Classification and staging of chronic liver disease from multimodal data

Chronic liver disease (CLD) is most of the time an asymptomatic, progressive, and ultimately potentially fatal disease. In this study, an automatic hierarchical procedure to stage CLD using ultrasound images, laboratory tests, and clinical records are described. The first stage of the proposed method, called clinical based classifier (CBC), discriminates healthy from pathologic conditions. When nonhealthy conditions are detected, the method refines the results in three exclusive pathologies in a hierarchical basis: 1) chronic hepatitis; 2) compensated cirrhosis; and 3) decompensated cirrhosis. The features used as well as the classifiers (Bayes, Parzen, support vector machine, and k-nearest neighbor) are optimally selected for each stage. A large multimodal feature database was specifically built for this study containing 30 chronic hepatitis cases, 34 compensated cirrhosis cases, and 36 decompensated cirrhosis cases, all validated after histopathologic analysis by liver biopsy. The CBC classification scheme outperformed the nonhierachical one against all scheme, achieving an overall accuracy of 98.67% for the normal detector, 87.45% for the chronic hepatitis detector, and 95.71% for the cirrhosis detector.

The usefulness of ultrasound in the classification of chronic liver disease

Chronic Liver Disease is a progressive, most of the time asymptomatic, and potentially fatal disease. In this paper, a semi-automatic procedure to stage this disease is proposed based on ultrasound liver images, clinical and laboratorial data. In the core of the algorithm two classifiers are used: a k nearest neighbor and a Support Vector Machine, with different kernels. The classifiers were trained with the proposed multi-modal feature set and the results obtained were compared with the laboratorial and clinical feature set. The results showed that using ultrasound based features, in association with laboratorial and clinical features, improve the classification accuracy. The support vector machine, polynomial kernel, outperformed the others classifiers in every class studied. For the Normal class we achieved 100% accuracy, for the chronic hepatitis with cirrhosis 73.08%, for compensated cirrhosis 59.26% and for decompensated cirrhosis 91.67%.

Chronic liver disease staging classification based on ultrasound, clinical and laboratorial data

In this work the identification and diagnosis of various stages of chronic liver disease is addressed. The classification results of a support vector machine, a decision tree and a k-nearest neighbor classifier are compared. Ultrasound image intensity and textural features are jointly used with clinical and laboratorial data in the staging process. The classifiers training is performed by using a population of 97 patients at six different stages of chronic liver disease and a leave-one-out cross-validation strategy. The best results are obtained using the support vector machine with a radial-basis kernel, with 73.20% of overall accuracy. The good performance of the method is a promising indicator that it can be used, in a non invasive way, to provide reliable information about the chronic liver disease staging.

Diffuse liver disease classification from ultrasound surface characterization, clinical and laboratorial data

In this work liver contour is semi-automatically segmented and quantified in order to help the identification and diagnosis of diffuse liver disease. The features extracted from the liver contour are jointly used with clinical and laboratorial data in the staging process. The classification results of a support vector machine, a Bayesian and a k-nearest neighbor classifier are compared. A population of 88 patients at five different stages of diffuse liver disease and a leave-one-out cross-validation strategy are used in the classification process. The best results are obtained using the k-nearest neighbor classifier, with an overall accuracy of 80.68%. The good performance of the proposed method shows a reliable indicator that can improve the information in the staging of diffuse liver disease.

Three-dimensial patchy lattice model: Ring formation and phase separation

We investigate the structural and thermodynamic properties of a model of particles with 2 patches of type A and 10 patches of type B. Particles are placed on the sites of a face centered cubic lattice with the patches oriented along the nearest neighbor directions. The competition between the self- assembly of chains, rings, and networks on the phase diagram is investigated by carrying out a systematic investigation of this class of models, using an extension ofWertheim's theory for associating fluids and Monte Carlo numerical simulations. We varied the ratio r epsilon(AB)/epsilon(AA) of the interaction between patches A and B, epsilon(AB), and between A patches, epsilon(AA) (epsilon(BB) is set to theta) as well as the relative position of the A patches, i.e., the angle. between the (lattice) directions of the A patches. We found that both r and theta (60 degrees, 90 degrees, or 120 degrees) have a profound effect on the phase diagram. In the empty fluid regime (r < 1/2) the phase diagram is reentrant with a closed miscibility loop. The region around the lower critical point exhibits unusual structural and thermodynamic behavior determined by the presence of relatively short rings. The agreement between the results of theory and simulation is excellent for theta = 120 degrees but deteriorates as. decreases, revealing the need for new theoretical approaches to describe the structure and thermodynamics of systems dominated by small rings. (C) 2014 AIP Publishing LLC.

