxplots (A, C) and ROC curves (B, C). A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects) and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal maxillary (B) digital cast files (DCFs) as non-equal and specificity of 100% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal mandibular (D) DCFs as non-equal and specificity of 100% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 100% for maxillary (B) and mandibular (D) DCFs.
underlying tissue,25 the age, and the weight of the victim.26 This information justifies the lack of literature consensus on the quantity of dental material involved in bitemarks (especially those inflicted) on human skin. Therefore, in the present study a systematic selection of the 3D quantity of remaining tooth parts of interest was established (parts 1, 2 and 3). Moreover, study part 4 was designed to mimic the currently overall established bitemark practice, which relies on the 2D registration of the bitemark, registering and comparing the contours of the bitemark and transparent overlays of the suspect's dental casts.19 In this part, bitemark contours were mimicked by selecting a thin slice (1 mm) of the DCFs cropped axially at maximally 3 mm from the incisal tooth edges. Specifically, the slice was selected at this region because theoretically it includes the bitemark contours that could be generated after impressing the incisal edges for 1 and 2 mm on the skin. Similarly, the literature describes the production of 2D comparison overlays from 3D dental casts.33 This procedure was enabled by a software package with tools specifically developed for slicing digital models and improving the 2D analysis in the bitemark practice.17 The authors matched the obtained 3D dental overlays digitally with photographs of bitemarks on pig skin.17 However, the 3D dental overlays were generated from the impression of dental casts on wax,17 creating an indirect procedure. In this context, the present study has the advantage of generating overlays (slices) directly from the 3D DCF, tackling the potential distortion inherent to indirect procedures.
The present study differed from earlier published studies9e11,13,16,34,35 using not exclusively DCFs from randomly-selected subjects, but also from three populations stratified on dental similarity. First, orthodontically treated subjects were selected because they present a “lower level of individuality”14 due to similar dental alignment. Second, twin pairs were studied because, theoretically, they have identical dentitions, due to a potential genetic control over dental morphology.36 However, it is currently known that morphological dental variations exist even among monozygotics,36e38 resulting in similar but not identical dental morphology. Third, a population of orthodontically treated monozygotic twins was examined. In this twin group, the acquired changes in the phenotype expression of the dental arrangement were orthodontically corrected. This stratification was probable the main reason contributing to the lack of evidence supporting UHD in the present study. Moreover the additional stratification, selecting only the five pairs of DCF with the smallest pair-wise Euclidean differences in each studied group, increased the degree of morphological similarity in the studied samples. Further on, another aspect contributing to the increase in morphological similarity between dentitions and consequent to the lack of evidence of UHD was the reduction in the quantity of dental material studied. The high level of sample stratification on dental similarity is reflected in the sensitivity and specificity obtained comparing each group with its corresponding threshold (Group V). A perfect classification of non-equal (sensitivity: 100%) and equal (specificity:
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
63
Fig. 5. Euclidean distances of the Groups I and V obtained pair-wise comparing DCFs with remaining tooth part of interest of 1 mm from the incisal edges (study part 3), expressed in Boxplots (A, C) and ROC curves (B, C) A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects) and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal maxillary (B) digital cast files (DCFs) as non-equal and specificity of 100% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal mandibular (D) DCFs as non-equal and specificity of 100% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 100% for maxillary (B) and mandibular (D) DCFs.
100%) DCFs was obtained in Group I (randomly-selected subjects). It reveals that even retaining the 5 most similar dentitions (minimal Euclidean distances), Group I presented a distinguishable discrepant dental morphology related to its respective threshold. In relation to the other groups, the sensitivity and specificity of group V decreased up to 80% and 70%, respectively. However, it maintained an acceptable discriminative accuracy for the threshold expressed in AUC values above 71%. The only exception was observed in the 2D bitemark mimicking registration (part 4), in which Group V presented a specificity of 55% and 67% for maxillary and mandibular DCFs, respectively. This may be explained by the influence of the limited quantity of tooth material considered in study part 4. Although in Group I no equal DCFs were found, even on bite mark level, no UHD may be claimed, because a randomly chosen individual can potentially be part of one of the stratified subgroups in which the evidence indicated that the UHD is not supported. In the present study, analysis of maxillary and mandibular dentitions resulted in similar outcomes in regard to the UHD. The findings suggest that no predilection for dental arches should be made in bitemark cases, indicating that all the available dental information should be analyzed.3 On the other hand, the present research was limited by the lack of mandibular DCFs from orthodontically treated subjects, encouraging the use of this material in further studies in the field. Another limitation was the possible error in the procedure of manually taking dental impressions and
creating dental casts in plaster (operator dependency). This limitation was addressed by incorporating the potential error during the manual procedure to take dental casts in the threshold. However, future research could be performed replacing this manual procedure by intra-oral scanning. Moreover, the present research was designed in the context of bitemarks without analyzing the bitemarks themselves, but investigating fundamentally uniqueness from analyzing dentitions. The analysis of bitemarks could be considered a following process towards the validation of the present findings within further studies of bitemarks on skin, foodstuff and objects. Evidence that the dentitions of subjects are indistinguishable using the techniques described indicates that a single bitemark could have been made by more than one individual. In the present study, the comparison between dentitions was performed using dental casts scanned as DCFs and analyzed by selecting as the area of interest the incisal edges of the anterior teeth. The quantity of dental material analyzed was diminished in each study step from 3 mm to 1 mm from the incisal edges based on the parts of the teeth assumed to be potentially involved in bitemarks on human skin. The dental material registered in a bitemark may reflect different morphological traits of the human dentitions, such as the size, shape, arrangement and position of the teeth in the arch.22 All of these morphological traits are found in the DCFs cropped systematically in this study. Specifically, the shape, size, arrangement and position of the incisal edges of the anterior teeth (portion of dental
64
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
Fig. 6. Euclidean distances of the Groups I, II, III, IV and V obtained pair-wise comparing DCFs with remaining tooth part of interest a slice of 1 mm not including the incisal edges (study part 4), expressed in Boxplots (A, C) and ROC curves (B, C). A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects), II (orthodontically treated subjects), III (twins), IV (orthodontically treated twins), and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 80% - for classifying non-equal maxillary (B) digital cast files (DCFs) as nonequal and specificity of 55% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 80% - for classifying non-equal mandibular (D) DCFs as non-equal and specificity of 66.7% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 71% and 76%, for maxillary (B) and mandibular (D) DCFs, respectively.
