Automated detection of hematological abnormalities through classification of flow cytometric data patterns

Flow Cytometry analyzers have become trusted companions due to their ability to perform fast and accurate analyses of human blood. The aim of these analyses is to determine the possible existence of abnormalities in the blood that have been correlated with serious disease states, such as infectious mononucleosis, leukemia, and various cancers. Though these analyzers provide important feedback, it is always desired to improve the accuracy of the results. This is evidenced by the occurrences of misclassifications reported by some users of these devices. It is advantageous to provide a pattern interpretation framework that is able to provide better classification ability than is currently available. Toward this end, the purpose of this dissertation was to establish a feature extraction and pattern classification framework capable of providing improved accuracy for detecting specific hematological abnormalities in flow cytometric blood data. ^ This involved extracting a unique and powerful set of shift-invariant statistical features from the multi-dimensional flow cytometry data and then using these features as inputs to a pattern classification engine composed of an artificial neural network (ANN). The contribution of this method consisted of developing a descriptor matrix that can be used to reliably assess if a donor’s blood pattern exhibits a clinically abnormal level of variant lymphocytes, which are blood cells that are potentially indicative of disorders such as leukemia and infectious mononucleosis. ^ This study showed that the set of shift-and-rotation-invariant statistical features extracted from the eigensystem of the flow cytometric data pattern performs better than other commonly-used features in this type of disease detection, exhibiting an accuracy of 80.7%, a sensitivity of 72.3%, and a specificity of 89.2%. This performance represents a major improvement for this type of hematological classifier, which has historically been plagued by poor performance, with accuracies as low as 60% in some cases. This research ultimately shows that an improved feature space was developed that can deliver improved performance for the detection of variant lymphocytes in human blood, thus providing significant utility in the realm of suspect flagging algorithms for the detection of blood-related diseases.^

Use of Medical Subject Headings (MeSH) in Portuguese for categorizing web-based healthcare content

Introduction: Internet users are increasingly using the worldwide web to search for information relating to their health. This situation makes it necessary to create specialized tools capable of supporting users in their searches. Objective: To apply and compare strategies that were developed to investigate the use of the Portuguese version of Medical Subject Headings (MeSH) for constructing an automated classifier for Brazilian Portuguese-language web-based content within or outside of the field of healthcare, focusing on the lay public. Methods: 3658 Brazilian web pages were used to train the classifier and 606 Brazilian web pages were used to validate it. The strategies proposed were constructed using content-based vector methods for text classification, such that Naive Bayes was used for the task of classifying vector patterns with characteristics obtained through the proposed strategies. Results: A strategy named InDeCS was developed specifically to adapt MeSH for the problem that was put forward. This approach achieved better accuracy for this pattern classification task (0.94 sensitivity, specificity and area under the ROC curve). Conclusions: Because of the significant results achieved by InDeCS, this tool has been successfully applied to the Brazilian healthcare search portal known as Busca Saude. Furthermore, it could be shown that MeSH presents important results when used for the task of classifying web-based content focusing on the lay public. It was also possible to show from this study that MeSH was able to map out mutable non-deterministic characteristics of the web. (c) 2010 Elsevier Inc. All rights reserved.

Recognition of noisy images by PLL networks

We use networks composed of three phase-locked loops (PLLs), where one of them is the master, for recognizing noisy images. The values of the coupling weights among the PLLs control the noise level which does not affect the successful identification of the input image. Analytical results and numerical tests are presented concerning the scheme performance. (c) 2008 Elsevier B.V. All rights reserved.

Refinement and variability techniques in model transformation of software requirements

Tese de Doutoramento em Tecnologias e Sistemas de Informação

Forearm surface EMG signals recognition and muscoloskeletal system dynamics identification using intelligent computational methods

Biosignals processing, Biological Nonlinear and time-varying systems identification, Electomyograph signals recognition, Pattern classification, Fuzzy logic and neural networks methods

Neuropeptide-inducible upregulation of proteasome activity precedes nuclear factor kappa B activation in androgen-independent prostate cancer cells.

ABSTRACT: BACKGROUND: Upregulation of nuclear factor kappa B (NFκB) activity and neuroendocrine differentiation are two mechanisms known to be involved in prostate cancer (PC) progression to castration resistance. We have observed that major components of these pathways, including NFκB, proteasome, neutral endopeptidase (NEP) and endothelin 1 (ET-1), exhibit an inverse and mirror image pattern in androgen-dependent (AD) and -independent (AI) states in vitro. METHODS: We have now investigated for evidence of a direct mechanistic connection between these pathways with the use of immunocytochemistry (ICC), western blot analysis, electrophoretic mobility shift assay (EMSA) and proteasome activity assessment. RESULTS: Neuropeptide (NP) stimulation induced nuclear translocation of NFκB in a dose-dependent manner in AI cells, also evident as reduced total inhibitor κB (IκB) levels and increased DNA binding in EMSA. These effects were preceded by increased 20 S proteasome activity at lower doses and at earlier times and were at least partially reversed under conditions of NP deprivation induced by specific NP receptor inhibitors, as well as NFκB, IκB kinase (IKK) and proteasome inhibitors. AD cells showed no appreciable nuclear translocation upon NP stimulation, with less intense DNA binding signal on EMSA. CONCLUSIONS: Our results support evidence for a direct mechanistic connection between the NPs and NFκB/proteasome signaling pathways, with a distinct NP-induced profile in the more aggressive AI cancer state.

