The positive identification of a decedent is paramount to a forensic investigation in which human remains have been recovered and must be identified. Due to increasing global mobility in the world's populations, it is not inconceivable that an individual might die far away from his or her home. Pinpointing an individual's geographic origin may contribute to definitive forensic identification of contemporary human skeletal remains in cases where dental records and other means of identification are either unavailable to law enforcement personnel or do not yield immediate results. Stable isotope analysis of biogenic tissues such as tooth enamel and bone mineral has become a well‐recognized and increasingly important method for determining the provenance of human remains, and it has been used successfully in bioarchaeological studies as well as forensic investigations. Both 18O and 2H stable isotope signatures are well established proxies as environmental indicators of climate (temperature) and source water and are therefore considered reliable indicators of geographic life trajectories of animals and humans. Similarly, 13C and 15N abundance data have distinguished dietary preferences in ancient human populations, and have been used to qualify 2H and 18O geolocational data that may be consistent with more than one location. Few if any studies have systematically investigated the multi-isotopic signatures in human tooth enamel and dentin from living individuals. Since 18O abundance values obtained from tooth enamel of late-erupting molars are a source of information on geographic origin of an individual during adolescence when crown formation takes place, it was hypothesized that: 1) the stable isotope abundance of 2H, 18O, and 13C in human tooth enamel and dentin is consistent with self-reported residential history and dietary preferences data, and 2) the isotopic variability evident between individuals with shared residential history and nutrient intake is quantifiable, and indicative of intra-individual variability. Two pilot studies were conducted to: 1) evaluate the feasibility of extracting and measuring the 2H composition of human tooth enamel and its suitability as a proxy for human geographic provenance, and 2) compare the isotopic abundance of isotopic abundance of 2H and 13C in human crown dentin collagen obtained from archaeological and modern teeth with the 18O and 13C isotopic composition of the corresponding tooth enamel carbonate. A protocol for preparing tooth enamel for 2H analysis was successfully devised and implemented; however, no correlation was observed between tooth enamel 2H abundance values as measured plotted against source water 2H abundance values. While unexpectedly low rates of hydrogen exchange within mineral hydroxyl groups were documented, and it was possible to analyze tooth enamel for its 2H isotopic composition, the seemingly fixed nature of 2H abundance in tooth enamel means that the 2H isotopic signature of tooth enamel cannot be used as an indicator of geographic provenance. Conversely, positive correlations between collagen 2H abundance values of primary dentin with 2H abundance values for source water and also with enamel 18O abundance values suggests that primary dentin collagen 2H values are linked to the isotopic composition of source water. Third molar tooth enamel was sampled from 10 living volunteers undergoing routine tooth extractions at University of Toronto affiliated dental clinics in Ontario, Canada. The mixed-sex group of patients was given questionnaires in which they provided detailed residential history and answered questions pertaining to dietary preferences (e.g., vegetarian) prior to donating all four third molars. Enamel was drilled from the crown of two third molars from each subject, chemically cleaned, and subjected to an acid digest before being analyzed for its 18O and 13C composition using Isotope Ratio Mass Spectrometry. Herewith, isotope abundance values are presented using the delta notation as delta values in per mil (‰). Mean d13CVPDB values for all samples ranged from -9.47 ‰ to -11.31 ‰ (pooled mean = -10.37 ‰), which suggested a persistent C4 plant dietary influence at the time the sampled tooth enamel was forming and is consistent with the typical North American diet. While inter-subject variation contributed the largest proportion of total d13C variability, differences were not significant. The pooled mean d18OVSMOW value for enamel samples was 24.39 ‰, while individual subject mean d18OVSMOW values ranged from 23.76 ‰ to 25.18 ‰. Marked offsets (0.01 ‰ - 0.51 ‰) in mean d18OVSMOW values for each pair of third molars were observed. While subject variation was significant (p=0.0034), neither diet nor sex significantly influenced the oxygen (or carbon) isotope data. Following conversion of d18OVSMOW values to d18OPhosphate values, drinking water values were calculated using the Daux et al. (2008) equation, and compared to their corresponding regional estimated annual average d18O values in precipitation retrieved from the Online Isotopes in Precipitation Calculator (OIPC) (Bowen 2014). Despite observed correlations between the drinking water d18OWater values and d18OOIPC values of four subjects, no statistically significant correlations were evident between the two limited data sets. However, when the isotopically similar d18O values of Toronto