crowns often involved in bitemarks) were investigated in study parts 1e4. Especially in study part 4 (slice of the anterior dentition), these morphological traits are brought to a more practical level, in which the parts of the teeth would not be completely visible but their contours would be shown e simulating a bitemark overlay tracing. So, the process to gradually decrease the quantity of dental material is significant to investigate whether or not the human dentition can create unique bite patterns on skin. However, the reduction of dental material in the present research was based on a specific bitemark pattern assumption, in which only the marks of the incisal edges of the six upper and lower anterior teeth were registered on human skin e while the literature indicates that other tooth parts may be registered in bitemarks7 such as the palatal surfaces of the anterior teeth. Further studies in the field are needed for the investigation of the UHD within the palatal surfaces of the anterior teeth. Other patterns may include the registration of premolars, palatal or lingual surfaces of anterior teeth, interrupted or uninterrupted arches, abrasions, contusions, as well as partial marks or multiple overlapping marks of dental arches.7 These variations in bitemark patterns not addressed in this research should be interpreted as limitations to be improved. Other studies in the field are necessary to verify exclusively whether dentitions are unique considering dental arches with more teeth and different tooth parts (including premolars and palatal or lingual surfaces of the anterior dentition). Preliminary evidence is observed in the literature39 indicating that the increase in the amount of dental
information tends to increase the detectable morphological differences between dentitions leading to potential uniqueness. In specific, the inclusion of first and second premolars increased the morphological difference between dentition without statistical significance (p > 0.05 for the maxilla and the mandible), but with a significant clinical impact. However, assumptions concerning uniqueness are feasible only with a corresponding threshold group for comparison. Additionally, computed tomography also may be investigated as a mean for analyzing in detail slices taken from the anterior dentition (<1 mm from the incisal edges) and their uniqueness. This technology enables an accurate analysis of the dental morphology with higher magnification than used in the present research. Thus, it would benefit not only the investigations on the uniqueness of the human dentition but also the research investigating the biting phenomenon itself. Considering the specific assumed pattern of a bitemark studied, this chapter demonstrated how highly similar the dentitions from different subjects can be when their teeth are well aligned (Group II) or when their teeth present similar morphology (Group III). These results indicate that forensic BM analysis must not be disregarded, but performed after case selection. This was also advised in the literature in the sense that “not every case is suitable for analysis”.40 The present chapter suggests that BM cases involving suspects with similar dental arrangement and morphology should be interpreted carefully. Further studies should be conducted exploring the limits of case selection in BM analysis. In fact, the case
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
selection should be based on distinctive individual tooth or arch characteristics to allow decisions based on the exclusion or nonexclusion of suspected biters associated with legal issues. On the other hand, uniqueness remains a fundamental question in the quest to assure that no individual could be falsely associated or linked to a bitemark. Finding a distinctive trait to exclude individuals is a difficult task when the suspected biters have well aligned teeth or similar dental morphology. It justifies the present chapter and confirms the importance of the UHD for legally undisputable bitemark scenarios. 5. Conclusion Scientific evidence for the UHD was not detected in the current research. Pairs of dentitions with morphological differences smaller than their respective measuring threshold were observed. This was related to the high level of sample stratification used. The included subjects had a high degree of similar dental alignment (orthodontically treated) and morphology (twins). Moreover, the standard within class characteristics of human bitemarks on human skin were mimicked, reducing the quantity of dental material analyzed in all study parts. Because the mimicked standard of bitemark patterns strongly vary in the forensic practice, further research increasing or even reducing the quantity of dental material analyzed is necessary. The current findings indicate that in forensic practice, investigations of bitemarks with the mimicked standards can only be performed in closed populations. Moreover, case selection should be based on distinctive individual tooth or dentition characteristics to allow decisions based on the exclusion or inclusion of suspected biters associated with legal issues. Conflict of interest None. Acknowledgements Author AF would like to express gratitude to the Coordination for the Improvement of Higher Education Personnel (CAPES) (Protocol number: 99999.003423/2015-08) for funding the present research. References 1. Lucas P. Dental Functional Morphology: How Teeth Work. Cambridge: Cambridge University Press; 2004. 2. Agrawal KR, Lucas PW. The mechanics of the first bite. Proc Biol Sci. 2003;270: 1277e1282. http://dx.doi.org/10.1098/rspb.2003.2361. 3. Dorion RB. Bite mark evidence. J Can Dent Assoc. 1982;48:795e798. 4. Nambiar P. Forensic odontology - bite marks - a review of the literature (part 1). Dent J Malays. 1994;14:32e36. 5. Whittaker D, McDonald D. An Atlas of Forensic Odontology. London: Wolfe Medical Publications; 1989. 6. Webster G. A suggested classification of bite marks in foodstuffs in forensic dental analysis. Forensic Sci Int. 1982;20:45e52. 7. Dorion R. Bite Mark Evidence: A Color Atlas and Text. second ed. Boca Raton: CRC Press; 2011. 8. Sheets HD, Bush PJ, Brzozowski C, Nawrocki LA, Ho P, Bush MA. Dental shape match rates in selected and orthodontically treated populations in New York State: a two-dimensional study. J Forensic Sci. 2011;56:621e626. http:// dx.doi.org/10.1111/j.1556-4029.2011.01731.x. 9. Sheets HD, Bush PJ, Bush MA. Patterns of variation and match rates of the anterior biting dentition: characteristics of a database of 3d-scanned dentitions. J Forensic Sci. 2013;58:60e68. http://dx.doi.org/10.1111/j.15564029.2012.02293.x. 10. Bush MA, Bush PJ, Sheets HD. Statistical evidence for the similarity of the human dentition. J Forensic Sci. 2011;56:118e123. http://dx.doi.org/10.1111/ j.1556-4029.2010.01531.x.
65
