Fast and accurate analysis of fluorescence in-situ hybridization (FISH) images for signal counting will depend mainly upon two components: a classifier to discriminate between artifacts and valid signals of several fluorophores (colors), and well discriminating features to represent the signals. Our previous work has focused on the first component. To investigate the second component, we evaluate candidate feature sets by illustrating the probability density functions (pdfs) and scatter plots for the features. The analysis provides first insight into dependencies between features, indicates the relative importance of members of a feature set, and helps in identifying sources of potential classification errors. Class separability yielded by different feature subsets is evaluated using the accuracy of several neural network (NN)-based classification strategies, some of them hierarchical, as well as using a feature selection technique making use of a scatter criterion. The complete analysis recommends several intensity and hue features for representing FISH signals. Represented by these features, around 90% of valid signals and artifacts of two fluorophores are correctly classified using the NN. Although applied to cytogenetics, the paper presents a comprehensive, unifying methodology of qualitative and quantitative evaluation of pattern feature representation essential for accurate image classification. This methodology is applicable to many other real-world pattern recognition problems.

research has indicated the significance of accurate classification of fluorescence in situ hybridisation (FISH) signals for the detection of genetic abnormalities. Based on well-discriminating features and a trainable neural network (NN) classifier, a previous system enabled highly-accurate classification of valid signals and artefacts of two fluorophores. However, since this system employed several features that are considered independent, the naive Bayesian classifier (NBC) is suggested here as an alternative to the NN. The NBC independence assumption permits the decomposition of the high-dimensional likelihood of the model for the data into a product of one-dimensional probability densities. The naive independence assumption together with the Bayesian methodology allow the NBC to predict a posteriori probabilities of class membership using estimated classconditional densities in a close and simple form. Since the probability densities are the only parameters of the NBC, the misclassification rate of the model is determined exclusively by the quality of density estimation. Densities are evaluated by three methods: single Gaussian estimation (SGE; parametric method), Gaussian mixture model assuming spherical covariance matrices (GMM; semi-parametric method) and kernel density estimation (KDE; non-parametric method). For lowdimensional densities, the GMM generally outperforms the KDE that tends to overfit the training set

In this study, a support vector machine (SVM) classifies real world data of cytogenetic signals measured from fluorescence in-situ hybridization (FISH) images in order to diagnose genetic syndromes. The study implements the SVM structural risk minimization concept in searching for the optimal setting of the classifier kernel and parameters. We propose thresholding the distance of tested patterns from the SVM separating hyperplane as a way of rejecting a percentage of the miss-classified patterns thereby allowing reduction of the expected risk. Results show accurate performance of the SVM in classifying FISH signals in comparison to other state-ofthe-art machine learning classifiers, indicating the potential of an SVM-based genetic diagnosis system. 1. FISH image analysis and signal representation̷ In recent years, fluorescence in-situ hybridization (FISH) has emerged as one of the most significant new developments in the analysis of human chromosomes. FISH offers numerous advantages compared with conventional cytogenetic techniques since it allows detection of numerical chromosome abnormalities during normal cell interphase. An important application of FISH is dot counting, i.e. the enumeration of signals (dots) within the nuclei, as the dots in the image represent the inspected chromosomes. Dot counting is used for diagnosing numerical chromosomal aberrations in, e.g., haematopoietic neoplasia, solid tumors and prenatal diagnosis [1]. However, a major limitation of the FISH technique for dot counting is the need to examine large numbers of cells. This is required for an accurate estimation of the distribution of chromosomes over cell population, especially in applications involving a relatively low frequency of abnormal cells. As visual evaluation by a trained cytogeneticist of large numbers of cells and enumeration of hybridization signals is

Research article An integrative ChIP-chip and gene expression profiling to model SMAD regulatory modules

Abstract. Since its inception in 1992, the fuzzy ARTMAP (FAM) neural network (NN) has attracted researchers ’ attention as a fast, accurate, off and online pattern classifier. Since then, many studies have explored different issues concerning FAM optimization, training and evaluation, e.g., model sensitivity to parameters, ordering strategy for the presentation of the training patterns, training method and method of predicting the classification accuracy. Other studies have suggested variants to FAM to improve its generalization capability or overcome the prime limitation of the model, which is category proliferation (i.e., model complexity that increases with data complexity). Category proliferation is pronounced in problems that are noisy or contain a large degree of class statistical overlapping. In many investigations, FAM was improved by incorporating elements of optimization theory, Bayes ’ decision theory, evolutionary learning, and cluster analysis. Due to its appealing characteristics, FAM and its variants have been applied extensively and successfully to real-world classification problems. Numerous applications were reported in, for example, the processing of signals from different sources, images, speech, and text; recognition of speakers, image objects, handwritten, and genetic abnormalities; and medical and fault diagnoses. When compared to other state-of-the-art machine learning classifiers, FAM and its variants showed superior speed and ease of training, and in most cases they delivered comparable classification accuracy. 1