Gladston J. P. Moreira

Voronoi distance based prospective space-time scans for point data sets: a dengue fever cluster analysis in a southeast Brazilian town

BackgroundThe Prospective Space-Time scan statistic (PST) is widely used for the evaluation of space-time clusters of point event data. Usually a window of cylindrical shape is employed, with a circular or elliptical base in the space domain. Recently, the concept of Minimum Spanning Tree (MST) was applied to specify the set of potential clusters, through the Density-Equalizing Euclidean MST (DEEMST) method, for the detection of arbitrarily shaped clusters. The original map is cartogram transformed, such that the control points are spread uniformly. That method is quite effective, but the cartogram construction is computationally expensive and complicated.ResultsA fast method for the detection and inference of point data set space-time disease clusters is presented, the Voronoi Based Scan (VBScan). A Voronoi diagram is built for points representing population individuals (cases and controls). The number of Voronoi cells boundaries intercepted by the line segment joining two cases points defines the Voronoi distance between those points. That distance is used to approximate the density of the heterogeneous population and build the Voronoi distance MST linking the cases. The successive removal of edges from the Voronoi distance MST generates sub-trees which are the potential space-time clusters. Finally, those clusters are evaluated through the scan statistic. Monte Carlo replications of the original data are used to evaluate the significance of the clusters. An application for dengue fever in a small Brazilian city is presented.ConclusionsThe ability to promptly detect space-time clusters of disease outbreaks, when the number of individuals is large, was shown to be feasible, due to the reduced computational load of VBScan. Instead of changing the map, VBScan modifies the metric used to define the distance between cases, without requiring the cartogram construction. Numerical simulations showed that VBScan has higher power of detection, sensitivity and positive predicted value than the Elliptic PST. Furthermore, as VBScan also incorporates topological information from the point neighborhood structure, in addition to the usual geometric information, it is more robust than purely geometric methods such as the elliptic scan. Those advantages were illustrated in a real setting for dengue fever space-time clusters.

Multi-objective dynamic programming for spatial cluster detection

The detection and inference of arbitrarily shaped spatial clusters in aggregated geographical areas is described here as a multi-objective combinatorial optimization problem. A multi-objective dynamic programming algorithm, the Geo Dynamic Scan, is proposed for this formulation, finding a collection of Pareto-optimal solutions. It takes into account the geographical proximity between areas, thus allowing a disconnected subset of aggregated areas to be included in the efficient solutions set. It is shown that the collection of efficient solutions generated by this approach contains all the solutions maximizing the spatial scan statistic. The plurality of the efficient solutions set is potentially useful to analyze variations of the most likely cluster and to investigate covariates. Numerical simulations are conducted to evaluate the algorithm. A study case with Chagas’ disease clusters in Brazil is presented, with covariate analysis showing strong correlation of disease occurrence with environmental data.

Inter-Patient ECG Heartbeat Classification with Temporal VCG Optimized by PSO

Classifying arrhythmias can be a tough task for a human being and automating this task is highly desirable. Nevertheless fully automatic arrhythmia classification through Electrocardiogram (ECG) signals is a challenging task when the inter-patient paradigm is considered. For the inter-patient paradigm, classifiers are evaluated on signals of unknown subjects, resembling the real world scenario. In this work, we explore a novel ECG representation based on vectorcardiogram (VCG), called temporal vectorcardiogram (TVCG), along with a complex network for feature extraction. We also fine-tune the SVM classifier and perform feature selection with a particle swarm optimization (PSO) algorithm. Results for the inter-patient paradigm show that the proposed method achieves the results comparable to state-of-the-art in MIT-BIH database (53% of Positive predictive (+P) for the Supraventricular ectopic beat (S) class and 87.3% of Sensitivity (Se) for the Ventricular ectopic beat (V) class) that TVCG is a richer representation of the heartbeat and that it could be useful for problems involving the cardiac signal and pattern recognition.

Variable neighborhood multiobjective genetic algorithm for the optimization of routes on IP networks

This paper proposes an algorithm to optimize multiple indices of Quality of Service of Multi Protocol Label Switching (MPLS) IP networks. The proposed algorithm, the Variable Neighborhood Multiobjective Genetic Algorithm (VN-MGA), is a Genetic Algorithm based on the NSGA-II, with the particular feature that different parts of a solution are encoded differently, at Level 1 and Level 2. In order to improve the results, both representations are needed. At Level 1, the first part of the solution is encoded, by considering as decision variables, the arrows that form the routes to be followed by each request (whilst the second part of the solution is kept constant), whereas at Level 2, the second part of the solution is encoded, by considering as decision variables, the sequence of requests, and first part is kept constant. The preliminary results shown here indicate that the proposed approach is promising, since the Pareto-fronts obtained by VN-MGA dominate the fronts obtained by fixed-neighborhood encoding schemes. Besides the potential benefits of the application of the proposed approach to the optimization of packet routing in MPLS networks, this work raises the theoretical issue of the systematic application of variable encodings, which allow variable neighborhood searches, as generic operators inside general evolutionary computation algorithms.