11. Bush MA, Bush PJ, Sheets HD. Similarity and match rates of the human dentition in three dimensions: relevance to bitemark analysis. Int J Leg Med. 2011;125:779e784. http://dx.doi.org/10.1007/s00414-010-0507-8. 12. Franco A, Willems G, Souza PHC, Bekkering GE, Thevissen P. The uniqueness of the human dentition as forensic evidence: a systematic review on the technological methodology. Int J Leg Med. 2015;129:1277e1283. http://dx.doi.org/ 10.1007/s00414-014-1109-7. 13. Rawson RD, Ommen RK, Kinard G, Johnson J, Yfantis A. Statistical evidence for the individuality of the human dentition. J Forensic Sci. 1984;29:245e253. 14. Kieser JA, Bernal V, Neil Waddell J, Raju S. The uniqueness of the human anterior dentition: a geometric morphometric analysis. J Forensic Sci. 2007;52: 671e677. http://dx.doi.org/10.1111/j.1556-4029.2007.00403.x. 15. Sognnaes RF, Rawson RD, Gratt BM, Nguyen NB. Computer comparison of bitemark patterns in identical twins. J Am Dent Assoc. 1982;105:449e451. 16. Blackwell SA, Taylor RV, Gordon I, et al. 3-D imaging and quantitative comparison of human dentitions and simulated bite marks. Int J Leg Med. 2007;121: 9e17. http://dx.doi.org/10.1007/s00414-005-0058-6. 17. Martin-de-las-Heras S, Tafur D. Comparison of simulated human dermal bitemarks possessing three-dimensional attributes to suspected biters using a proprietary three-dimensional comparison. Forensic Sci Int. 2009;190:33e37. http://dx.doi.org/10.1016/j.forsciint.2009.05.007. 18. Martin-de-Las-Heras S, Tafur D, Bravo M. A quantitative method for comparing human dentition with tooth marks using three-dimensional technology and geometric morphometric analysis. Acta Odontol Scand. 2014;72:331e336. http://dx.doi.org/10.3109/00016357.2013.826383. 19. Bowers C, Johansen R. Digital Analysis of Bite Mark Evidence. Indianapolis: Forensic Imaging Inst; 2000. 20. Casella G. Statistical Design. New York: Springer; 2008. 21. Tukey J. Comparing individual means in the analysis of variance. Biometrics. 1949;5:99e114. 22. Senn DR, Weems M. Manual of Forensic Odontology. fifth ed. CRC Press; 2013. 23. Committee on Identifying the Needs of the Forensic Sciences Community NRC. Strengthening Forensic Science in the United States: A Path Forward. 2009. 24. Tuceryan M, Li F, Blitzer HL, Parks ET, Platt JA. A framework for estimating probability of a match in forensic bite mark identification. J Forensic Sci. 2011;56. http://dx.doi.org/10.1111/j.1556-4029.2010.01571.x. 25. Bush MA, Thorsrud K, Miller RG, Dorion RBJ, Bush PJ. The response of skin to applied stress: investigation of bitemark distortion in a cadaver model. J Forensic Sci. 2010;55:71e76. http://dx.doi.org/10.1111/j.15564029.2009.01235.x. 26. Bush MA, Miller RG, Bush PJ, Dorion RBJ. Biomechanical factors in human dermal bitemarks in a cadaver model. J Forensic Sci. 2009;54:167e176. http:// dx.doi.org/10.1111/j.1556-4029.2008.00908.x. 27. Miller RG, Bush PJ, Dorion RBJ, Bush MA. Uniqueness of the dentition as impressed in human skin: a cadaver model. J Forensic Sci. 2009;54:909e914. http://dx.doi.org/10.1111/j.1556-4029.2009.01076.x. 28. Sheets HD, Bush PJ, Bush MA. Bitemarks: distortion and covariation of the maxillary and mandibular dentition as impressed in human skin. Forensic Sci Int. 2012;223:202e207. http://dx.doi.org/10.1016/j.forsciint.2012.08.044. 29. Bowers CM, Pretty IA. Expert disagreement in bitemark casework. J Forensic Sci. 2009;54:915e918. http://dx.doi.org/10.1111/j.1556-4029.2009.01073.x. 30. Page M, Taylor J, Blenkin M. Expert interpretation of bitemark injuriesea contemporary qualitative study. J Forensic Sci. 2013;58:664e672. http:// dx.doi.org/10.1111/1556-4029.12108. 31. Webster M, Sheets H. A practical introduction to landmark-based geometric morphometrics. Paleontol Soc Pap. 2010;16:163e188. 32. Paphangkorakit J, Osborn JW. The effect of pressure on a maximum incisal bite force in man. Arch Oral Biol. 1997;42:11e17. 33. Las Heras S, Valenzuela A, Ogayar C, Valverde J, Torres J. Computer-based production of comparison overlays from 3D-scanned dental casts for bite mark analysis. J Forensic Sci. 2005;50:127e133. 34. Nambiar P, Bridges TE, Brown KA. Quantitative forensic evaluation of bite marks with the aid of a shape analysis computer program: Part 1; the development of “SCIP” and the similarity index. J Forensic Odontostomatol. 1995;13: 18e25. 35. Nambiar P, Bridges TE, Brown KA. Quantitative forensic evaluation of bite marks with the aid of a shape analysis computer program: Part 2; “SCIP” and bite marks in skin and foodstuffs. J Forensic Odontostomatol. 1995;13:26e32. 36. Osborne R, Horowitz S, De George F. Genetic variation in tooth dimensions: a twin study of the permanent anterior teeth. Physiol Res. 1958;10:350e356. 37. Lundstrom A. Tooth morphology as a basis for distinguishing monozygotic and dizygotic twins. Am J Hum Genet. 1963;15:34e43. 38. Townsend G, Richards L, Brown T, Burgess V. Twin zygosity determination on the basis of dental morphology. J Forensic Odontostomatol. 1988;6:1e15. 39. Franco A, Willems G, Couto Souza PH, Coucke W, Thevissen P. Three-dimensional validation of the impact of the quantity of teeth or tooth parts on the morphological difference between twin dentitions. J Forensic Odontostomatol. 2016;1:27e37. 40. Barsley R, Freeman A, Metcalf R, Senn D, Wright F. Bitemark analysis. J Am Dent Assoc. 2012;143:444e446. author reply 446, 448.
Contents lists available at ScienceDirect
Journal of Forensic and Legal Medicine j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j fl m
Uniqueness of the anterior dentition three-dimensionally assessed for forensic bitemark analysis A. Franco a, b, *, G. Willems a, PHC Souza b, W. Coucke c, P. Thevissen a a
Department of Oral Health Sciences e Forensic Dentistry, KU Leuven & Dentistry, University Hospitals Leuven, Belgium , Brazil lica do Parana Department of Dentistry e Stomatology, School of Life Sciences, Pontifícia Universidade Cato c Free-lance Statistician, Heverlee, Belgium b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 17 June 2016 Received in revised form 17 November 2016 Accepted 19 January 2017 Available online 21 January 2017
The uniqueness of the human dentition (UHD) is an important concept in the comparative process in bitemark analysis. During this analysis, the incisal edges of the suspects' teeth are matched with the bitemarks collected from the victim's body or crime scenes. Despite playing an essential part to exclude suspects, the UHD contained in the involved incisal tooth edges remains an assumption on bitemark level. The present study was aimed, first, to investigate three-dimensionally (3D) the UHD within different quantities of dental material from the incisal edges; second, to test these outcomes in a bidimensional (2D) simulation. Four-hundred forty-five dental casts were collected to compose 4 study groups: I e randomly-selected subjects, II e orthodontically treated subjects, III e twins and IV e orthodontically treated twins. Additionally, 20 dental casts were included to create threshold groups on subjects from whom the dental impressions were taken at 2 different moments (Group V). All the dental ~o Paulo, casts were digitalized with an automated motion device (XCAD 3D® (XCADCAM Technology®, Sa SP, Brazil). The digital cast files (DCF) were integrated in Geomagic Studio® (3D Systems®, Rock Hill, SC, USA) software package (GS) for cropping, automated superimposition and pair-wise comparisons. All the DCF were cropped remaining 3 mm (part 1), 2 mm (part 2) and 1 mm (part 3) from the incisal edges of the anterior teeth. For a 2D validation, slices of 1 mm, not including incisal edges (part 4), were also cropped. These procedures were repeated in Group V, creating specific thresholds for each of the study parts. The 4 study groups were compared with its respective threshold using ANOVA test with statistical significance of 5%. Groups I, II and III did not differ from the corresponding threshold (Group V) in all study parts (p > 0.05). Scientific evidence to support the UHD was not observed in the current study. Bitemark analysis should not be disregarded but considered carefully when the suspects present similar dental alignment and morphology, such as in orthodontically treated subjects and twins, respectively. © 2017 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved.