area residents were averaged and combined with the remaining Canadian data and the 18O data from 5 enamel samples analyzed in the 2H enamel pilot study, strong positive correlations were evident between d18OPhosphate valuesand d18OOIPC values (R2 = 0.87). Moreover, an equally strong linear relationship was observed between modeled annual precipitation 18O and calculated source water 18O (R2 = 0.87). Based on the d2H, d18O, and d13C values measured in tooth enamel and dentin, it is possible to infer and confirm geographic provenance and dietary intake. However, while enamel carbonate d13C values were consistent with self-reported dietary intake information and residential history, as could be expected vegetarians could not be distinguished from those who consumed meat without determining C/N isotopic ratios in dentin collagen. Further, the presence of strong linear relationships between the 18O composition of enamel carbonate and modeled source water 18O in the merged dataset illustrates the importance of considering site-specific isotopic complexities and using multi-isotope data obtained from multiple tissues when investigating the geographic origins of humans in an archaeological or forensic context. It is not possible to quantify intra-individual isotopic variability without sampling from larger, geographically diverse populations and controlling for as many variables as possible. The construction of regional databases containing 18O and 2H isotopic data obtained from a variety of environmental and human and faunal tissue samples, and the application of such data to individual cases in which geographic origins are desired, is recommended provided the following caveats are considered: 1) whenever possible, d18OPhosphate values should be used when comparing d18O values in tooth enamel with those in precipitation in order to avoid the unmitigated error associated with the application of carbonate conversion equations to enamel phosphate d18O data, 2) the continuous consumption of food sourced elsewhere is liable to affect the overall enamel carbonate d18O values, 3) d18O values in precipitation are not inclusive of the variety of postprecipitation and hydrological processes unique to a particular location, nor are they an accurate representation of the isotopically mixed nature of tap water sourced from distant reservoirs. Hence, extrapolations should be made with caution. Enamel sequential microsampling methods may be advisable for narrowing down a geographic timeline if the overlapping of isotopic signals at various points along the tooth’s longitudinal plane can be quantified and validated. While standardization of analytical methodology is critical to appropriate interpretations of the data, stable isotope profiling is not a standalone method and should be used in conjunction with other lines of evidence in determinations of human provenance.

Made available in DSpace on 2014-06-11T19:29:40Z (GMT). No. of bitstreams: 0 Previous issue date: 2011-11-25Bitstream added on 2014-06-13T18:59:21Z : No. of bitstreams: 1 barboza_eb_me_sjrp.pdf: 1261122 bytes, checksum: 0b2e650e6acac4f6e02e1fdae512dccc (MD5) / Na sociedade atual, a identificação exata e rápida dos indivíduos é uma necessidade. Nas aplicações forenses, a identificação por meio de características biométricas é bastante usual e muitas vezes é a única alternativa. A arcada dentária é uma das mais importantes e populares características biométricas utilizadas pela Odontologia Legal na área forense. O principal objetivo da Odontologia Legal é identificar indivíduos falecidos para os quais os outros meios de identificação biométrica (impressões digitais, faces, etc.) não são possíveis de serem aplicados. Em geral, o perito humano realiza a comparação manual entre registros dentários antemortem e postmortem anotando as diferenças encontradas em cada dente, o que demanda muito tempo e torna essa prática suscetível a erros, daí a importância do desenvolvimento de sistemas automáticos ou semiautomáticos para identificação humana forense. O objetivo dessa dissertação de mestrado é avaliar métodos baseados na Transformada Imagem Floresta para segmentação semiautomática de dentes e suas restaurações em imagens de radiografias panorâmicas de arcadas dentárias, utilizando os contornos obtidos para definir descritores biométricos para a identificação humana forense. Os descritores de formas baseados nos métodos Contexto da Forma (Shape Context) e Estatística dos Ângulos de Raios (BAS – Beam Angle Statistics) foram implementados e avaliados para o reconhecimento dos dentes. A técnica da Distância de Edição avaliou o reconhecimento das restaurações baseada nos códigos dentais gerados. Resultados experimentais obtidos sobre uma base de 40 imagens de radiografias de arcadas dentárias, contendo no total 1126 imagens de dentes, mostram que o método de segmentação baseado na Transformada Imagem Floresta Diferencial... / Nowadays, the fast and accurate individual’s identification is a great necessity. In forensic applications, identification through biometric features is often usual and in some cases is the only alternative. The dental arch is one of the most important and popular biometric characteristic used in the dental forensic field. The primary goal of forensic dentistry is to identify deceased individuals for whom other forms of biometric identification (fingerprints, faces, etc.) are