A novel MAP-MRF approach for multispectral image contextual classification using combination of suboptimal iterative algorithms

In this paper we present a novel approach for multispectral image contextual classification by combining iterative combinatorial optimization algorithms. The pixel-wise decision rule is defined using a Bayesian approach to combine two MRF models: a Gaussian Markov Random Field (GMRF) for the observations (likelihood) and a Potts model for the a priori knowledge, to regularize the solution in the presence of noisy data. Hence, the classification problem is stated according to a Maximum a Posteriori (MAP) framework. In order to approximate the MAP solution we apply several combinatorial optimization methods using multiple simultaneous initializations, making the solution less sensitive to the initial conditions and reducing both computational cost and time in comparison to Simulated Annealing, often unfeasible in many real image processing applications. Markov Random Field model parameters are estimated by Maximum Pseudo-Likelihood (MPL) approach, avoiding manual adjustments in the choice of the regularization parameters. Asymptotic evaluations assess the accuracy of the proposed parameter estimation procedure. To test and evaluate the proposed classification method, we adopt metrics for quantitative performance assessment (Cohen`s Kappa coefficient), allowing a robust and accurate statistical analysis. The obtained results clearly show that combining sub-optimal contextual algorithms significantly improves the classification performance, indicating the effectiveness of the proposed methodology. (C) 2010 Elsevier B.V. All rights reserved.

Classification of breast masses using a committee machine of artificial neural networks

This work proposes a new approach using a committee machine of artificial neural networks to classify masses found in mammograms as benign or malignant. Three shape factors, three edge-sharpness measures, and 14 texture measures are used for the classification of 20 regions of interest (ROIs) related to malignant tumors and 37 ROIs related to benign masses. A group of multilayer perceptrons (MLPs) is employed as a committee machine of neural network classifiers. The classification results are reached by combining the responses of the individual classifiers. Experiments involving changes in the learning algorithm of the committee machine are conducted. The classification accuracy is evaluated using the area A. under the receiver operating characteristics (ROC) curve. The A, result for the committee machine is compared with the A, results obtained using MLPs and single-layer perceptrons (SLPs), as well as a linear discriminant analysis (LDA) classifier Tests are carried out using the student's t-distribution. The committee machine classifier outperforms the MLP SLP, and LDA classifiers in the following cases: with the shape measure of spiculation index, the A, values of the four methods are, in order 0.93, 0.84, 0.75, and 0.76; and with the edge-sharpness measure of acutance, the values are 0.79, 0.70, 0.69, and 0.74. Although the features with which improvement is obtained with the committee machines are not the same as those that provided the maximal value of A(z) (A(z) = 0.99 with some shape features, with or without the committee machine), they correspond to features that are not critically dependent on the accuracy of the boundaries of the masses, which is an important result. (c) 2008 SPIE and IS&T.

Combining Legendre's Polynomials and Genetic Algorithm in the Solution of Nonlinear Initial-Value Problems

This work develops a method for solving ordinary differential equations, that is, initial-value problems, with solutions approximated by using Legendre's polynomials. An iterative procedure for the adjustment of the polynomial coefficients is developed, based on the genetic algorithm. This procedure is applied to several examples providing comparisons between its results and the best polynomial fitting when numerical solutions by the traditional Runge-Kutta or Adams methods are available. The resulting algorithm provides reliable solutions even if the numerical solutions are not available, that is, when the mass matrix is singular or the equation produces unstable running processes.

DECISION TREE CLASSIFIERS FOR STAR/GALAXY SEPARATION

We study the star/galaxy classification efficiency of 13 different decision tree algorithms applied to photometric objects in the Sloan Digital Sky Survey Data Release Seven (SDSS-DR7). Each algorithm is defined by a set of parameters which, when varied, produce different final classification trees. We extensively explore the parameter space of each algorithm, using the set of 884,126 SDSS objects with spectroscopic data as the training set. The efficiency of star-galaxy separation is measured using the completeness function. We find that the Functional Tree algorithm (FT) yields the best results as measured by the mean completeness in two magnitude intervals: 14 <= r <= 21 (85.2%) and r >= 19 (82.1%). We compare the performance of the tree generated with the optimal FT configuration to the classifications provided by the SDSS parametric classifier, 2DPHOT, and Ball et al. We find that our FT classifier is comparable to or better in completeness over the full magnitude range 15 <= r <= 21, with much lower contamination than all but the Ball et al. classifier. At the faintest magnitudes (r > 19), our classifier is the only one that maintains high completeness (> 80%) while simultaneously achieving low contamination (similar to 2.5%). We also examine the SDSS parametric classifier (psfMag - modelMag) to see if the dividing line between stars and galaxies can be adjusted to improve the classifier. We find that currently stars in close pairs are often misclassified as galaxies, and suggest a new cut to improve the classifier. Finally, we apply our FT classifier to separate stars from galaxies in the full set of 69,545,326 SDSS photometric objects in the magnitude range 14 <= r <= 21.

