Atiq Islam

Multifractal Texture Estimation for Detection and Segmentation of Brain Tumors

A stochastic model for characterizing tumor texture in brain magnetic resonance (MR) images is proposed. The efficacy of the model is demonstrated in patient-independent brain tumor texture feature extraction and tumor segmentation in magnetic resonance images (MRIs). Due to complex appearance in MRI, brain tumor texture is formulated using a multiresolution-fractal model known as multifractional Brownian motion (mBm). Detailed mathematical derivation for mBm model and corresponding novel algorithm to extract spatially varying multifractal features are proposed. A multifractal feature-based brain tumor segmentation method is developed next. To evaluate efficacy, tumor segmentation performance using proposed multifractal feature is compared with that using Gabor-like multiscale texton feature. Furthermore, novel patient-independent tumor segmentation scheme is proposed by extending the well-known AdaBoost algorithm. The modification of AdaBoost algorithm involves assigning weights to component classifiers based on their ability to classify difficult samples and confidence in such classification. Experimental results for 14 patients with over 300 MRIs show the efficacy of the proposed technique in automatic segmentation of tumors in brain MRIs. Finally, comparison with other state-of-the art brain tumor segmentation works with publicly available low-grade glioma BRATS2012 dataset show that our segmentation results are more consistent and on the average outperforms these methods for the patients where ground truth is made available.

Learning people annotation from the web via consistency learning

The phenomenal growth of Image/Video on the web and the increasing sparseness of meta information to go along with forces us to look for signals from the Image/Video content for Search / Information Retrieval and Browsing based corpus exploration. One of the prominent type of information that users look for while searching/browsing through such corpora is information around the people present in the Image/Video. While face recognition has matured to some extent over the past few years, this problem remains a hard one due to a) absence of labelled data for such a large set of celebrities that users look for and b) the variability of age/makeup/expressions/pose in the target corpus. We propose a learning paradigm which we refer to as consistency learning to address both these issues by posing the problem of learning from weakly labelled training set. We use the text-image co-occurrence on the web as a weak signal of relevance and learn the set of consistent face models from this very large and noisy training set. The resulting system learns face models for a large set of celebrities directly from the web and uses it to tag Image/Video for better retrieval. While the proposed method has been applied to faces, we see it broadly applicable in any learning problem with a suitable similarity metric defined. We present results on learning from a very large dataset of 37 million images resulting in a validation accuracy of 92.68%.

Multifractal modeling, segmentation, prediction, and statistical validation of posterior fossa tumors

In this paper, we characterize the tumor texture in pediatric brain magnetic resonance images (MRIs) and exploit these features for automatic segmentation of posterior fossa (PF) tumors. We focus on PF tumor because of the prevalence of such tumor in pediatric patients. Due to varying appearance in MRI, we propose to model the tumor texture with a multi-fractal process, such as a multi-fractional Brownian motion (mBm). In mBm, the time-varying Holder exponent provides flexibility in modeling irregular tumor texture. We develop a detailed mathematical framework for mBm in two-dimension and propose a novel algorithm to estimate the multi-fractal structure of tissue texture in brain MRI based on wavelet coefficients. This wavelet based multi-fractal feature along with MR image intensity and a regular fractal feature obtained using our existing piecewise-triangular-prism-surface-area (PTPSA) method, are fused in segmenting PF tumor and non-tumor regions in brain T1, T2, and FLAIR MR images respectively. We also demonstrate a non-patient-specific automated tumor prediction scheme based on these image features. We experimentally show the tumor discriminating power of our novel multi-fractal texture along with intensity and fractal features in automated tumor segmentation and statistical prediction. To evaluate the performance of our tumor prediction scheme, we obtain ROCs and demonstrate how sharply the curves reach the specificity of 1.0 sacrificing minimal sensitivity. Experimental results show the effectiveness of our proposed techniques in automatic detection of PF tumors in pediatric MRIs.

Assessing product image quality for online shopping

Assessing product-image quality is important in the context of online shopping. A high quality image that conveys more information about a product can boost the buyers confidence and can get more attention. However, the notion of image quality for product-images is not the same as that in other domains. The perception of quality of product-images depends not only on various photographic quality features but also on various high level features such as clarity of the foreground or goodness of the background etc. In this paper, we define a notion of product-image quality based on various such features. We conduct a crowd-sourced experiment to collect user judgments on thousands of eBays images. We formulate a multi-class classification problem for modeling image quality by classifying images into good, fair and poor quality based on the guided perceptual notions from the judges. We also conduct experiments with regression using average crowd-sourced human judgments as target. We compute a pseudo-regression score with expected average of predicted classes and also compute a score from the regression technique. We design many experiments with various sampling and voting schemes with crowd-sourced data and construct various experimental image quality models. Most of our models have reasonable accuracies (greater or equal to 70%) on test data set. We observe that our computed image quality score has a high (0.66) rank correlation with average votes from the crowd sourced human judgments.

DIALOG ACT CLASSIFICATION USING ACOUSTIC AND DISCOURSE INFORMATION OF MAPTASK DATA

Dialog act (DA) classification is useful to understand the intentions of a human speaker. An effective classification of DA can be exploited for realistic implementation of expert systems. In this work, we investigate DA classification using both acoustic and discourse information for HCRC MapTask data. We extract several different acoustic features and exploit these features using a Hidden Markov Model (HMM) network to classify acoustic information. For discourse feature extraction, we propose a novel parts-of-speech (POS) tagging technique that effectively reduces the dimensionality of discourse features. To classify discourse information, we exploit two classifiers such as a HMM and Support Vector Machine (SVM). We further obtain classifier fusion between HMM and SVM to improve discourse classification. Finally, we perform an efficient decision-level classifier fusion for both acoustic and discourse information to classify 12 different DAs in MapTask data. We obtain 65.2% and 55.4% DA classification rates using acoustic and discourse information, respectively. Furthermore, we obtain combined accuracy of 68.6% for DA classification using both acoustic and discourse information. These accuracy rates of DA classification are either comparable or better than previously reported results for the same data set. For average precision and recall, we obtain accuracy rates of 74.89% and 69.83%, respectively. Therefore, we obtain much better precision and recall rates for most of the classified DAs when compared to existing works on the same HCRC MapTask data set.