not able to be applied. In general, the human expert performs the manual comparison between antemortem and postmortem dental records noting the differences found in each tooth, which requires much time and practice, but this manual practice is more susceptible to errors. Hence, the importance of the development of automatic or semiautomatic forensic human identification systems. The goal of this Master dissertation is to evaluate methods based on Image Forest Transform for semiautomatic segmentation of teeth and their restorations in panoramic radiographs images of dental arches, using the contours obtained to define biometric descriptors for forensic human identification. The shape descriptors based on the Shape Context method and Beam Angle Statistics (BAS) were implemented and evaluated for the teeth recognition. The Edition Distance technique was used to evaluate the restorations and all dental work through the generated dental codes. Experimental results obtained on a database of 40 radiographs images, containing a total of 1126 teeth images, show that the segmentation method based on Differential Image Foresting Transform presented good and promising results for the semiautomatic teeth segmentation. They also showed that the Shape Context and BAS descriptors had similar results, with a slight advantage for the... (Complete abstract click electronic access below)

Orientador: Aparecido Nilceu Marana / Banca: Evandro Luís Linhari Rodrigues / Banca: Izabel Regina F. R. de Bullen / Resumo: Na sociedade atual, a identificação exata e rápida dos indivíduos é uma necessidade. Nas aplicações forenses, a identificação por meio de características biométricas é bastante usual e muitas vezes é a única alternativa. A arcada dentária é uma das mais importantes e populares características biométricas utilizadas pela Odontologia Legal na área forense. O principal objetivo da Odontologia Legal é identificar indivíduos falecidos para os quais os outros meios de identificação biométrica (impressões digitais, faces, etc.) não são possíveis de serem aplicados. Em geral, o perito humano realiza a comparação manual entre registros dentários antemortem e postmortem anotando as diferenças encontradas em cada dente, o que demanda muito tempo e torna essa prática suscetível a erros, daí a importância do desenvolvimento de sistemas automáticos ou semiautomáticos para identificação humana forense. O objetivo dessa dissertação de mestrado é avaliar métodos baseados na Transformada Imagem Floresta para segmentação semiautomática de dentes e suas restaurações em imagens de radiografias panorâmicas de arcadas dentárias, utilizando os contornos obtidos para definir descritores biométricos para a identificação humana forense. Os descritores de formas baseados nos métodos Contexto da Forma (Shape Context) e Estatística dos Ângulos de Raios (BAS - Beam Angle Statistics) foram implementados e avaliados para o reconhecimento dos dentes. A técnica da Distância de Edição avaliou o reconhecimento das restaurações baseada nos códigos dentais gerados. Resultados experimentais obtidos sobre uma base de 40 imagens de radiografias de arcadas dentárias, contendo no total 1126 imagens de dentes, mostram que o método de segmentação baseado na Transformada Imagem Floresta Diferencial... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Nowadays, the fast and accurate individual's identification is a great necessity. In forensic applications, identification through biometric features is often usual and in some cases is the only alternative. The dental arch is one of the most important and popular biometric characteristic used in the dental forensic field. The primary goal of forensic dentistry is to identify deceased individuals for whom other forms of biometric identification (fingerprints, faces, etc.) are not able to be applied. In general, the human expert performs the manual comparison between antemortem and postmortem dental records noting the differences found in each tooth, which requires much time and practice, but this manual practice is more susceptible to errors. Hence, the importance of the development of automatic or semiautomatic forensic human identification systems. The goal of this Master dissertation is to evaluate methods based on Image Forest Transform for semiautomatic segmentation of teeth and their restorations in panoramic radiographs images of dental arches, using the contours obtained to define biometric descriptors for forensic human identification. The shape descriptors based on the Shape Context method and Beam Angle Statistics (BAS) were implemented and evaluated for the teeth recognition. The Edition Distance technique was used to evaluate the restorations and all dental work through the generated dental codes. Experimental results obtained on a database of 40 radiographs images, containing a total of 1126 teeth images, show that the segmentation method based on Differential Image Foresting Transform presented good and promising results for the semiautomatic teeth segmentation. They also showed that the Shape Context and BAS descriptors had similar results, with a slight advantage for the... (Complete abstract click electronic access below) / Mestre