Keywords: Bitemark Uniqueness Anterior dentition Forensic dentistry 3D morphometrics
1. Introduction The morphology of the human anterior teeth is associated with the teeth of primates from the Anthropoidea taxonomic group.1,2 Differently from most of the vertebrates that use their teeth exclusively for ingestion,1 humans consciously may bite while attacking or defending under violent circumstances.3 The patterned injury inflicted by the teeth on skin is named bitemark.3 The biomechanics behind the biting phenomenon is complex,4
* Corresponding author. Department of Oral Health Sciences e Forensic Dentistry, KU Leuven and Dentistry, University Hospitals Leuven, Kapucijnenvoer 7, block B, 3000 Leuven, Belgium. E-mail address: [email protected] (A. Franco).
involving not only the action of teeth, but also the position of the tongue,3 the occlusion, the relation between dental arches,4 the bite intention,5 the reaction of the injured person,4 and the type of the bitten material and its underlying structure.6 Bitemarks on skin may be typically presented with two arches facing each other, which often correspond to the anterior maxillary and mandibular dentitions.7 The dental information registered on the bitten tissue may be compared with the suspects' dentitions. The comparative approach is based on the assumption that the human anterior dentition is unique. The assumed uniqueness of the human dentition (UHD) guarantees that the dentition of no other person could have inflicted the bite. However, recent studies pointed out that the incisal edges of dentitions from different subjects may not be unique.8e11 A critical evaluation of these studies,12 indicated that
http://dx.doi.org/10.1016/j.jflm.2017.01.005 1752-928X/© 2017 Elsevier Ltd and Faculty of Forensic and Legal Medicine. All rights reserved.
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
methodological limitations, such as the sample selection (randomly performed13) and imaging techniques applied (bidimensionally),14 potentially biased the results. In the present research, the UHD was investigated using methodological improvements supported by the literature, such as stratifying the sample on orthodontically treated subjects8 and twins,15 and three-dimensionally (3D) registering the dental arches.16 These improvements restrict the search for uniqueness in populations with similar dentitions and register the dental morphology in a realistic manner. Additional methodological improvements applied, were the automated image superimposition, the creation of a threshold combining the possible measurement errors, and the analysis of dental material within different quantities. The last is of special importance in human bitemark analysis, because it enables the systematic assessment of the incisal edges using dental parts possibly involved in bitemarks. Moreover, due to the factors involved in the biomechanics of the biting phenomenon,3e6 the quantity of dental parts involved highly varies. Recent studies with 3D dental registration analyzed the anterior dental crown morphology from 1 up to 3 mm of the incisal edges.17,18 Studies using 2D technology were limited to investigate geometric contours of the incisal edges in occlusal view.8,14 Based on that, the 3D investigation on the UHD within small quantities of the anterior teeth, namely 1e3 mm of the incisal edges, becomes important to bring the research to a specific bitemark level. Moreover, the analysis of slices from the considered small quantities of the anterior teeth mimic the current status of the bitemark analysis, which is mainly based on the 2D superimposition of dental overlays from the anterior dentition in occlusal view.19 Knowing if the UHD is unique both in the 3D and in the 2D simulation set up is essential for bitemark purposes. The present study aims to assess the UHD in the anterior dentition based on systematically reducing the 3D quantity of dental material analyzed; and using slices of the analyzed dental material.
2. Material and methods The current research was conducted after the approval of the local Committee of Ethics in Research, under the protocol number: 19575613.2.0000.0020. Four-hundred forty-five dental casts were digitalized with an automated motion device with angular laser scanning (XCAD 3D® (XCADCAM Technology®, S~ ao Paulo, SP, Brazil) in resolution of <20 mm. The digital cast files (DCFs) were divided in 4 study Groups (I, II, III and IV) (Table 1). The included DCFs presented at least a completely clinically erupted permanent anterior dentition (from canine to canine). The DCFs with restorations, prostheses, and orthodontic retainers in the anterior teeth were excluded. In Group II, a high prevalence of
59
orthodontic retainers was detected in the mandibular arch, therefore the mandibular DCFs were excluded for analysis. All the DCFs were imported in Geomagic Studio® (3D Systems®, Rock Hill, SC, USA) software package (GS) for cropping the clinically visible dental crowns of the anterior teeth as tooth part of interest. Pair-wise DCF comparisons were performed in each group, separately for the maxilla and the mandible, using automated GS superimposition. In total, 2013 comparisons were performed, 110 in Group I; 1711 in Group II, 172 in Group III, 10 in Group IV. The outcomes of each pair-wise DCF comparison resulted in morphological differences quantified by GS as 4 measures, expressed in millimeters as the maximum positive deviation (max.þ), the maximum negative deviation (max.-), the average and the standard deviation (SD). The four quantified measures were converted in a single variable (Euclidean distance from zero), applying the following formula:
Distance ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Maxþ2 þMax2 þAverage2 þStandard deviation2 :
This procedure was essential to rank the most similar (closest to zero) and the most different (farthest from zero) pairs of DCF in each study group. The 5 most similar pairs of DCF of each group were selected and used in this study. Twenty additional DCFs were obtained from 5 subjects from whom dental impressions and related casts were made twice within a period of 7 days. This Reference Group (Group V) served to establish thresholds, allowing quantification of the errors obtained and included during the casts and DCFs acquisition. In part 1, all the DCFs were cropped leaving maximally 3 mm from the incisal edges of the anterior teeth (calculated from the incisal edge of the highest tooth). In part 2, only the groups with Euclidean distances above the threshold (considered unique) in part 1were used. The DCFs of these groups were re-imported in the software and cropped, retaining a tooth portion of maximally 2 mm from the incisal edges of the anterior teeth. In part 3, the same DCFs were re-cropped keeping maximally 1 mm of tooth material from the incisal edges (Fig. 1). In part 4, all the groups were used and the DCFs were cropped keeping a slice of 1 mm not including the incisal edges (Fig. 2). In each study part, a threshold value was established using Group V. Statistically, the comparison between the studied groups was performed applying One-way ANOVA20 with log-transformed distances, separately for maxilla and mandible. A correction for simultaneous hypothesis testing was applied according to Tukey's range test.21 The human dentition was considered unique when the mean Euclidean distance of any studied Group was statistically significantly higher than the respective reference Group. A Receiver Operator Characteristic (ROC) analysis was performed to assess the threshold potential of Group V e differently cropped in each study
Table 1 Sample distribution of the studied groups per dental arch, sex and zygosity. Group
I II III IV V
Sample
Random subjects Orthodontically treated subjects Twins Orthodontically treated twins Threshold subjectsa
Dental arch
Sex
Zygosity
n
Maxillary
Mandibular
Male
Female
Mono-
Di-
11 59 172 10 10
11 0 172 10 10
7 32 38 2 2
4 27 48 8 3
n/a n/a 39 5 n/a
n/a n/a 47 0 n/a
22 59 344 20 20
DCF: digital cast files; n/a: not applicable; n: sample size; A total of 2.013 pair-wise comparisons were performed: 110 in Group I; 1.711 in Group II, 172 in Group III, 10 in Group IV, and 10 in Group V. In Groups I and II the pair-wise comparisons were performed matching all the DCF in the sample, while in Groups III and IV the comparisons were performed exclusively between twin siblings. Yet in Group V, the pair-wise comparisons were performed with the DCF obtained from the same patients in moment 1 and 2. a In Group V, the threshold was developed performing the dental impression and digitalization of dental casts of five subjects in two different times. The sample distribution for twins (Groups III and IV) based on sex is expressed as pairs of monozygotic (mono-) and dizygotic (di-)siblings.