Combining Monte Carlo simulation and density-functional theory to describe the spectral changes of Na(2) in liquid helium

Spectral changes of Na(2) in liquid helium were studied using the sequential Monte Carlo-quantum mechanics method. Configurations composed by Na(2) surrounded by explicit helium atoms sampled from the Monte Carlo simulation were submitted to time-dependent density-functional theory calculations of the electronic absorption spectrum using different functionals. Attention is given to both line shift and line broadening. The Perdew, Burke, and Ernzerhof (PBE1PBE, also known as PBE0) functional, with the PBE1PBE/6-311++G(2d,2p) basis set, gives the spectral shift, compared to gas phase, of 500 cm(-1) for the allowed X (1)Sigma(+)(g) -> B (1)Pi(u) transition, in very good agreement with the experimental value (700 cm(-1)). For comparison, cluster calculations were also performed and the first X (1)Sigma(+)(g) -> A (1)Sigma(+)(u) transition was also considered.

Study of the optical and magnetic properties of pyrimidine in water combining PCM and QM/MM methodologies

The solvent effects on the low-lying absorption spectrum and on the (15)N chemical shielding of pyrimidine in water are calculated using the combined and sequential Monte Carlo simulation and quantum mechanical calculations. Special attention is devoted to the solute polarization. This is included by an iterative procedure previously developed where the solute is electrostatically equilibrated with the solvent. In addition, we verify the simple yet unexplored alternative of combining the polarizable continuum model (PCM) and the hybrid QM/MM method. We use PCM to obtain the average solute polarization and include this in the MM part of the sequential QM/MM methodology, PCM-MM/QM. These procedures are compared and further used in the discrete and the explicit solvent models. The use of the PCM polarization implemented in the MM part seems to generate a very good description of the average solute polarization leading to very good results for the n-pi* excitation energy and the (15)N nuclear chemical shield of pyrimidine in aqueous environment. The best results obtained here using the solute pyrimidine surrounded by 28 explicit water molecules embedded in the electrostatic field of the remaining 472 molecules give the statistically converged values for the low lying n-pi* absorption transition in water of 36 900 +/- 100 (PCM polarization) and 36 950 +/- 100 cm(-1) (iterative polarization), in excellent agreement among one another and with the experimental value observed with a band maximum at 36 900 cm(-1). For the nuclear shielding (15)N the corresponding gas-water chemical shift obtained using the solute pyrimidine surrounded by 9 explicit water molecules embedded in the electrostatic field of the remaining 491 molecules give the statistically converged values of 24.4 +/- 0.8 and 28.5 +/- 0.8 ppm, compared with the inferred experimental value of 19 +/- 2 ppm. Considering the simplicity of the PCM over the iterative polarization this is an important aspect and the computational savings point to the possibility of dealing with larger solute molecules. This PCM-MM/QM approach reconciles the simplicity of the PCM model with the reliability of the combined QM/MM approaches.

Combining Mager and Steinmetz: The Effect of Grain Size and Maximum Induction on Hysteresis Energy Loss

Twelve samples with different grain sizes were prepared by normal grain growth and by primary recrystallization, and the hysteresis dissipated energy was measured by a quasi-static method. Results showed a linear relation between hysteresis energy loss and the inverse of grain size, which is here called Mager`s law, for maximum inductions from 0.6 to 1.5 T, and a Steinmetz power law relation between hysteresis loss and maximum induction for all samples. The combined effect is better described by a Mager`s law where the coefficients follow Steinmetz law.