A simple randomized algorithm for consistent sequential prediction of ergodic time series

We present a simple randomized procedure for the prediction of a binary sequence. The algorithm uses ideas from recent developments of the theory of the prediction of individual sequences. We show that if thesequence is a realization of a stationary and ergodic random process then the average number of mistakes converges, almost surely, to that of the optimum, given by the Bayes predictor.

Model selection and error estimation

We study model selection strategies based on penalized empirical loss minimization. We point out a tight relationship between error estimation and data-based complexity penalization: any good error estimate may be converted into a data-based penalty function and the performance of the estimate is governed by the quality of the error estimate. We consider several penalty functions, involving error estimates on independent test data, empirical {\sc vc} dimension, empirical {\sc vc} entropy, andmargin-based quantities. We also consider the maximal difference between the error on the first half of the training data and the second half, and the expected maximal discrepancy, a closely related capacity estimate that can be calculated by Monte Carlo integration. Maximal discrepancy penalty functions are appealing for pattern classification problems, since their computation is equivalent to empirical risk minimization over the training data with some labels flipped.

Atypical EPO profiles in anti-doping analyses

Résumé : Erythropoietin (EPO) is a glycoprotein hormone endogenously produced by the kidney, whose main physiological role is the stimulation of erythropoiesis. Since the beginning of the nineties, recombinant human EPO (rhEPO), a potent anti-anaemia treatment drug, has been manufactured by pharmaceutical industries. However, the erythropoiesis stimulating power of rhEPO was rapidly misused by unscrupulous athletes in order to improve their performances in endurance sports. Endogenous EPO has the same amino-acid backbone as most of recombinant forms; the molecules however differ through their respective glycosylation patterns. This difference constitutes the basis of the usual EPO screening test (IEF) developed in 2000 and still currently used in all anti-doping laboratories of the world. Nowadays, 3 EPO generations have been commercialized. The fight against EPO abuse is a continuous challenge for anti-doping laboratories. The diversity of recombinant EPO forms and the continuous development of new ones considerably confuse the identification of EPO doping. Several facets of this fight were investigated in this work. One of the limiting aspects of doping agents screening is the availability of positive samples. Therefore, 2nd and 3rd generation EPOS, namely NESP and C.E.R.A., were injected to healthy subjects in the frame of pilot clinical studies. These latter allowed to review the current EPO identification criteria defined by the World Anti-Doping Agency (WADA) in the case of NESP and to validate and implement a new assay targeting C.E.R.A. in human serum. Both studies resulted in the determination of the respective detection windows of NESP and C.E.R.A. in biological fluids. Following that, Dynepo, a 1st generation EPO presenting similarities with the endogenous form, was also in the centre of a similar clinical study. Our work aimed to overcome the actual identification criteria, which are not adapted to Dynpeo, and to propose an alternative pattern classification method based on the discriminant analysis of IEF EPO profiles. This method might be validated for other EPO forms in the future. The detection window of this molecule was also determined. Under particular conditions, confounding effects can complicate the identification of EPO in biological matrices. For example, athletes having performed a strenuous physical effort can excrete modified isoforms of endogenous EPO, making it very similar to some recombinant forms. Such phenomena, called effort urines, were reproduced under controlled conditions and, after characterization of effort EPO, an urinary biochemical marker was proposed to unequivocally identify effort urines. It also happens that EPO analyses fail to detect endogenous levels of EPO. Such profiles were thoroughly investigated and potential causes identified. Natural reasons relying on urine properties and test specificity were underlined, but the possible addition of adulterant agents in urine samples was also considered. Therefore, a simple biochemical assay targeting the suspected substances was set up. Our work was based on the characterization of atypical EPO profiles from different origins. Therefore, 3 EPO molecules representing the 3 generations of the drug and 2 confounding effects confusing the results interpretation were studied. These studies resulted in tangible applications for the laboratory, the best example of which being the C.E.R.A. assay, but also in scientific findings allowing to improve our comprehension of EPO doping in sport.

Discriminant ECOC: a heuristic method for application dependent design of error correcting output codes

We present a heuristic method for learning error correcting output codes matrices based on a hierarchical partition of the class space that maximizes a discriminative criterion. To achieve this goal, the optimal codeword separation is sacrificed in favor of a maximum class discrimination in the partitions. The creation of the hierarchical partition set is performed using a binary tree. As a result, a compact matrix with high discrimination power is obtained. Our method is validated using the UCI database and applied to a real problem, the classification of traffic sign images.