An unsupervised approach to feature discretization and selection

Many learning problems require handling high dimensional datasets with a relatively small number of instances. Learning algorithms are thus confronted with the curse of dimensionality, and need to address it in order to be effective. Examples of these types of data include the bag-of-words representation in text classification problems and gene expression data for tumor detection/classification. Usually, among the high number of features characterizing the instances, many may be irrelevant (or even detrimental) for the learning tasks. It is thus clear that there is a need for adequate techniques for feature representation, reduction, and selection, to improve both the classification accuracy and the memory requirements. In this paper, we propose combined unsupervised feature discretization and feature selection techniques, suitable for medium and high-dimensional datasets. The experimental results on several standard datasets, with both sparse and dense features, show the efficiency of the proposed techniques as well as improvements over previous related techniques.

ECG biometrics using a dissimilarity space representation

Electrocardiogram (ECG) biometrics are a relatively recent trend in biometric recognition, with at least 13 years of development in peer-reviewed literature. Most of the proposed biometric techniques perform classifi-cation on features extracted from either heartbeats or from ECG based transformed signals. The best representation is yet to be decided. This paper studies an alternative representation, a dissimilarity space, based on the pairwise dissimilarity between templates and subjects' signals. Additionally, this representation can make use of ECG signals sourced from multiple leads. Configurations of three leads will be tested and contrasted with single-lead experiments. Using the same k-NN classifier the results proved superior to those obtained through a similar algorithm which does not employ a dissimilarity representation. The best Authentication EER went as low as 1:53% for a database employing 503 subjects. However, the employment of extra leads did not prove itself advantageous.

Text classification using compression-based dissimilarity measures

Arguably, the most difficult task in text classification is to choose an appropriate set of features that allows machine learning algorithms to provide accurate classification. Most state-of-the-art techniques for this task involve careful feature engineering and a pre-processing stage, which may be too expensive in the emerging context of massive collections of electronic texts. In this paper, we propose efficient methods for text classification based on information-theoretic dissimilarity measures, which are used to define dissimilarity-based representations. These methods dispense with any feature design or engineering, by mapping texts into a feature space using universal dissimilarity measures; in this space, classical classifiers (e.g. nearest neighbor or support vector machines) can then be used. The reported experimental evaluation of the proposed methods, on sentiment polarity analysis and authorship attribution problems, reveals that it approximates, sometimes even outperforms previous state-of-the-art techniques, despite being much simpler, in the sense that they do not require any text pre-processing or feature engineering.

Contributo para o estabelecimento de níveis de referência de diagnóstico em equipamentos de TC multidetectores

