Tamanna Howlader

Noise Reduction of cDNA Microarray Images Using Complex Wavelets

Noise reduction is an essential step of cDNA microarray image analysis for obtaining better-quality gene expression measurements. Wavelet-based denoising methods have shown significant success in traditional image processing. The complex wavelet transform (CWT) is preferred to the classical discrete wavelet transform for denoising of microarray images due to its improved directional selectivity for better representation of the circular edges of spots and near shift-invariance property. Existing CWT-based denoising methods are not efficient for microarray image processing because they fail to take into account the signal as well as noise correlations that exist between red and green channel images. In this paper, two bivariate estimators are developed for the CWT-based denoising of microarray images using the standard maximum a posteriori and linear minimum mean squared error estimation criteria. The proposed denoising methods are capable of taking into account both the interchannel signal and noise correlations. Significance of the proposed denoising methods is assessed by examining the effect of noise reduction on the estimation of the log-intensity ratio. Extensive experimentations are carried out to show that the proposed methods provide better noise reduction of microarray images leading to more accurate estimation of the log-intensity ratios as compared to the other CWT-based denoising methods.

Association of Low-Birth Weight with Malnutrition in Children under Five Years in Bangladesh: Do Mother's Education, Socio-Economic Status, and Birth Interval Matter?

Background Malnutrition in children under five years remains a significant problem in Bangladesh, despite substantial socio-economic progress and a decade of interventions aimed at improving it. Although several studies have been conducted to identify the important risk factors of malnutrition, none of them assess the role of low birth weight (LBW) despite its high prevalence (36%). This study examines the association between LBW and malnutrition using data from the Bangladesh Demographic and Health Survey (BDHS) 2011 and provides practical guidelines for improving nutritional status of children. Methods Malnutrition in children is measured in terms of their height-for-age, weight-for-height, and weight-for-age. Children whose Z-scores for either of these indices are below two standard deviations (–2SD) from median of WHO’s reference population are considered as stunted, wasted or underweight, respectively. The association between malnutrition and LBW was investigated by calculating adjusted risk-ratio (RR), which controls for potential confounders such as child’s age and sex, mother’s education and height, length of preceding-birth-interval, access to food, area of residence, household socio-economic status. Adjusted RR was calculated using both Cochran-Mantel-Haenszel approach and multivariable logistic regression models controlling for confounder. Results The prevalence of malnutrition was markedly higher in children with LBW than those with normal birth-weights (stunting: 51% vs 39%; wasting: 25% vs 14% and underweight: 52% vs 33%). While controlling for the known risk factors, children with LBW had significantly increased risk of becoming malnourished compared to their counter part with RR 1.23 (95% CI:1.16–1.30), 1.71 (95% CI:1.53–1.92) and 1.47 (95% CI: 1.38–1.56) for stunting, wasting and underweight, respectively. The observed associations were not modified by factors known to reduce the prevalence of malnutrition, such as higher education of mother, better household socio-economic conditions and longer birth-interval. Conclusions Higher education of mother, better household socio-economic conditions and prolonged birth intervals alone are not sufficient in bringing about substantial reductions in prevalence of child malnutrition in Bangladesh. Targeted interventions should be designed to reduce prevalence of LBW in addition to improving mother’s education and other socio-demographic conditions.

On the selection of 2D Krawtchouk moments for face recognition

Sparse representation of images using orthogonal two-dimensional Krawtchouk moments (2D KCMs) for face recognition is motivated by their ability to capture region-based higher-order hidden nonlinear structures from discrete coordinates of finitely supported images and the invariance of affine transformations of these moments to common geometric distortions. This paper presents the effectiveness of selecting the discriminatory set of KCMs as the global and local face features as opposed to traditional features obtained from heuristic choice of fixed-order moments or projection of the moments for recognizing an identity. The selection of significantly sparse 2D KCM-based features according to the proposed approach results in highly efficient face recognition method as compared to the other methods that use orthogonal moments such as the 2D Zernike, 2D Tchebichef or 2D Gaussian-Hermite. Experiments on challenging databases (viz., FRGC and CK-AUC) and comparisons with the well established projection, texture, and moment-based methods indicate superior recognition performance in terms of mean accuracy and robustness of the proposed holistic- or hybrid-type discriminative KCM-based method, especially when sample sizes are small and the intraclass faces have significant variations due to expressions. HighlightsHybrid-type face recognition using discriminative selection of Krawtchouk moments.A comparative study of different moment-based discriminative face features.Experiments on two challenging databases show superiority of the proposed selection.Fisher-scoring is preferable to LDA projection for moment-based face recognition.Effective facial parts are identified for recognition of expression variant faces.

Image fusion technique using multivariate statistical model for wavelet coefficients

Wavelet-based image fusion techniques have been highly successful in combining important features such as edges and textures of source images. In this work, a new discrete wavelet transform (DWT)-based fusion algorithm is proposed using a locally-adaptive multivariate statistical model for the wavelet coefficients of the source images as well as that of the fused image. The multivariate model is proposed based on the fact that the DWT coefficients of source images are correlated not only with each other but also with the fused image. By using this model as a joint prior function, an estimate of the fused coefficients is derived via the Bayesian maximum a posteriori estimation technique. Experimental results show that performance of the proposed fusion method is better than that of the other methods in terms of commonly-used metrics such as structural similarity, peak signal-to-noise ratio, and cross-entropy.