To select or to weigh: a comparative study of linear combination schemes for superparent-one-dependence estimators

We conduct a large-scale comparative study on linearly combining superparent-one-dependence estimators (SPODEs), a popular family of seminaive Bayesian classifiers. Altogether, 16 model selection and weighing schemes, 58 benchmark data sets, and various statistical tests are employed. This paper's main contributions are threefold. First, it formally presents each scheme's definition, rationale, and time complexity and hence can serve as a comprehensive reference for researchers interested in ensemble learning. Second, it offers bias-variance analysis for each scheme's classification error performance. Third, it identifies effective schemes that meet various needs in practice. This leads to accurate and fast classification algorithms which have an immediate and significant impact on real-world applications. Another important feature of our study is using a variety of statistical tests to evaluate multiple learning methods across multiple data sets.

La plateforme Bubbles : un outil d'investigation des différences individuelles de stratégies de reconnaissance de l'identité des visages.

L’objectif de cette recherche est la création d’une plateforme en ligne qui permettrait d’examiner les différences individuelles de stratégies de traitement de l’information visuelle dans différentes tâches de catégorisation des visages. Le but d’une telle plateforme est de récolter des données de participants géographiquement dispersés et dont les habiletés en reconnaissance des visages sont variables. En effet, de nombreuses études ont montré qu’il existe de grande variabilité dans le spectre des habiletés à reconnaître les visages, allant de la prosopagnosie développementale (Susilo & Duchaine, 2013), un trouble de reconnaissance des visages en l’absence de lésion cérébrale, aux super-recognizers, des individus dont les habiletés en reconnaissance des visages sont au-dessus de la moyenne (Russell, Duchaine & Nakayama, 2009). Entre ces deux extrêmes, les habiletés en reconnaissance des visages dans la population normale varient. Afin de démontrer la faisabilité de la création d’une telle plateforme pour des individus d’habiletés très variables, nous avons adapté une tâche de reconnaissance de l’identité des visages de célébrités utilisant la méthode Bubbles (Gosselin & Schyns, 2001) et avons recruté 14 sujets contrôles et un sujet présentant une prosopagnosie développementale. Nous avons pu mettre en évidence l’importance des yeux et de la bouche dans l’identification des visages chez les sujets « normaux ». Les meilleurs participants semblent, au contraire, utiliser majoritairement le côté gauche du visage (l’œil gauche et le côté gauche de la bouche).

Prediction of scoliosis curve type based on the analysis of trunk surface topography

Scoliosis treatment strategy is generally chosen according to the severity and type of the spinal curve. Currently, the curve type is determined from X-rays whose acquisition can be harmful for the patient. We propose in this paper a system that can predict the scoliosis curve type based on the analysis of the surface of the trunk. The latter is acquired and reconstructed in 3D using a non invasive multi-head digitizing system. The deformity is described by the back surface rotation, measured on several cross-sections of the trunk. A classifier composed of three support vector machines was trained and tested using the data of 97 patients with scoliosis. A prediction rate of 72.2% was obtained, showing that the use of the trunk surface for a high-level scoliosis classification is feasible and promising.

Notes on PCA, Regularization, Sparsity and Support Vector Machines

We derive a new representation for a function as a linear combination of local correlation kernels at optimal sparse locations and discuss its relation to PCA, regularization, sparsity principles and Support Vector Machines. We first review previous results for the approximation of a function from discrete data (Girosi, 1998) in the context of Vapnik"s feature space and dual representation (Vapnik, 1995). We apply them to show 1) that a standard regularization functional with a stabilizer defined in terms of the correlation function induces a regression function in the span of the feature space of classical Principal Components and 2) that there exist a dual representations of the regression function in terms of a regularization network with a kernel equal to a generalized correlation function. We then describe the main observation of the paper: the dual representation in terms of the correlation function can be sparsified using the Support Vector Machines (Vapnik, 1982) technique and this operation is equivalent to sparsify a large dictionary of basis functions adapted to the task, using a variation of Basis Pursuit De-Noising (Chen, Donoho and Saunders, 1995; see also related work by Donahue and Geiger, 1994; Olshausen and Field, 1995; Lewicki and Sejnowski, 1998). In addition to extending the close relations between regularization, Support Vector Machines and sparsity, our work also illuminates and formalizes the LFA concept of Penev and Atick (1996). We discuss the relation between our results, which are about regression, and the different problem of pattern classification.

Increasing detection rate of user-to-root attacks using genetic algorithms

An extensive set of machine learning and pattern classification techniques trained and tested on KDD dataset failed in detecting most of the user-to-root attacks. This paper aims to provide an approach for mitigating negative aspects of the mentioned dataset, which led to low detection rates. Genetic algorithm is employed to implement rules for detecting various types of attacks. Rules are formed of the features of the dataset identified as the most important ones for each attack type. In this way we introduce high level of generality and thus achieve high detection rates, but also gain high reduction of the system training time. Thenceforth we re-check the decision of the user-to- root rules with the rules that detect other types of attacks. In this way we decrease the false-positive rate. The model was verified on KDD 99, demonstrating higher detection rates than those reported by the state- of-the-art while maintaining low false-positive rate.