60
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
Fig. 1. Tooth parts of interest of the DCF used in study parts 1, 2 and 3 (3D analysis). Images A, B and C correspond to the tooth parts of interest of the digital cast files (DCFs) remaining 3 (part 1), 2 (part 2) and 1 mm (part 3) from the incisal edges, respectively.
part, for detecting non-equal (sensitivity) and equal dentitions (specificity). In this context, the Area Under the Curve (AUC) was obtained by means of bootstrapping. The statistical tests were performed with significance rate of 5% using Sþ® 8.0 (Tibco®, Palo Alto, California, USA) software package. 3. Results In part 1 (3 mm at incisal edges), only the mean Euclidean distances obtained for the maxillary (4.35) and the mandibular (4.83) DCFs of Group I (randomly-selected subjects) was higher than, and statistically different from the Euclidean distances from the respective threshold (Group V) (p < 0.05) (Table 2). The ROC analysis revealed sensitivity and specificity of 100% and 70% for maxillary DCFs (87%), and 80% and 73.3% for mandibular DCFs (AUC: 78%), respectively (Fig. 3). In parts 2 and 3 (2 mm or 1 mm at incisal edges, respectively), the mean Euclidean distances of the maxillary (part 2: 3.36, part 3: 2.86) and mandibular (part 2: 3.99; part 3: 3.94) DCFs of Group I remained above the respective thresholds with statistical significant difference (p < 0.05) (Table 2). In both parts the sensitivity and specificity reached 100% for maxillary and mandibular DCFs (AUC: 100%) (Figs. 4 and 5). In part 4 (slice of 1 mm) only Group I was above the respective threshold and statistically significant different from it (p < 0.05) (Table 2). The ROC analysis revealed sensitivity and specificity of 80% and 55% for maxillary DCFs (AUC: 71%), and 80% and 67% for mandibular DCFs (AUC: 76%), respectively (Fig. 6). 4. Discussion The UHD is an important concept in comparative processes of forensic odontology, especially in the fields of dental identification and bitemark analysis.22 Specifically in the last, a comparison is performed between the dentition of a suspect and the bitemark left on the victim or crime scene.3 However, a perfect match between the dentition and the bitemark solely indicates the convergence of evidences. If scientifically confirmed, the UHD would guarantee that only one suspect could inflict the related bitemark. In the
absence of uniqueness, the perpetrators could claim that other person(s) committed the crime. Despite having an important role in bitemarks, the UHD remains an assumption.12 Criticism on the validity of the bitemark evidence in Court was officially established in 200923, triggering scientific researches in three levels: the existence of the UHD8e11,24; the distortion of bitemarks on skin25e28; and the technical interpretation of the bitemark evidence.29,30 Investigations on the UHD were performed over time, presenting a highly varied methodology,12 showing that the UHD could be assessed using 2D (flatbed scanning and photography) and 3D (laser scanning) image registration; using dental casts or indentations on wax; and using the contour of the incisal edges (extracted through manual hand-tracing or digital landmarking) or the complete morphology of the incisal edges.12 Early studies, dating from 198215 and 198413, were characterized by the use of 2D image registration of bite impressions in wax. At that time, the human dentition was considered unique both on a stratified sample of twins15 and on a random population.13 Especially in a random population, the quantification of the possible combinations of tooth positions was performed indicating that “a general population sample demonstrates the uniqueness of the human dentition beyond any reasonable doubt”.13 However, these studies were mostly limited by the registration of the dental morphology in wax and its analysis in 2D. In 2007, geometric morphometrics was introduced in the field, allowing to study the UHD on flatbed scanner registration of the dental casts14 and suggesting evidence towards the existence of the UHD. The data registration was still
Table 2 Significance level of comparison between the Study Groups I, II, III and IV and the respective Threshold Group (Group V) per study part separate for each dental arch. Part
Dental arch
Comparison
Mean (SD)
1
Maxilla
GI vs. GV GII vs. GV GIII vs. GV GIV vs. GV GI vs. GV GIII vs. GV GIV vs. GV GI vs. GV GI vs. GV GI vs. GV GI vs. GV GI vs. GV GII vs. GV GIII vs. GV GIV vs. GV GI vs. GV GIII vs. GV GIV vs. GV
4.35 1.82 1.90 1.88 4.83 1.86 1.34 3.36 3.99 2.86 3.94 4.46 2.00 2.21 2.20 5.89 1.66 1.45
Mandible
2 3 4
Maxilla Mandible Maxilla Mandible Maxilla
Mandible
Fig. 2. Tooth part of interest of the DCF in study part 4 (simulated 2D analysis). In study part 4, the digital cast files (DCFs) cropped 3 mm from the incisal edges (part 1) were cropped at their gingival portion (A), remaining a slice of 1 mm (B). In axial (occlusal) view (C), these slices mimic the bidimensional registration of forensic bitemark indentations.
(1.55) (0.45) (0.07) (0.35) (0.77) (0.60) (0.15) (1.01) (0.30) (0.86) (0.56) (1.12) (0.48) (0.57) (0.69) (1.50) (0.52) (0.19)
p vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
1.31 1.31 1.31 1.31 1.48 1.48 1.48 1.12 1.41 1.07 1.08 1.78 1.78 1.78 1.78 1.45 1.45 1.45
(0.45) (0.45) (0.45) (0.45) (0.63) (0.63) (0.63) (0.61) (0.24) (0.60) (0.45) (0.82) (0.82) (0.82) (0.82) (0.40) (0.40) (0.40)
0.0001* 0.1836 0.0835 0.1041 0.0001* 0.4959 0.9910 0.0035* 0.0001* 0.0049* 0.0004* 0.0018* 0.8981 0.6663 0.7095 0.0001* 0.8505 0.9990
Part 1: digital cast files (DCF) cropped with remaining tooth part of interest of 3 mm from the incisal edges; Part 2: DCF cropped with remaining tooth part of interest of 2 mm from the incisal edges; Part 3: DCF cropped with remaining tooth part of interest of 1 mm from the incisal edges; Part 4: DCF cropped with remaining tooth part of interest of 1 mm slice excluding the incisal edges. GI: randomly-selected subjects; GII: orthodontically treated subjects; GIII: twins: GIV: orthodontically treated twins; GV: respective threshold (obtained specifically for each study part).Study parts 2 and 3 only compared GI because this was the only study group in which the UHD was observed in study part 1. Mean and standard deviation (SD) of the Euclidean distances obtained in each group. P value calculated from ANOVA test with significance rate of 5%. *Statistically significant difference (p < 0.05).