A aproximação fisionômica é o método que busca, a partir do crânio, simular a fotografia de um indivíduo quando em vida. Deve ser empregada como último recurso, na busca de desaparecidos, quando não houver possibilidade de aplicação de um método válido de identificação. O objetivo deste estudo foi obter a aproximação fisionômica, a partir de um crânio seco e de tomografia computadorizada multislice de indivíduos vivos, através da função de base radial hermitiana (FBRH). Constituiu-se também em avaliar o resultado da mesma quanto ao reconhecimento. Na primeira etapa do estudo, foi utilizada a imagem escaneada de um crânio seco, de origem desconhecida, com o intuito de avaliar se a quantidade de pontos obtidos seria suficiente para aplicação da FBRH e consequente reconstrução da superfície facial. Na segunda fase, foram utilizadas três tomografias de indivíduos vivos, para análise da semelhança alcançada entre a face escaneada e as aproximações faciais. Nesta etapa, foi aplicada uma associação de diferentes metodologias já publicadas, para reconstrução de uma mesma região da face, a partir de um mesmo crânio. Na última etapa, foram simuladas situações de reconhecimento com familiares e amigos dos indivíduos doadores das tomografias. Observou-se que a metodologia de FBRH pode ser empregada em aproximação fisionômica. Houve reconhecimento positivo nos três sujeitos estudados, sendo que, em dois deles, os resultados foram ainda mais significativos. Desta forma, conclui-se que a metodologia é rápida, objetiva e proporciona o reconhecimento. Esta permite a criação de múltiplas versões de aproximações fisionômicas a partir do mesmo crânio, o que amplia as possibilidades de reconhecimento. Observou-se ainda que a técnica não exige habilidade artística do profissional. / Facial approximation works by building the visual face up from the skull. This method should be performed as last resort, to carry out for missing persons, when there is no other primary identification method avaliable. The purpose of this study was to introduce a new computerized method with hermite radial basis function (HRBF) for facial approximation using dry skull and computed tomography (CT). The same was also evaluated as a result of the recognition. Firstly, a scan of a dry unidentified skull image was used in order to assess if the amount of points would be sufficient for HRBF methodology and subsequent reconstruction of the facial surface. In second, three CT scans of living individuals were used to evaluate the similarity achieved between the real face scanned and facial approximations. An association of different facial structures reconstruction techniques already published for the same region of the face was applied for the same skull. Moreover, some situations from developed facial approximations were simulated, as recognition by a relative or parent, on a face pool-test. Results from the study showed that the purposed methodology can be used for facial approximation. At the three cases a correct approximation identification as one of a few possible matches to the missing person happened. In two of them, the results were consistently better at identifying the correct approximation. In conclusion, the proposed methodology is fast, objective and reaches visual identification. It is possible to perform multiple versions of the same skull, changing the selected data into the system, which maximizes the chances of establishing recognition of the target face. It was also observed that the technique does not need artistic interpretation.

Made available in DSpace on 2014-06-11T19:29:40Z (GMT). No. of bitstreams: 0 Previous issue date: 2008-06-23Bitstream added on 2014-06-13T19:18:04Z : No. of bitstreams: 1 falguera_jr_me_sjrp.pdf: 2234581 bytes, checksum: 19293ff7ecaf5caa8cf4417a59cb11fa (MD5) / Diversos métodos biométricos baseados em características físicas do corpo humano como impressão digital, face, íris e retina têm sido propostos para identificação humana. No entanto, para a identificação post-mortem, tais características biométricas podem não estar disponíveis. Nestes casos, partes do esqueleto do corpo humano podem ser utilizadas para identificação, tais como dentes, tórax, vértebras, ombros e os sinus frontais. Investigações anteriores mostraram, por meio de técnicas manuais para extração de características, que os padrões dos sinus frontais são altamente variáveis entre indivíduos distintos e únicos para cada indivíduo. Esta dissertação de mestrado tem por objetivo propor um método computacional para o reconhecimento de sinus frontais para identificação humana post-mortem em aplicações forenses. Para tanto, foram avaliados métodos de segmentação de imagens de radiografias anteroposteriores de sinus frontais. O método baseado na Transformada Imagem-Floresta demonstrou ser eficiente para segmentação dos sinus frontais das imagens de radiografias, exigindo mínima intervenção humana. Foram também investigadas e implementadas técnicas para extração de descritores geométricos e descritores baseados nas formas dos sinus frontais. Experimentos realizados em um banco de imagens contendo 90 radiografias anteroposteriores de 29 indivíduos mostraram que a técnica de extração de características baseada nos descritores de contexto da forma foi a mais eficaz, propiciando taxas de erro igual (EER) e de recuperações corretas (CRR) de 3,73% e 95,5%, respectivamente. Os resultados obtidos nos experimentos corroboram os encontrados na literatura sobre a individualidade dos sinus frontais e sua viabilidade em termos de