Preliminary study of human breast tissue using synchrotron radiation combining WAXS and SAXS techniques

Using synchrotron radiation, we combined simultaneously wide angle X-ray scattering (WAXS) and small angle X-ray scattering (SAXS) techniques to obtain the scattering profiles of normal and neoplastic breast tissu-es samples at the momentum transfer range 6.28 nm(-1) <= Q(=4 pi.sin(theta/2)lambda) <= 50.26 nm(-1) and 0.15 nm(-1) <= Q <= 1.90 nm(-1), respectively. The results obtained show considerable differences between the scattering profiles of these tissues. We verified that the combination of some parameters (ratio between glandular and adipose peak intensity and third-order axial peak intensity) extracted from scattering profiles can be used for identifying breast cancer. (c) 2009 Elsevier Ltd. All rights reserved.

A Previously Undescribed Syndrome Combining Fibular Agenesis/Hypoplasia, Oligodactylous Clubfeet, Anonychia/Ungual Hypoplasia, and Other Defects

We describe an apparently new genetic syndrome in six members of a family living in a remote area in Northeastern Brazil. This syndrome comprises: short stature Clue to a marked decrease in the length of the lower limbs (predominantly mesomelic with fibular agenesis/marked hypoplasia), grossly malformed/deformed clubfeet with severe oligodactyly, tipper limbs with acromial dimples and variable motion limitation of the forearms and/or hands, severe nail hypoplasia/anonychia sometimes associated with mild brachydactyly and occasionally with pre-axial polydactyly. This syndrome is apparently distinct from the syndrome of brachydactyly-ectrodactyly with fibular aplasia or hypoplasia (OMIM 113310), the syndrome of fibular aplasia or hypoplasia, femoral bowing and poly-, syn-, and oligodactyly (OMIM 228930), and from other previously described conditions exhibiting fibular agenesis/hvpoplasia. (C) 2008 Wiley-Liss, Inc.

Improved detection of incipient vascular changes by a biotechnological platform combining post mortem MRI in situ with neuropathology

The histopathological counterpart of white matter hyperintensities is a matter of debate. Methodological and ethical limitations have prevented this question to be elucidated. We want to introduce a protocol applying state-of-the-art methods in order to solve fundamental questions regarding the neuroimaging-neuropathological uncertainties comprising the most common white matter hyperintensities [WMHs] seen in aging. By this protocol, the correlation between signal features in in situ, post mortem MRI-derived methods, including DTI and MTR and quantitative and qualitative histopathology can be investigated. We are mainly interested in determining the precise neuroanatomical substrate of incipient WMHs. A major issue in this protocol is the exact co-registration of small lesion in a tridimensional coordinate system that compensates tissue deformations after histological processing. The protocol is based on four principles: post mortem MRI in situ performed in a short post mortem interval, minimal brain deformation during processing, thick serial histological sections and computer-assisted 3D reconstruction of the histological sections. This protocol will greatly facilitate a systematic study of the location, pathogenesis, clinical impact, prognosis and prevention of WMHs. (C) 2009 Elsevier B.V. All rights reserved.

Combining two potential causes of metalloproteinase secretion causes abdominal aortic aneurysms in rats: a new experimental model

P>Progress in understanding the pathophysiology of abdominal aortic aneurysms (AAA) is dependent in part on the development and application of effective animal models that recapitulate key aspects of the disease. The objective was to produce an experimental model of AAA in rats by combining two potential causes of metalloproteinase (MMP) secretion: inflammation and turbulent blood flow. Male Wistar rats were randomly divided in four groups: Injury, Stenosis, Aneurysm and Control (40/group). The Injury group received a traumatic injury to the external aortic wall. The Stenosis group received an extrinsic stenosis at a corresponding location. The Aneurysm group received both the injury and stenosis simultaneously, and the Control group received a sham operation. Animals were euthanized at days 1, 3, 7 and 15. Aorta and/or aneurysms were collected and the fragments were fixed for morphologic, immunohistochemistry and morphometric analyses or frozen for MMP assays. AAAs had developed by day 3 in 60-70% of the animals, reaching an aortic dilatation ratio of more than 300%, exhibiting intense wall remodelling initiated at the adventitia and characterized by an obvious inflammatory infiltrate, mesenchymal proliferation, neoangiogenesis, elastin degradation and collagen deposition. Immunohistochemistry and zymography studies displayed significantly increased expressions of MMP-2 and MMP-9 in aneurysm walls compared to other groups. The haemo-dynamic alterations caused by the stenosis may have provided additional contribution to the MMPs liberation. This new model illustrated that AAA can be multifactorial and confirmed the key roles of MMP-2 and MMP-9 in this dynamic remodelling process.

