Ocular pathology is one of the main health problems worldwide. The number of people with retinopathy symptoms has increased considerably in recent years. Early adequate treatment has demonstrated to be effective to avoid the loss of the vision. The analysis of fundus images is a non intrusive option for periodical retinal screening. Different models designed for the analysis of retinal images are based on supervised methods, which require of hand labelled images and processing time as part of the training stage. On the other hand most of the methods have been designed under the basis of specific characteristics of the retinal images (e.g. field of view, resolution). This compromises its performance to a reduce group of retinal image with similar features. For these reasons an unsupervised model for the analysis of retinal image is required, a model that can work without human supervision or interaction. And that is able to perform on retinal images with different characteristics. In this research, we have worked on the development of this type of model. The system locates the eye structures (e.g. optic disc and blood vessels) as first step. Later, these structures are masked out from the retinal image in order to create a clear field to perform the lesion detection. We have selected the Graph Cut technique as a base to design the retinal structures segmentation methods. This selection allows incorporating prior knowledge to constraint the searching for the optimal segmentation. Different link weight assignments were formulated in order to attend the specific needs of the retinal structures (e.g. shape). This research project has put to work together the fields of image processing and ophthalmology to create a novel system that contribute significantly to the state of the art in medical image analysis. This new knowledge provides a new alternative to address the analysis of medical images and opens a new panorama for researchers exploring this research area.

O presente estudo teve como objetivo avaliar o impacto de um modelo 3D, sobre a formação e progressão de lesões de cárie como objeto de aprendizagem, no desempenho teórico/prático e grau de satisfação de alunos de graduação em odontologia de diferentes contextos, na detecção de lesões de cárie utilizando o ICDAS. Foi conduzido um estudo multicêntrico controlado randomizado envolvendo uma amostra por conveniência de alunos de graduação em odontologia de cinco instituições (1 nacional e 4 internacionais). Inicialmente, os alunos receberam uma aula teórica tradicional e responderam uma primeira avaliação teórica. Posteriormente, foram aleatoriamente alocados em dois grupos: 1) grupo teste: que recebeu uma aula teórica tradicional expositiva mais acesso ao modelo 3D e 2) grupo controle: que recebeu unicamente a aula teórica tradicional expositiva. Depois, os alunos do grupo controle saíram da sala e um vídeo de 6 minutos (modelo 3D) foi projetado. Após o vídeo os alunos do grupo controle regressaram à sala e ambos os grupos foram submetidos a uma avaliação teórico/prática com o propósito de avaliar o desempenho após a intervenção e grau de satisfação da atividade. Análises de regressão linear e de Poisson multinível foram realizadas para analisar o impacto do objeto de aprendizagem no desempenho teórico-prático do aluno. Análises descritivas foram realizadas para avaliar o grau de satisfação do aluno. Um total de 307 alunos participou do estudo. Alunos que tiveram melhor desempenho na avaliação teórica inicial obtiveram melhores notas na média teórica final (OR=1,11; 95%IC=1,02-1,21). Alunos do grupo teste tiveram um melhor desempenho teórico em comparação com os do grupo controle (p=0,04), principalmente para questões relacionadas à correlação histológica do ICDAS com características clínicas dos diferentes estágios de progressão. Não houve diferença estatisticamente significante na avaliação prática entre grupos e, um alto nível de satisfação da atividade foi observado na amostra. Conclui-se que, a atividade avaliada teve um impacto satisfatório no desenvolvimento de competências teóricas relacionadas à detecção de lesões de cárie utilizando o ICDAS. / This study aimed at evaluating the impact of a 3D model as a learning object in the training/teaching and satisfaction degree of undergraduate dental students from different contexts for the detection of caries lesions using the ICDAS. A multicenter controlled randomized trial was conducted, involving a convenience sample of undergraduate dental students from five institutions (1 national and 4 international). Firstly, students attended a traditional theoretical lecture and answered a first theoretical test. Then, they were randomly allocated into two groups as follows: 1) test group: receiving the theoretical lecture and accessing the 3D model, and 2) control group: receiving the theoretical lecture only. Afterwards, control group students left the room and a 6-minute video was projected (3D model). Once the video had finished, control group students returned to the room and both groups were submitted to a theoretical/practical test to evaluate their performance after intervention as well as their satisfaction degree. Multilevel linear and Poisson regression analyses were done, to analyze the learning object impact in the students´ theoretical/practical performance. Descriptive analyses were conducted to assess the students´ satisfaction degree. Three hundred and seven students participated. Those having a better performance in the initial theoretical test also had better grades in the final theoretical assessment (OR=1,11; 95%IC=1,02-1,21). Test group students had a better theoretical performance in comparison to control group ones (p=0,04), mainly in relation to questions regarding the ICDAS histological correlation with clinical features on each severity caries stage. There were no statistically significant differences regarding practical assessment between groups, and a high level of activity satisfaction was observed. In conclusion, the assessed activity had a satisfactory impact in the developing of theoretical skills in relation to the detection of caries lesions using the ICDAS.