precisão e usabilidade para a identificação humana post-mortem. Palavras-chave: Biometria, identificação... / Several methods based on Biometrics such as fingerprint, face, iris, and retina have been proposed for person identification. However, for postmortem identification such biometric measurements may not be available. In such cases, parts of the human skeleton can be used for identification, such as teeth, thorax, vertebrae, shoulders, and frontal sinus. Previous investigations showed, by means of manual features extraction techniques, that frontal sinus patterns are highly variable for distinctive individuals and unique for each one. The objective of this master thesis is to propose a computational method for frontal sinus recognition for postmortem human identification in forensic applications. In order to achieve this, methods for frontal sinus segmentation from anteroposterior radiographs were evaluated. The method based on Image-Foresting Transform has shown itself efficient in frontal sinus segmentation from radiograph images, demanding minimal human intervention. After the segmentation, techniques for extracting frontal sinus geometrical and shape-based descriptors were investigated and implemented. Experiments over a database containing 90 anteroposterior radiograph images from 29 individuals have shown that the features extraction techniques based on shape context descriptors were the most efficient, providing equal error (EER) and correct retrievals (CRR) rates of 3.73% and 95,5%, respectively. The results obtained in our experiments confirm the outcomes described in literature about the individuality of the frontal sinus and its feasibility in terms of precision and usability for postmortem human identification. Keywords: Biometrics, forensics human... (Complete abstract click electronic access below)

A aproximação fisionômica é o método que busca, a partir do crânio, simular a fotografia de um indivíduo quando em vida. Deve ser empregada como último recurso, na busca de desaparecidos, quando não houver possibilidade de aplicação de um método válido de identificação. O objetivo deste estudo foi obter a aproximação fisionômica, a partir de um crânio seco e de tomografia computadorizada multislice de indivíduos vivos, através da função de base radial hermitiana (FBRH). Constituiu-se também em avaliar o resultado da mesma quanto ao reconhecimento. Na primeira etapa do estudo, foi utilizada a imagem escaneada de um crânio seco, de origem desconhecida, com o intuito de avaliar se a quantidade de pontos obtidos seria suficiente para aplicação da FBRH e consequente reconstrução da superfície facial. Na segunda fase, foram utilizadas três tomografias de indivíduos vivos, para análise da semelhança alcançada entre a face escaneada e as aproximações faciais. Nesta etapa, foi aplicada uma associação de diferentes metodologias já publicadas, para reconstrução de uma mesma região da face, a partir de um mesmo crânio. Na última etapa, foram simuladas situações de reconhecimento com familiares e amigos dos indivíduos doadores das tomografias. Observou-se que a metodologia de FBRH pode ser empregada em aproximação fisionômica. Houve reconhecimento positivo nos três sujeitos estudados, sendo que, em dois deles, os resultados foram ainda mais significativos. Desta forma, conclui-se que a metodologia é rápida, objetiva e proporciona o reconhecimento. Esta permite a criação de múltiplas versões de aproximações fisionômicas a partir do mesmo crânio, o que amplia as possibilidades de reconhecimento. Observou-se ainda que a técnica não exige habilidade artística do profissional. / Facial approximation works by building the visual face up from the skull. This method should be performed as last resort, to carry out for missing persons, when there is no other primary identification method avaliable. The purpose of this study was to introduce a new computerized method with hermite radial basis function (HRBF) for facial approximation using dry skull and computed tomography (CT). The same was also evaluated as a result of the recognition. Firstly, a scan of a dry unidentified skull image was used in order to assess if the amount of points would be sufficient for HRBF methodology and subsequent reconstruction of the facial surface. In second, three CT scans of living individuals were used to evaluate the similarity achieved between the real face scanned and facial approximations. An association of different facial structures reconstruction techniques already published for the same region of the face was applied for the same skull. Moreover, some situations from developed facial approximations were simulated, as recognition by a relative or parent, on a face pool-test. Results from the study showed that the purposed methodology can be used for facial approximation. At the three cases a correct approximation identification as one of a few possible matches to the missing person happened. In two of them, the results were consistently better at identifying the correct approximation. In conclusion, the proposed methodology is fast, objective and reaches visual identification. It is possible to perform multiple versions of the same skull, changing the selected data into the system, which maximizes the chances of establishing recognition of the target face. It was also observed that the technique does not need artistic interpretation.