Wavelet-Based Noise Reduction by Joint Statistical Modeling of cDNA Microarray Images

Complementary DNA (cDNA) microarray experiments involve a large number of error-prone steps, which result in a high level of noise in the resulting red and green channel images. Removal of noise is a crucial step, since it makes further image processing easier and results in accurate gene expression measurements. The wavelet transform has shown significant success in the denoising of images including cDNA microarrays. Existing wavelet-based denoising methods process each image individually ignoring the information in the other channel. In this paper, a noise reduction technique is proposed that exploits the dependency between the wavelet transform coefficients of the two channels by using a locally-adaptive joint statistical model. The maximum a posteriori criterion is used to derive a joint estimator for the noise-free coefficients assuming suitable priors for the local variances. Significance of the proposed method is assessed by examining its effect on estimation of the log-intensity ratio. Experiments show that the proposed method provides an improved noise reduction performance in terms of mean squared error and yields log-intensity ratios that are close to the true values as compared to that of the standard denoising methods.

Bayesian Fusion of Ensemble of Multifocused Noisy Images

This paper addresses the problem of simultaneous fusion and denoising of an ensemble of multifocused noisy source images using statistical approach. The central theme of the paper is to develop a novel generalized Bayesian framework based on maximum a posteriori (MAP) estimation technique to obtain the fused image from the noisy observations using a multiscale wavelet transform. A mathematically tractable multivariate a priori function is used in the MAP estimator to derive the closed-form expression of the fusion rule for the wavelet coefficients of noisy images. Experiments are carried out on a number of test-sets having an ensemble of multifocused source images with varying noise strengths to evaluate the performance of the proposed MAP-based fusion method as compared to the existing methods. Results show that the performance of the proposed method is better than that of the other wavelet or principal component analysis-based methods in terms of various metrics such as the structural similarity, peak signal-to-noise ratio and cross-entropy, uses of which are common both in the areas of fusion and denoising. In addition, the proposed method yields excellent results in terms of visual quality even in the case of non-Gaussian noise as well as computational load.

Intelligent Recognition of Spontaneous Expression Using Motion Magnification of Spatio-temporal Data

The challenges of recognition of spontaneous expressions from spatio-temporal data include the characterization of subtle changes of facial textures, which in many cases occur for a very brief duration. In this context, the paper presents an intelligent approach for spontaneous expression recognition algorithm, wherein adaptive magnification of motion of spatio-temporal data is applied prior to the extraction of features of expression. The proposed magnification enhances the low-intensity facial activities without introducing notable artifacts for the high-intensity activities. The local binary patterns extracted from three-orthogonal planes of the Eulerian magnified spatio-temporal data are used as features of spontaneous expressions. The extracted features are classified using the well-known support vector machine classifier. Experiments are conducted on commonly-referred spatio-temporal databases such as the SMIC and MMI that have spontaneous expressions representing the micro- and meso-level facial activities, respectively. Experimental results reveal that the proposed approach of motion magnification prior to feature extraction significantly improves the detection and classification accuracy at the expense of acceptable robustness.

Pooled shrinkage estimator for quadratic discriminant classifier: an analysis for small sample sizes in face recognition

The quadratic discriminant classifier (QDC) is a well-known parametric Bayesian classifier that has been successfully applied to statistical pattern recognition problems. One such application is in automatic face recognition where the number of training images per subject is often found to be much less than the length of the facial features. In such a case, the QDC cannot be used because the class-specific covariance matrix on which it depends is either poorly estimated or singular thereby resulting in unacceptable classifier performance. High dimensional covariance estimation techniques such as shrinkage can alleviate this problem but only to a certain extent. This paper presents a computationally simple yet effective solution for further improving the QDC performance in small sample size scenarios. The proposed technique adopts a strategy of combining the class-specific shrinkage estimates of the covariance matrix to obtain a pooled shrinkage estimate, which is then plugged into the QDC. Experiments indicate that the proposed classifier leads to remarkable improvement in face recognition accuracy as compared to the existing classifiers such as the nearest neighbor, support vector machine and naive Bayes, irrespective of the nature of the database and feature extraction method. Monte Carlo simulations reveal that this improvement is due to the much lower mean squared error of the pooled shrinkage estimator which offers greater stability to the QDC.

On Wavelet-Based Methods for Noise Reduction of cDNA Microarray Images

Denoising is recognized as one of the mandatory preprocessing tasks in microarray image analysis. Sparse representations of image pixels are commonly exploited to develop efficient image denoising algorithms. Existing approaches to transform image pixels into sparse representations require computationally demanding optimization techniques or a huge amount of prior knowledge to learn the kernels. Nevertheless, due to the mathematical elegancy, different types of multiresolution analysis, in particular, the variants of wavelet transforms such as the discrete wavelet transform, stationary wavelet transform, and complex wavelet transform have been employed successfully to develop many high-performance microarray array image denoising algorithms. This article presents a review of the sequential development of the wavelet-based methods for microarray image denoising. The useful and well-known properties of wavelet coefficients have led to the development of these algorithms by exploiting the statistical nature of the coefficients of the image and noise. The objective of this article is to summarize the key features of these algorithms and provide constructive analysis through categorization and comparison. The surveyed methods are discussed with respect to algorithmic issues such as the type of wavelet transforms used, statistical models employed, computational complexity, and denoising performance metrics.