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
61
Fig. 3. Euclidean distances of the Groups I, II, III, IV and V obtained pair-wise comparing DCFs with remaining tooth part of interest of 3 mm from the incisal edges (study part 1), expressed in Boxplots (A, C) and ROC curves (B, C). A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects), II (orthodontically treated subjects), III (twins), IV (orthodontically treated twins), and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal maxillary (B) digital cast files (DCFs) as nonequal and specificity of 70% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 80% - for classifying equal mandibular (D) DCFs as non-equal and specificity of 73.3% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 87% and 78%, for maxillary (B) and mandibular (D) DCFs, respectively.
performed in 2D, using flatbed scanners. The 3D technology was used four years later, revealing a polemic finding, namely the lack of the UHD.9e11 Despite observing the lack of the UHD with sound statistical approach, these studies presented discussable methodology.12 The comparisons of dentitions relied on manually landmarking the incisal edges of the anterior teeth.9e11 The main restriction in this procedure is founded on operator-dependency, which comprehends the bias inherent to manually positioning landmarks in each examined dental cast. In the literature this bias was improved testing statistically operator reproducibility,8,14 and implementing it in equations as criteria for determining match or mismatch between dentitions.9 Moreover, the comparisons between dentitions were performed using Geometric Morphometrics e a technique that registers manually positioned landmarks as Cartesian coordinates31 and compares their spatial relation31 pairwise between dentitions. According to each study, a various number of landmarks were used, namely 148, 2414, and 3016. The location and number of these landmarks were chosen specially on the incisal edges of the frontal teeth. The use of a higher number of landmarks is observed in more recent studies, in an attempt to register more morphological information from the contour of the incisal edges of the dentitions.9 In the present study GS was used e a 3D modeling software package that enables the automatic morphological match by superimposing DCFs and searching for
their best fitting overlap. Similarly to Geometric Morphometrics, this software considers the 3D DCFs as sets of points that can be compared mutually. Differently from the literature, the comparison is not founded on manually positioned landmarks and the number of points in each 3D DCF sets normally exceeds 1.000 per six anterior teeth. Consequently, more morphological information was assessed in an operator-independent procedure. Class characteristics were used to describe the standard evidences found in human bitemarks on human skin. They are observed as two oval semi-arches facing each other with rectangular and circular contusions or indentations, remaining after the biting contact of incisors and canines with the skin.7 However, the literature has no consensus on the exact quantity of dental material that is involved in bitemarks, but studies with bite impressions on foodstuff6 and wax17 suggest that the incisal edges may be registered in these bitten surfaces from 1 to 3 mm. In the living, the quantity of dental material involved depends not only on the biting force applied, but also on the visco-elastic properties of the skin. A human bite may reach power of 37.72kg32, depending on the size of the muscle, the bone morphology and the articulation between the maxilla and the mandible. On the other hand, the elasticity of the skin decreases considerably25 from a subjected power of 13.5 kg and responds to skin rupture with higher biting forces. However, the human skin responds heterogeneously, depending on the
62
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
Fig. 4. Euclidean distances of the Groups I and V obtained pair-wise comparing DCFs with remaining tooth part of interest of 2 mm from the incisal edges (study part 2), expressed in Boxplots (A, C) and ROC curves (B, C). A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects) and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal maxillary (B) digital cast files (DCFs) as non-equal and specificity of 100% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal mandibular (D) DCFs as non-equal and specificity of 100% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 100% for maxillary (B) and mandibular (D) DCFs.
underlying tissue,25 the age, and the weight of the victim.26 This information justifies the lack of literature consensus on the quantity of dental material involved in bitemarks (especially those inflicted) on human skin. Therefore, in the present study a systematic selection of the 3D quantity of remaining tooth parts of interest was established (parts 1, 2 and 3). Moreover, study part 4 was designed to mimic the currently overall established bitemark practice, which relies on the 2D registration of the bitemark, registering and comparing the contours of the bitemark and transparent overlays of the suspect's dental casts.19 In this part, bitemark contours were mimicked by selecting a thin slice (1 mm) of the DCFs cropped axially at maximally 3 mm from the incisal tooth edges. Specifically, the slice was selected at this region because theoretically it includes the bitemark contours that could be generated after impressing the incisal edges for 1 and 2 mm on the skin. Similarly, the literature describes the production of 2D comparison overlays from 3D dental casts.33 This procedure was enabled by a software package with tools specifically developed for slicing digital models and improving the 2D analysis in the bitemark practice.17 The authors matched the obtained 3D dental overlays digitally with photographs of bitemarks on pig skin.17 However, the 3D dental overlays were generated from the impression of dental casts on wax,17 creating an indirect procedure. In this context, the present study has the advantage of generating overlays (slices) directly from the 3D DCF, tackling the potential distortion inherent to indirect procedures.
The present study differed from earlier published studies9e11,13,16,34,35 using not exclusively DCFs from randomly-selected subjects, but also from three populations stratified on dental similarity. First, orthodontically treated subjects were selected because they present a “lower level of individuality”14 due to similar dental alignment. Second, twin pairs were studied because, theoretically, they have identical dentitions, due to a potential genetic control over dental morphology.36 However, it is currently known that morphological dental variations exist even among monozygotics,36e38 resulting in similar but not identical dental morphology. Third, a population of orthodontically treated monozygotic twins was examined. In this twin group, the acquired changes in the phenotype expression of the dental arrangement were orthodontically corrected. This stratification was probable the main reason contributing to the lack of evidence supporting UHD in the present study. Moreover the additional stratification, selecting only the five pairs of DCF with the smallest pair-wise Euclidean differences in each studied group, increased the degree of morphological similarity in the studied samples. Further on, another aspect contributing to the increase in morphological similarity between dentitions and consequent to the lack of evidence of UHD was the reduction in the quantity of dental material studied. The high level of sample stratification on dental similarity is reflected in the sensitivity and specificity obtained comparing each group with its corresponding threshold (Group V). A perfect classification of non-equal (sensitivity: 100%) and equal (specificity:
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
63
Fig. 5. Euclidean distances of the Groups I and V obtained pair-wise comparing DCFs with remaining tooth part of interest of 1 mm from the incisal edges (study part 3), expressed in Boxplots (A, C) and ROC curves (B, C) A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects) and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal maxillary (B) digital cast files (DCFs) as non-equal and specificity of 100% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 100% - for classifying non-equal mandibular (D) DCFs as non-equal and specificity of 100% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 100% for maxillary (B) and mandibular (D) DCFs.