O presente estudo teve como objetivo avaliar o impacto de um modelo 3D, sobre a formação e progressão de lesões de cárie como objeto de aprendizagem, no desempenho teórico/prático e grau de satisfação de alunos de graduação em odontologia de diferentes contextos, na detecção de lesões de cárie utilizando o ICDAS. Foi conduzido um estudo multicêntrico controlado randomizado envolvendo uma amostra por conveniência de alunos de graduação em odontologia de cinco instituições (1 nacional e 4 internacionais). Inicialmente, os alunos receberam uma aula teórica tradicional e responderam uma primeira avaliação teórica. Posteriormente, foram aleatoriamente alocados em dois grupos: 1) grupo teste: que recebeu uma aula teórica tradicional expositiva mais acesso ao modelo 3D e 2) grupo controle: que recebeu unicamente a aula teórica tradicional expositiva. Depois, os alunos do grupo controle saíram da sala e um vídeo de 6 minutos (modelo 3D) foi projetado. Após o vídeo os alunos do grupo controle regressaram à sala e ambos os grupos foram submetidos a uma avaliação teórico/prática com o propósito de avaliar o desempenho após a intervenção e grau de satisfação da atividade. Análises de regressão linear e de Poisson multinível foram realizadas para analisar o impacto do objeto de aprendizagem no desempenho teórico-prático do aluno. Análises descritivas foram realizadas para avaliar o grau de satisfação do aluno. Um total de 307 alunos participou do estudo. Alunos que tiveram melhor desempenho na avaliação teórica inicial obtiveram melhores notas na média teórica final (OR=1,11; 95%IC=1,02-1,21). Alunos do grupo teste tiveram um melhor desempenho teórico em comparação com os do grupo controle (p=0,04), principalmente para questões relacionadas à correlação histológica do ICDAS com características clínicas dos diferentes estágios de progressão. Não houve diferença estatisticamente significante na avaliação prática entre grupos e, um alto nível de satisfação da atividade foi observado na amostra. Conclui-se que, a atividade avaliada teve um impacto satisfatório no desenvolvimento de competências teóricas relacionadas à detecção de lesões de cárie utilizando o ICDAS. / This study aimed at evaluating the impact of a 3D model as a learning object in the training/teaching and satisfaction degree of undergraduate dental students from different contexts for the detection of caries lesions using the ICDAS. A multicenter controlled randomized trial was conducted, involving a convenience sample of undergraduate dental students from five institutions (1 national and 4 international). Firstly, students attended a traditional theoretical lecture and answered a first theoretical test. Then, they were randomly allocated into two groups as follows: 1) test group: receiving the theoretical lecture and accessing the 3D model, and 2) control group: receiving the theoretical lecture only. Afterwards, control group students left the room and a 6-minute video was projected (3D model). Once the video had finished, control group students returned to the room and both groups were submitted to a theoretical/practical test to evaluate their performance after intervention as well as their satisfaction degree. Multilevel linear and Poisson regression analyses were done, to analyze the learning object impact in the students´ theoretical/practical performance. Descriptive analyses were conducted to assess the students´ satisfaction degree. Three hundred and seven students participated. Those having a better performance in the initial theoretical test also had better grades in the final theoretical assessment (OR=1,11; 95%IC=1,02-1,21). Test group students had a better theoretical performance in comparison to control group ones (p=0,04), mainly in relation to questions regarding the ICDAS histological correlation with clinical features on each severity caries stage. There were no statistically significant differences regarding practical assessment between groups, and a high level of activity satisfaction was observed. In conclusion, the assessed activity had a satisfactory impact in the developing of theoretical skills in relation to the detection of caries lesions using the ICDAS.