Texture Estimation for Abnormal Tissue Segmentation in Brain MRI

This chapter discusses multi-fractal texture estimation and characterization of brain lesions (necrosis, edema, enhanced tumor, non-enhanced tumor, etc.) in magnetic resonance (MR) images. This work formulates the complex texture of tumor in MR images using a stochastic model known as multi-fractional Brownian motion (mBm). Mathematical derivations of the mBm model and corresponding algorithm to extract the spatially varying multi-fractal texture feature are discussed. Extracted multi-fractal texture feature is fused with other effective features to enhance the tissue characteristics. Segmentation of the tissues is performed by using a feature-based classification method. The efficacy of the mBm texture feature in segmenting different abnormal tissues is demonstrated using a large-scale publicly available clinical dataset. Experimental results and performance of the methods confirm the efficacy of the proposed technique in an automatic segmentation of abnormal tissues in multimodal (T1, T2, Flair, and T1contrast) brain MRIs.

Class specific gene expression estimation and classification in microarray data

In this work, we characterize genes using an oligonucleotide affymetrix gene expression dataset and propose a novel gene selection method based on samples from the posterior distributions of class-specific gene expression measures. We construct a hierarchical Bayesian framework for a random effect ANOVA model that allows us to obtain the posterior distributions of the class-specific gene expressions. We also formalize a novel class prediction scheme based on the samples from new posterior distributions of group specific gene expressions. Our experimental results show the class-discriminating power of the selected genes. Furthermore, we demonstrate that our prediction scheme classifies tissue samples into appropriate treatment groups with high accuracy. The computations are implemented by using Gibbs sampling. We compare the efficacy of our proposed gene selection and prediction methods with that of Pomeroy et. al (Nature, 2002) on the same CNS tumor sample dataset.

Algorithmic content generation for products

Content is one of the most essential parts of products on e-commerce websites such as eBay. It not only drives user-engagement but also traffic from various search engine websites based on the relevance. Generating the content for the products, however comes with a wide set of challenges, due to the complexity of commerce at scale, and requires new applications in text processing and information extraction to address some core issues. Some of the factors which need to be addressed are: scalability (millions of products), dynamism (products change with time), removal of item-specific or seller specific information (maintain generality), size of the content etc. Generally, curators are hired for writing the product descriptions manually, which is not cost-effective and is not scalable. In the current work, an algorithmic framework based on Natural Language Processing and Deep Learning is proposed and used to generate the content for ecommerce products. Seller descriptions for multiple items aggregated at a product level are used for content generation. Furthermore, a combination of behavioral and text signals such as search queries are also used to understand the user intent. Two different approaches are proposed in this work: Extraction (sentence retrieval) and Abstraction (sentence generation). The results of both the methods are analyzed and it is depicted that algorithmic content generation is scalable, fast and has potential to cut down the manualcuration cost dramatically.

Gene expression based prototype for automatic tumor prediction

Background Automatic detection of tumors is a challenging task due to the heterogeneous phenotypic and genotypic behaviors of cells within tumor types [1-3]. In recent years, a number of research endeavors have been reported in literatures that exploit microarray gene expression data to predict tissue/tumor types with high confidence [3-14]. However, in predicting tissue types, the above mentioned works neither explicitly considered correlation among the genes nor the probable subgroups within the known groups. In this work, our primary objective is to develop an automated prediction scheme for tumors based on DNA microarray gene expressions of tissue samples.

Gene Expression Based CNS Tumor Prototype for Automatic Tumor Detection

Tumors of central nervous system (CNS) represent a unique challenge in diagnosis and treatment because of their heterogeneous phenotypic and genotypic behavior. Unambiguous characterization of these tumors is essential towards accurate prognosis and therapy. Rapid advancements in microarray technologies have made it very promising to achieve this unambiguous characterization. However, because of the noisy nature of measured gene expression levels from microarray chips, careful preprocessing of gene expression data are necessary before statistical analysis can proceed.. In this paper, we propose a procedure for classifying central nervous system (CNS) tumors based on DNA microarray gene expressions of samples from patients with a variety of CNS tumor types. , After obtaining the tumor specific gene expression estimates, significantly expressed (marker) genes are located and clustered using a complete linkage hierarchical algorithm. The algorithm involves clustering together all genes that show high correlation in their expression measures across the samples.. From such gene-cluster, eigengene expressions are obtained by projecting the genes expressions within same cluster onto their first three principal components. In the final step of building prototype for any particular tumor type, the corresponding tissue samples with eigengene expressions are divided into subgroups using self-organizing map (SOM). The centroid of the with eigengenes expression is used as the prototype of the corresponding tumor type for each subgroup. In predicting the tumor type of a new tissue sample, distances are calculated between the new sample and all the centroid of all the tumor prototypes. The new tissue sample is classified to the tumor type of the nearest centroid. Experimental results reported in this paper strongly support the histological categorization of the tumors and the current knowledge of their molecular definitions.