100%) DCFs was obtained in Group I (randomly-selected subjects). It reveals that even retaining the 5 most similar dentitions (minimal Euclidean distances), Group I presented a distinguishable discrepant dental morphology related to its respective threshold. In relation to the other groups, the sensitivity and specificity of group V decreased up to 80% and 70%, respectively. However, it maintained an acceptable discriminative accuracy for the threshold expressed in AUC values above 71%. The only exception was observed in the 2D bitemark mimicking registration (part 4), in which Group V presented a specificity of 55% and 67% for maxillary and mandibular DCFs, respectively. This may be explained by the influence of the limited quantity of tooth material considered in study part 4. Although in Group I no equal DCFs were found, even on bite mark level, no UHD may be claimed, because a randomly chosen individual can potentially be part of one of the stratified subgroups in which the evidence indicated that the UHD is not supported. In the present study, analysis of maxillary and mandibular dentitions resulted in similar outcomes in regard to the UHD. The findings suggest that no predilection for dental arches should be made in bitemark cases, indicating that all the available dental information should be analyzed.3 On the other hand, the present research was limited by the lack of mandibular DCFs from orthodontically treated subjects, encouraging the use of this material in further studies in the field. Another limitation was the possible error in the procedure of manually taking dental impressions and
creating dental casts in plaster (operator dependency). This limitation was addressed by incorporating the potential error during the manual procedure to take dental casts in the threshold. However, future research could be performed replacing this manual procedure by intra-oral scanning. Moreover, the present research was designed in the context of bitemarks without analyzing the bitemarks themselves, but investigating fundamentally uniqueness from analyzing dentitions. The analysis of bitemarks could be considered a following process towards the validation of the present findings within further studies of bitemarks on skin, foodstuff and objects. Evidence that the dentitions of subjects are indistinguishable using the techniques described indicates that a single bitemark could have been made by more than one individual. In the present study, the comparison between dentitions was performed using dental casts scanned as DCFs and analyzed by selecting as the area of interest the incisal edges of the anterior teeth. The quantity of dental material analyzed was diminished in each study step from 3 mm to 1 mm from the incisal edges based on the parts of the teeth assumed to be potentially involved in bitemarks on human skin. The dental material registered in a bitemark may reflect different morphological traits of the human dentitions, such as the size, shape, arrangement and position of the teeth in the arch.22 All of these morphological traits are found in the DCFs cropped systematically in this study. Specifically, the shape, size, arrangement and position of the incisal edges of the anterior teeth (portion of dental
64
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
Fig. 6. Euclidean distances of the Groups I, II, III, IV and V obtained pair-wise comparing DCFs with remaining tooth part of interest a slice of 1 mm not including the incisal edges (study part 4), expressed in Boxplots (A, C) and ROC curves (B, C). A, C: Boxplots showing the difference in Euclidean distances between the study Groups I (randomly-selected subjects), II (orthodontically treated subjects), III (twins), IV (orthodontically treated twins), and V (threshold) combining the four quantification values (maximum positive deviation, maximum negative deviation, average, standard deviation). B: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 80% - for classifying non-equal maxillary (B) digital cast files (DCFs) as nonequal and specificity of 55% - for classifying equal maxillary (B) DCFs as equal (specificity). D: The maximized potential prediction (grey circle) indicates that Group V reached sensitivity of 80% - for classifying non-equal mandibular (D) DCFs as non-equal and specificity of 66.7% - for classifying equal mandibular (D) DCFs as equal (specificity). The Area Under the Curve (AUC) reached accuracy of 71% and 76%, for maxillary (B) and mandibular (D) DCFs, respectively.
crowns often involved in bitemarks) were investigated in study parts 1e4. Especially in study part 4 (slice of the anterior dentition), these morphological traits are brought to a more practical level, in which the parts of the teeth would not be completely visible but their contours would be shown e simulating a bitemark overlay tracing. So, the process to gradually decrease the quantity of dental material is significant to investigate whether or not the human dentition can create unique bite patterns on skin. However, the reduction of dental material in the present research was based on a specific bitemark pattern assumption, in which only the marks of the incisal edges of the six upper and lower anterior teeth were registered on human skin e while the literature indicates that other tooth parts may be registered in bitemarks7 such as the palatal surfaces of the anterior teeth. Further studies in the field are needed for the investigation of the UHD within the palatal surfaces of the anterior teeth. Other patterns may include the registration of premolars, palatal or lingual surfaces of anterior teeth, interrupted or uninterrupted arches, abrasions, contusions, as well as partial marks or multiple overlapping marks of dental arches.7 These variations in bitemark patterns not addressed in this research should be interpreted as limitations to be improved. Other studies in the field are necessary to verify exclusively whether dentitions are unique considering dental arches with more teeth and different tooth parts (including premolars and palatal or lingual surfaces of the anterior dentition). Preliminary evidence is observed in the literature39 indicating that the increase in the amount of dental
information tends to increase the detectable morphological differences between dentitions leading to potential uniqueness. In specific, the inclusion of first and second premolars increased the morphological difference between dentition without statistical significance (p > 0.05 for the maxilla and the mandible), but with a significant clinical impact. However, assumptions concerning uniqueness are feasible only with a corresponding threshold group for comparison. Additionally, computed tomography also may be investigated as a mean for analyzing in detail slices taken from the anterior dentition (<1 mm from the incisal edges) and their uniqueness. This technology enables an accurate analysis of the dental morphology with higher magnification than used in the present research. Thus, it would benefit not only the investigations on the uniqueness of the human dentition but also the research investigating the biting phenomenon itself. Considering the specific assumed pattern of a bitemark studied, this chapter demonstrated how highly similar the dentitions from different subjects can be when their teeth are well aligned (Group II) or when their teeth present similar morphology (Group III). These results indicate that forensic BM analysis must not be disregarded, but performed after case selection. This was also advised in the literature in the sense that “not every case is suitable for analysis”.40 The present chapter suggests that BM cases involving suspects with similar dental arrangement and morphology should be interpreted carefully. Further studies should be conducted exploring the limits of case selection in BM analysis. In fact, the case
A. Franco et al. / Journal of Forensic and Legal Medicine 46 (2017) 58e65
selection should be based on distinctive individual tooth or arch characteristics to allow decisions based on the exclusion or nonexclusion of suspected biters associated with legal issues. On the other hand, uniqueness remains a fundamental question in the quest to assure that no individual could be falsely associated or linked to a bitemark. Finding a distinctive trait to exclude individuals is a difficult task when the suspected biters have well aligned teeth or similar dental morphology. It justifies the present chapter and confirms the importance of the UHD for legally undisputable bitemark scenarios. 5. Conclusion Scientific evidence for the UHD was not detected in the current research. Pairs of dentitions with morphological differences smaller than their respective measuring threshold were observed. This was related to the high level of sample stratification used. The included subjects had a high degree of similar dental alignment (orthodontically treated) and morphology (twins). Moreover, the standard within class characteristics of human bitemarks on human skin were mimicked, reducing the quantity of dental material analyzed in all study parts. Because the mimicked standard of bitemark patterns strongly vary in the forensic practice, further research increasing or even reducing the quantity of dental material analyzed is necessary. The current findings indicate that in forensic practice, investigations of bitemarks with the mimicked standards can only be performed in closed populations. Moreover, case selection should be based on distinctive individual tooth or dentition characteristics to allow decisions based on the exclusion or inclusion of suspected biters associated with legal issues. Conflict of interest None. Acknowledgements Author AF would like to express gratitude to the Coordination for the Improvement of Higher Education Personnel (CAPES) (Protocol number: 99999.003423/2015-08) for funding the present research. References 1. Lucas P. Dental Functional Morphology: How Teeth Work. Cambridge: Cambridge University Press; 2004. 2. Agrawal KR, Lucas PW. The mechanics of the first bite. Proc Biol Sci. 2003;270: 1277e1282. http://dx.doi.org/10.1098/rspb.2003.2361. 3. Dorion RB. Bite mark evidence. J Can Dent Assoc. 1982;48:795e798. 4. Nambiar P. Forensic odontology - bite marks - a review of the literature (part 1). Dent J Malays. 1994;14:32e36. 5. Whittaker D, McDonald D. An Atlas of Forensic Odontology. London: Wolfe Medical Publications; 1989. 6. Webster G. A suggested classification of bite marks in foodstuffs in forensic dental analysis. Forensic Sci Int. 1982;20:45e52. 7. Dorion R. Bite Mark Evidence: A Color Atlas and Text. second ed. Boca Raton: CRC Press; 2011. 8. Sheets HD, Bush PJ, Brzozowski C, Nawrocki LA, Ho P, Bush MA. Dental shape match rates in selected and orthodontically treated populations in New York State: a two-dimensional study. J Forensic Sci. 2011;56:621e626. http:// dx.doi.org/10.1111/j.1556-4029.2011.01731.x. 9. Sheets HD, Bush PJ, Bush MA. Patterns of variation and match rates of the anterior biting dentition: characteristics of a database of 3d-scanned dentitions. J Forensic Sci. 2013;58:60e68. http://dx.doi.org/10.1111/j.15564029.2012.02293.x. 10. Bush MA, Bush PJ, Sheets HD. Statistical evidence for the similarity of the human dentition. J Forensic Sci. 2011;56:118e123. http://dx.doi.org/10.1111/ j.1556-4029.2010.01531.x.