[pt] Em aproximadamente um terço dos pacientes com epilepsia, a cirurgia é única forma de intervenção para diminuição dos impactos ou término das crises. Em pacientes sem um foco lesional na imagem por ressonância magnética, essa intervenção depende de outros métodos investigativos, que nem sempre estão prontamente disponíveis. Nesses casos, métodos avançados de pós-processamento e de sequências de imagens podem ajudar a detectar lesões. O objetivo dessa revisão sistemática foi resumir a disponibilidade e taxas de sucesso dessas técnicas. De acordo com as diretrizes PRISMA, usando as bases de dados PubMED, Web of Science, PsycNET e CENTRAL, uma busca por artigos foi conduzida até o dia 12 de janeiro de 2021. No total, a busca retornou 4.024 artigos, com 49 permanecendo após a revisão. Vinte e cinco artigos usaram alguma forma de voxel-based morphometry, 14 usaram machine learning e 10 usaram técnicas avançadas de MRI. Apenas um artigo descreveu um estudo prospectivo. A taxa de detecção de lesões variou bastante entre estudos, com técnicas de machine learning demonstrando taxas mais consistentes, todas acima de 50 por cento em grupos de pacientes com imagem negativa. Isso pode ser útil em centros onde outros métodos investigativos, como PET, SPECT, MEG ou sEEG não estão prontamente acessíveis. / [en] In approximately one third of patients with epilepsy, surgery is the only form of intervention to diminish seizure burden or achieve seizure freedom. In patients without a lesional focus on MRI, surgical intervention depends on other investigative methods, not always readily accessible. Advanced MRI postprocessing and acquisition methods may help with lesion localization in those cases. The aim of this systematic review was to summarize the availability and success rate of such MRI techniques. In accordance with the PRISMA guidelines, using PubMED, Web of Science, PsycNET, and CENTRAL, a search for papers was performed until the 12th of January of 2021. In total, the search returned 4,024 papers, of which 49 remained after revision. Twenty-five used a form of voxelbased morphometry, 14 used machine learning techniques, and 10 used advanced MRI sequences not commonly part of the standard MRI-protocol. Only one paper described a prospective study. The lesion detection rate greatly varied between studies, with machine learning techniques showing a more consistent rate, all above 50 percent in MRI-negative groups. This could be particularly helpful in center where other investigative methods, including PET, SPECT, MEG and stereo EEG are not readily available.

Cette étude vise à développer un système d’aide au diagnostic (CAD) pour la détection de lésions épileptogènes, reposant sur l’analyse de données de neuroimagerie, notamment, l’IRM T1 et FLAIR. L’approche adoptée, introduite précédemment par Azami et al., 2016, consiste à placer la tâche de détection dans le cadre de la détection de changement à l'échelle du voxel, basée sur l’apprentissage d’un modèle one-class SVM pour chaque voxel dans le cerveau. L'objectif principal de ce travail est de développer des mécanismes d’apprentissage de représentations, qui capturent les informations les plus discriminantes à partir de l’imagerie multimodale. Les caractéristiques manuelles ne sont pas forcément les plus pertinentes pour la tâche visée. Notre première contribution porte sur l'intégration de différents réseaux profonds non-supervisés, pour extraire des caractéristiques dans le cadre du problème de détection de changement. Nous introduisons une nouvelle configuration des réseaux siamois, mieux adaptée à ce contexte. Le système CAD proposé a été évalué sur l’ensemble d’images IRM T1 des patients atteints d'épilepsie. Afin d'améliorer la performance obtenue, nous avons proposé d'étendre le système pour intégrer des données multimodales qui possèdent des informations complémentaires sur la pathologie. Notre deuxième contribution consiste donc à proposer des stratégies de combinaison des différentes modalités d’imagerie dans un système pour la détection de changement. Ce système multimodal a montré une amélioration importante sur la tâche de détection de lésions épileptogènes sur les IRM T1 et FLAIR. Notre dernière contribution se focalise sur l'intégration des données TEP dans le système proposé. Etant donné le nombre limité des images TEP, nous envisageons de synthétiser les données manquantes à partir des images IRM disponibles. Nous démontrons que le système entraîné sur les données réelles et synthétiques présente une amélioration importante par rapport au système entraîné sur les images réelles uniquement. / This work represents one attempt to develop a computer aided diagnosis system for epilepsy lesion detection based on neuroimaging data, in particular T1-weighted and FLAIR MR sequences. Given the complexity of the task and the lack of a representative voxel-level labeled data set, the adopted approach, first introduced in Azami et al., 2016, consists in casting the lesion detection task as a per-voxel outlier detection problem. The system is based on training a one-class SVM model for each voxel in the brain on a set of healthy controls, so as to model the normality of the voxel. The main focus of this work is to design representation learning mechanisms, capturing the most discriminant information from multimodality imaging. Manual features, designed to mimic the characteristics of certain epilepsy lesions, such as focal cortical dysplasia (FCD), on neuroimaging data, are tailored to individual pathologies and cannot discriminate a large range of epilepsy lesions. Such features reflect the known characteristics of lesion appearance; however, they might not be the most optimal ones for the task at hand. Our first contribution consists in proposing various unsupervised neural architectures as potential feature extracting mechanisms and, eventually, introducing a novel configuration of siamese networks, to be plugged into the outlier detection context. The proposed system, evaluated on a set of T1-weighted MRIs of epilepsy patients, showed a promising performance but a room for improvement as well. To this end, we considered extending the CAD system so as to accommodate multimodality data which offers complementary information on the problem at hand. Our second contribution, therefore, consists in proposing strategies to combine representations of different imaging modalities into a single framework for anomaly detection. The extended system showed a significant improvement on the task of epilepsy lesion detection on T1-weighted and FLAIR MR images. Our last contribution focuses on the integration of PET data into the system. Given the small number of available PET images, we make an attempt to synthesize PET data from the corresponding MRI acquisitions. Eventually we show an improved performance of the system when trained on the mixture of synthesized and real images.