Actualmente a Tomografia Computorizada (TC) é um dos métodos de diagnóstico por imagem que tem uma maior contribuição para a dose de radiação X recebida pelos pacientes. Pretende-se com este estudo avaliar as doses praticadas em TC e contribuir para o estabelecimento de Níveis de Referência de Diagnóstico (NRD) na região da Grande Lisboa, Portugal. Foram efectuadas medições de dose em 5 equipamentos de TC multidetectores, considerando o abdómen como área anatómica de interesse. Recorreu-se a uma câmara de ionização e a um fantoma para obter o índice de dose de TC (CTDI) e o produto dose-comprimento (DLP), que permitem determinar os NRD. Estes valores foram comparados com os NRD propostos pela Guideline Europeia e com os estudos desenvolvidos em outros países, como o Reino Unido, Grécia e Taiwan. Os resultados revelaram que os valores de NRD obtidos neste estudo (16,7 mGy para o CTDIvol e 436,5 mGy·cm para o DLP) são discrepantes relativamente à Guideline Europeia (±50%), mas muito próximos relativamente aos NRD estabelecidos nos países considerados. Estes valores podem ser eventualmente explicados pelos equipamentos em análise e pela utilização de protocolos de exame adoptados pelos profissionais de Radiologia nas instituições analisadas. ABSTRACT - Nowadays Computed Tomography (CT) is one of the imaging techniques which have a large contribution to radiation dose received by patients. The purpose of this study is to evaluate CT doses and contribute to the establishment of Diagnostic Reference Levels (DRL) in Lisbon, Portugal. Dose measurements on 5 multidetector CT scanners have been performed, considering the abdomen as the anatomic region of interest. All measurements were performed using an ionization chamber and a phantom to obtain the index CT dose (CTDI) and the dose-length product (DLP), which are used to determine DRL. These values were compared not only with European reference dose values but also with DRL studies developed in other countries like United Kingdom, Greece and Taiwan. The results revealed that DRL values obtained in this study (CTDIvol is 16,7 mGy and DLP is 436,5 mGy·cm) have a higher discrepancy to European Guideline (±50%), while the DRL´s of other countries are nearest to values obtained in this study. Those differences may be eventually explained by the type of the evaluated equipments but also by the exam protocols used by the Radiology professionals on the analyzed institutions.

Stratification of the earth beneath the Azores from P and S receiver functions

Seismic recordings of IRIS/IDA/GSN station CMLA and of several temporary stations in the Azores archipelago are processed with P and S receiver function (PRF and SRF) techniques. Contrary to regional seismic tomography these methods provide estimates of the absolute velocities and of the Vp/Vs ratio up to a depth of similar to 300 km. Joint inversion of PRFs and SRFs for a few data sets consistently reveals a division of the subsurface medium into four zones with a distinctly different Vp/Vs ratio: the crust similar to 20 km thick with a ratio of similar to 1.9 in the lower crust, the high-Vs mantle lid with a strongly reduced VpNs velocity ratio relative to the standard 1.8, the low-velocity zone (LVZ) with a velocity ratio of similar to 2.0, and the underlying upper-mantle layer with a standard velocity ratio. Our estimates of crustal thickness greatly exceed previous estimates (similar to 10 km). The base of the high-Vs lid (the Gutenberg discontinuity) is at a depth of-SO km. The LVZ with a reduction of S velocity of similar to 15% relative to the standard (IASP91) model is terminated at a depth of similar to 200 km. The average thickness of the mantle transition zone (TZ) is evaluated from the time difference between the S410p and SKS660p, seismic phases that are robustly detected in the S and SKS receiver functions. This thickness is practically similar to the standard IASP91 value of 250 km. and is characteristic of a large region of the North Atlantic outside the Azores plateau. Our data are indicative of a reduction of the S-wave velocity of several percent relative to the standard velocity in a depth interval from 460 to 500 km. This reduction is found in the nearest vicinities of the Azores, in the region sampled by the PRFs, but, as evidenced by SRFs, it is missing at a distance of a few hundred kilometers from the islands. We speculate that this anomaly may correspond to the source of a plume which generated the Azores hotspot. Previously, a low S velocity in this depth range was found with SRF techniques beneath a few other hotspots.

Modelação da contribuição do tráfego automóvel para a qualidade do ar em zona industrial