65
11. Bush MA, Bush PJ, Sheets HD. Similarity and match rates of the human dentition in three dimensions: relevance to bitemark analysis. Int J Leg Med. 2011;125:779e784. http://dx.doi.org/10.1007/s00414-010-0507-8. 12. Franco A, Willems G, Souza PHC, Bekkering GE, Thevissen P. The uniqueness of the human dentition as forensic evidence: a systematic review on the technological methodology. Int J Leg Med. 2015;129:1277e1283. http://dx.doi.org/ 10.1007/s00414-014-1109-7. 13. Rawson RD, Ommen RK, Kinard G, Johnson J, Yfantis A. Statistical evidence for the individuality of the human dentition. J Forensic Sci. 1984;29:245e253. 14. Kieser JA, Bernal V, Neil Waddell J, Raju S. The uniqueness of the human anterior dentition: a geometric morphometric analysis. J Forensic Sci. 2007;52: 671e677. http://dx.doi.org/10.1111/j.1556-4029.2007.00403.x. 15. Sognnaes RF, Rawson RD, Gratt BM, Nguyen NB. Computer comparison of bitemark patterns in identical twins. J Am Dent Assoc. 1982;105:449e451. 16. Blackwell SA, Taylor RV, Gordon I, et al. 3-D imaging and quantitative comparison of human dentitions and simulated bite marks. Int J Leg Med. 2007;121: 9e17. http://dx.doi.org/10.1007/s00414-005-0058-6. 17. Martin-de-las-Heras S, Tafur D. Comparison of simulated human dermal bitemarks possessing three-dimensional attributes to suspected biters using a proprietary three-dimensional comparison. Forensic Sci Int. 2009;190:33e37. http://dx.doi.org/10.1016/j.forsciint.2009.05.007. 18. Martin-de-Las-Heras S, Tafur D, Bravo M. A quantitative method for comparing human dentition with tooth marks using three-dimensional technology and geometric morphometric analysis. Acta Odontol Scand. 2014;72:331e336. http://dx.doi.org/10.3109/00016357.2013.826383. 19. Bowers C, Johansen R. Digital Analysis of Bite Mark Evidence. Indianapolis: Forensic Imaging Inst; 2000. 20. Casella G. Statistical Design. New York: Springer; 2008. 21. Tukey J. Comparing individual means in the analysis of variance. Biometrics. 1949;5:99e114. 22. Senn DR, Weems M. Manual of Forensic Odontology. fifth ed. CRC Press; 2013. 23. Committee on Identifying the Needs of the Forensic Sciences Community NRC. Strengthening Forensic Science in the United States: A Path Forward. 2009. 24. Tuceryan M, Li F, Blitzer HL, Parks ET, Platt JA. A framework for estimating probability of a match in forensic bite mark identification. J Forensic Sci. 2011;56. http://dx.doi.org/10.1111/j.1556-4029.2010.01571.x. 25. Bush MA, Thorsrud K, Miller RG, Dorion RBJ, Bush PJ. The response of skin to applied stress: investigation of bitemark distortion in a cadaver model. J Forensic Sci. 2010;55:71e76. http://dx.doi.org/10.1111/j.15564029.2009.01235.x. 26. Bush MA, Miller RG, Bush PJ, Dorion RBJ. Biomechanical factors in human dermal bitemarks in a cadaver model. J Forensic Sci. 2009;54:167e176. http:// dx.doi.org/10.1111/j.1556-4029.2008.00908.x. 27. Miller RG, Bush PJ, Dorion RBJ, Bush MA. Uniqueness of the dentition as impressed in human skin: a cadaver model. J Forensic Sci. 2009;54:909e914. http://dx.doi.org/10.1111/j.1556-4029.2009.01076.x. 28. Sheets HD, Bush PJ, Bush MA. Bitemarks: distortion and covariation of the maxillary and mandibular dentition as impressed in human skin. Forensic Sci Int. 2012;223:202e207. http://dx.doi.org/10.1016/j.forsciint.2012.08.044. 29. Bowers CM, Pretty IA. Expert disagreement in bitemark casework. J Forensic Sci. 2009;54:915e918. http://dx.doi.org/10.1111/j.1556-4029.2009.01073.x. 30. Page M, Taylor J, Blenkin M. Expert interpretation of bitemark injuriesea contemporary qualitative study. J Forensic Sci. 2013;58:664e672. http:// dx.doi.org/10.1111/1556-4029.12108. 31. Webster M, Sheets H. A practical introduction to landmark-based geometric morphometrics. Paleontol Soc Pap. 2010;16:163e188. 32. Paphangkorakit J, Osborn JW. The effect of pressure on a maximum incisal bite force in man. Arch Oral Biol. 1997;42:11e17. 33. Las Heras S, Valenzuela A, Ogayar C, Valverde J, Torres J. Computer-based production of comparison overlays from 3D-scanned dental casts for bite mark analysis. J Forensic Sci. 2005;50:127e133. 34. Nambiar P, Bridges TE, Brown KA. Quantitative forensic evaluation of bite marks with the aid of a shape analysis computer program: Part 1; the development of “SCIP” and the similarity index. J Forensic Odontostomatol. 1995;13: 18e25. 35. Nambiar P, Bridges TE, Brown KA. Quantitative forensic evaluation of bite marks with the aid of a shape analysis computer program: Part 2; “SCIP” and bite marks in skin and foodstuffs. J Forensic Odontostomatol. 1995;13:26e32. 36. Osborne R, Horowitz S, De George F. Genetic variation in tooth dimensions: a twin study of the permanent anterior teeth. Physiol Res. 1958;10:350e356. 37. Lundstrom A. Tooth morphology as a basis for distinguishing monozygotic and dizygotic twins. Am J Hum Genet. 1963;15:34e43. 38. Townsend G, Richards L, Brown T, Burgess V. Twin zygosity determination on the basis of dental morphology. J Forensic Odontostomatol. 1988;6:1e15. 39. Franco A, Willems G, Couto Souza PH, Coucke W, Thevissen P. Three-dimensional validation of the impact of the quantity of teeth or tooth parts on the morphological difference between twin dentitions. J Forensic Odontostomatol. 2016;1:27e37. 40. Barsley R, Freeman A, Metcalf R, Senn D, Wright F. Bitemark analysis. J Am Dent Assoc. 2012;143:444e446. author reply 446, 448.