Risk stratification in neonates and infants submitted to cardiac surgery with cardiopulmonary bypass: A multimarker approach combining inflammatory mediators, N-terminal pro-B-type natriuretic peptide and troponin I

Low cardiac output syndrome (LCOS) is a common problem following cardiac surgery with cardiopulmonary bypass (CPB) in neonates and infants, and its early recognition remains a challenging task. We aimed to test whether a multimarker approach combining inflammatory and cardiac markers provides complementary information for prediction of LCOS and death in children submitted to cardiac surgery with CPB. Forty-six children younger than 18 months with congenital heart defects were prospectively enrolled. No intervention was made. Blood samples were collected pre-operatively, during CPB and post-operatively (PO) for measurement of interleukin (IL)-6, IL-8, IL-10, tumor necrosis factor (TNF)-alpha, cardiac troponin I (cTnI) and N-terminal pro-B-type natriuretic peptide (NT-proBNP). Clinical data and outcome variables were recorded. Logistic regression was used to identify predictors of LCOS and death. Multivariate logistic regression identified pre-operative NT-proBNP and IL-8 4 h PO as independent predictors of LCOS, while cTnI 4 h PO and CPB length were independent predictors of death. The use of inflammatory and cardiac markers in combination improved sensitivity, negative predictive value and accuracy of the models. In conclusion, the combined assessment of inflammatory and cardiac biochemical markers can be useful for identifying young children at increased risk for LCOS and death after heart surgery with CPB. (C) 2008 Elsevier Ltd. All rights reserved.

Bayesian network classifiers: Beyond classification accuracy

This work proposes and discusses an approach for inducing Bayesian classifiers aimed at balancing the tradeoff between the precise probability estimates produced by time consuming unrestricted Bayesian networks and the computational efficiency of Naive Bayes (NB) classifiers. The proposed approach is based on the fundamental principles of the Heuristic Search Bayesian network learning. The Markov Blanket concept, as well as a proposed ""approximate Markov Blanket"" are used to reduce the number of nodes that form the Bayesian network to be induced from data. Consequently, the usually high computational cost of the heuristic search learning algorithms can be lessened, while Bayesian network structures better than NB can be achieved. The resulting algorithms, called DMBC (Dynamic Markov Blanket Classifier) and A-DMBC (Approximate DMBC), are empirically assessed in twelve domains that illustrate scenarios of particular interest. The obtained results are compared with NB and Tree Augmented Network (TAN) classifiers, and confinn that both proposed algorithms can provide good classification accuracies and better probability estimates than NB and TAN, while being more computationally efficient than the widely used K2 Algorithm.

Building binary-tree-based multiclass classifiers using separability measures

Various popular machine learning techniques, like support vector machines, are originally conceived for the solution of two-class (binary) classification problems. However, a large number of real problems present more than two classes. A common approach to generalize binary learning techniques to solve problems with more than two classes, also known as multiclass classification problems, consists of hierarchically decomposing the multiclass problem into multiple binary sub-problems, whose outputs are combined to define the predicted class. This strategy results in a tree of binary classifiers, where each internal node corresponds to a binary classifier distinguishing two groups of classes and the leaf nodes correspond to the problem classes. This paper investigates how measures of the separability between classes can be employed in the construction of binary-tree-based multiclass classifiers, adapting the decompositions performed to each particular multiclass problem. (C) 2010 Elsevier B.V. All rights reserved.

A review on the combination of binary classifiers in multiclass problems

Several real problems involve the classification of data into categories or classes. Given a data set containing data whose classes are known, Machine Learning algorithms can be employed for the induction of a classifier able to predict the class of new data from the same domain, performing the desired discrimination. Some learning techniques are originally conceived for the solution of problems with only two classes, also named binary classification problems. However, many problems require the discrimination of examples into more than two categories or classes. This paper presents a survey on the main strategies for the generalization of binary classifiers to problems with more than two classes, known as multiclass classification problems. The focus is on strategies that decompose the original multiclass problem into multiple binary subtasks, whose outputs are combined to obtain the final prediction.