A preocupação sobre a qualidade do ar nas zonas industriais confere aos estudos sobre a qualidade do ar uma importância acrescida. Este trabalho teve como objectivo saber qual a contribuição dos principais poluentes provenientes do tráfego automóvel para a qualidade do ar na zona do parque industrial da Sapec, da Península da Mitrena, concelho de Setúbal, recorrendo ao modelo meteorológico e de qualidade do ar, TAPM (The Air Pollution Model). Neste trabalho analisaram-se dados da estação de monitorização da qualidade do ar, mais próxima da zona de estudo (Subestação) por forma a caracterizar-se a zona em causa, a nível meteorológico e da qualidade do ar. Os dados metereológico desta estação também foram utilizados com o objectivo de se validar os resultados meteorológicos obtidos pelo modelo. Na avaliação da contribuição do tráfego para a qualidade do ar, recorreu-se a um estudo de tráfego realizado pela Estradas de Portugal (EP) em 2004. Este estudo realizou a contagem dos veículos que se dirigiram ao parque industrial nos dias 14 e 15 de Dezembro, num período de 24 horas. A partir dessa contagem e de factores de emissão foi possível determinar a contribuição, de cada classe de veículo, para as concentrações atmosféricas de PM10 (resultantes de processos de combustão e ressuspensão), NOx, CO e HC. A comparação entre os dados meteorológicos simulados e medidos mostram que o modelo teve um bom comportamento, isto é, as discrepâncias entre os valores simulados e medidos foram mínimas. Relativamente à contribuição de cada categoria de veículos para a qualidade do ar, verificou-se que a classe de pesados de mercadorias foi aquela que mais contribui para as emissões de PM10, NOx e HC, enquanto que para as emissões de CO foram os veículos ligeiros de passageiros que tiveram uma maior contribuição. As classes dos motociclos e ciclomotores foram aquelas que tiveram uma menor contribuição para as concentrações atmosféricas de poluentes. Comparando as emissões de PM10 provenientes dos processos de combustão e de ressuspensão conclui-se que a maior percentagem provem da ressuspensão.

Classificação semântica de imagens com k-NN e RN

Um dos maiores desafios tecnológicos no presente é o de se conseguir gerar e manter, de uma maneira eficiente e consistente, uma base de dados de objectos multimédia, em particular, de imagens. A necessidade de desenvolver métodos de pesquisa automáticos baseados no conteúdo semântico das imagens tornou-se de máxima importância. MPEG-7 é um standard que descreve o contudo dos dados multimédia que suportam estes requisitos operacionais. Adiciona um conjunto de descritores audiovisuais de baixo nível. O histograma é a característica mais utilizada para representar as características globais de uma imagem. Neste trabalho é usado o “Edge Histogram Descriptor” (EHD), que resulta numa representação de baixo nível que permite a computação da similaridade entre imagens. Neste trabalho, é obtida uma caracterização semântica da imagem baseada neste descritor usando dois métodos da classificação: o algoritmo k Nearest Neighbors (k-NN) e uma Rede Neuronal (RN) de retro propagação. No algoritmo k-NN é usada a distância Euclidiana entre os descritores de duas imagens para calcular a similaridade entre imagens diferentes. A RN requer um processo de aprendizagem prévia, que inclui responder correctamente às amostras do treino e às amostras de teste. No fim deste trabalho, será apresentado um estudo sobre os resultados dos dois métodos da classificação.

Embedded platform for ECG biometric recognition

Trabalho Final de Mestrado para obtenção do grau de Mestre em Engenharia de Electrónica e Telecomunicações

Colloidal dipolar interactions in 2D smectic-C films

We use a two-dimensional (2D) elastic free energy to calculate the effective interaction between two circular disks immersed in smectic-C films. For strong homeotropic anchoring, the distortion of the director field caused by the disks generates topological defects that induce an effective interaction between the disks. We use finite elements, with adaptive meshing, to minimize the 2D elastic free energy. The method is shown to be accurate and efficient for inhomogeneities on the length scales set by the disks and the defects, that differ by up to 3 orders of magnitude. We compute the effective interaction between two disk-defect pairs in a simple (linear) configuration. For large disk separations, D, the elastic free energy scales as similar to D-2, confirming the dipolar character of the long-range effective interaction. For small D the energy exhibits a pronounced minimum. The lowest energy corresponds to a symmetrical configuration of the disk-defect pairs, with the inner defect at the mid-point between the disks. The disks are separated by a distance that, is twice the distance of the outer defect from the nearest disk. The latter is identical to the equilibrium distance of a defect nucleated by an isolated disk.

Optimização da lista de células vizinhas em redes LTE

Trabalho Final de Mestrado para obtenção do grau de Mestre em Engenharia de Electrónica e Telecomunicações