A CMA stochastic differential equation approach for many-objective optimization

In multiobjective optimization problems, Pareto dominance-based search techniques are known to lose their efficiency in problems with a large number of objective functions - the many-objective problems. This paper proposes an algorithm based on a stochastic differential equation approach combined with an evolutionary strategy for dealing with such problems. The proposed algorithm is intended to both allow the determination of tight Pareto-optimal solutions in many-objective problems (which is a difficult task for usual evolutionary algorithms) and to find a solution set that performs a relatively uniform sampling of the Pareto-optimal set (which is a deficiency of the known stochastic differential equation approach). The proposed algorithm is shown to attain such goals at a relatively low computational cost.

Automatic cardiac arrhythmia detection and classification using vectorcardiograms and complex networks.

This paper intends to bring new insights in the methods for extracting features for cardiac arrhythmia detection and classification systems. We explore the possibility for utilizing vectorcardiograms (VCG) along with electrocardiograms (ECG) to get relevant informations from the heartbeats on the MIT-BIH database. For this purpose, we apply complex networks to extract features from the VCG. We follow the ANSI/AAMI EC57:1998 standard, for classifying the beats into 5 classes (N, V, S, F and Q), and de Chazals scheme for dataset division into training and test set, with 22 folds validation setup for each set. We used the Support Vector Machinhe (SVM) classifier and the best result we chose had a global accuracy of 84.1%, while still obtaining relatively high Sensitivities and Positive Predictive Value and low False Positive Rates, when compared to other papers that follows the same evaluation methodology that we do.

A multi-objective approach for calibration and detection of cervical cells nuclei

The automation process of Pap smear analysis holds the potential to address womens health care in the face of an increasing population and respective collected data. A fundamental step for automating analysis is cell detection from light microscopy images. Such information serves as input to cell classification algorithms and diagnostic recommendation tools. This paper describes an approach to nuclei cell segmentation, which critically impacts the following steps for cell analyses. We developed an algorithm combining clustering and genetic algorithms to detect image regions with high diagnostic value. A major problem when performing the segmentation of images is the cellular overlay. We introduce a new nuclear targeting approach using heuristics associated with a multi-objective genetic algorithm. Our experiments show results using a public 45-image dataset, including comparison to other cell detection approaches. The findings suggest an improvement in the nuclei segmentation and promise to support more sophisticated schemes for data quality control.

Deep periocular representation aiming video surveillance

Abstract Usually, in the deep learning community, it is claimed that generalized representations that yielding outstanding performance / effectiveness require a huge amount of data for learning, which directly affect biometric applications. However, recent works combining transfer learning from other domains have surmounted such data application constraints designing interesting and promising deep learning approaches in diverse scenarios where data is not so abundant. In this direction, a biometric system for the periocular region based on deep learning approach is designed and applied on two non-cooperative ocular databases. Impressive representation discrimination is achieved with transfer learning from the facial domain (a deep convolutional network, called VGG) and fine tuning in the specific periocular region domain. With this design, our proposal surmounts previous state-of-the-art results on NICE (mean decidability of 3.47 against 2.57) and MobBio (equal error rate of 5.42% against 8.73%) competition databases.

Improving automatic cardiac arrhythmia classification: Joining temporal-VCG, complex networks and SVM classifier.

The classification of heartbeats using electrocardiogram (ECG) aiming arrhythmia detection is a well researched subject and still there are room for improvements concerning the recommended databases. In this sense, aiming to classify heartbeats for arrhythmia detection, we extend a previous ours proposal that uses vectorcardiogram, a bi-dimensional representation of two ECG leads, by incorporating the time component producing a three-dimensional representation, the temporal vectorcardiogram. Along with the new representation, also we apply complex networks to extract features from the temporal VCG. The new proposed features feed then a Support Vector Machines (SVM) classifier. The temporal VCG have increased in the global accuracy, and have better results classifying the N and S classes, when it is compared with the best result to our previous work with VCG. We conclude that new techniques to extract 3D features from the Temporal VCG could be an interesting research direction.

Optimizing acceptance frontier using PSO and GA for multiple signature iris recognition

In the last three decades, the eye iris has been investigated as the most unique phenotype feature in the biometric literature. In the iris recognition literature, works achieving outstanding accuracies in very well behavior environments, i.e., when the subjects eye is in a well lightened environment and at a fixed distance for image acquisition, are known since a decade. Nonetheless, in noncooperative environments, where the image acquisition aiming iris location/segmentation is not straightforward, the iris recognition problem has many open issues and has been well researched in recent works. A promising strategy in this context employs a classification approach using multiple signature extraction. Representations are extracted from overlapping and different parts of the iris region. Aiming a robust and noisy invariant classification, an increasing set of acceptance thresholds (frontier) are required for dealing with multiple signatures for iris recognition in the iris matching process. Usually this frontier is estimated using brute force algorithms and a specific step resolution playing an important trade-off between runtime and accuracy of recognition in terms of false acceptance and false rejection rates, measured using half total error rate (HTER). In this sense, this work aims to use Genetic Algorithms (GA) and Particle Swarm Optimization (PSO) for finding such frontier. Moreover, we also employ a robust feature extraction technique proposed by us in a previous work. The experiments showed that the use of these evolutionary algorithms provides similar, if not better, effectiveness in very little runtime in a complex database well-known in the literature. Furthermore, it is shown that the frontier obtained by PSO is more stable than the one obtained by GA.