Application of artificial neural networks for the classification of liver lesions by image texture parameters
TL;DR: In this article, a multilayered back-propagation neural network was used for liver lesion classification using B-scan ultrasound images for normal, hemangioma and malignant livers.
Abstract: Ultrasound imaging is a powerful tool for characterizing the state of soft tissues; however, in some cases, where only subtle differences in images are seen as in certain liver lesions such as hemangioma and malignancy, existing B-scan methods are inadequate. More detailed analyses of image texture parameters along with artificial neural networks can be utilized to enhance differentiation. From B-scan ultrasound images, 11 texture parameters comprising of first, second and run length statistics have been obtained for normal, hemangioma and malignant livers. Tissue characterization was then performed using a multilayered backpropagation neural network. The results for 113 cases have been compared with a classification based on discriminant analysis. For linear discriminant analysis, classification accuracy is 79.6% and with neural networks the accuracy is 100%. The present results show that neural networks classify better than discriminant analysis, demonstrating a much potential for clinical application.
Citations
More filters
22 May 2008
TL;DR: The purpose of this study is to characterize the evolution of the liver diseases towards hepatocellular carcinoma (HCC), by finding the relevant textural features obtained from ultrasound images, which accurately surprise the changes of the Liver tissue in the context of this evolution.
Abstract: The purpose of this study is to characterize the evolution of the liver diseases towards hepatocellular carcinoma (HCC), by finding the relevant textural features obtained from ultrasound images, which accurately surprise the changes of the liver tissue in the context of this evolution. For the computation of the textural features, the following methods are used: first and second order statistics, edge-based statistics, fractal-based methods and multiresolution methods; specific methods for feature-selection are applied in order to determine the exhaustive set of independent and relevant features for each evolution phase. The specific values of these parameters, corresponding to each evolution phase, will be estimated through statistical methods. We will focus on modeling the cirrhosis and hepatocellular carcinoma, but the normal state and chronic viral hepatitis (CVH) are also taken into consideration. The final purpose is that of providing a reliable method for non-invasive characterization of the evolution towards HCC, in order to prevent this malignant tumor.
4 citations
30 Oct 2014
TL;DR: A texture analysis method based on the Textural Microstructure Cooccurrence Matrix of order two and three is developed and its role in abdominal tumor recognition is assessed, and feature selection methods are used in order to determine the most important features and to improve the classification process.
Abstract: The automatic, non-invasive diagnosis of the abdominal tumors is an important issue in nowadays research. We develop computerized methods for this purpose, based on ultrasound images. We previously defined the textural model of these tumors, consisting of the relevant textural features that best characterize them and of their specific values. In this work, we developed a texture analysis method based on the Textural Microstructure Cooccurrence Matrix (TMCM) of order two and three, and we assessed its role in abdominal tumor recognition. We used feature selection methods in order to determine the most important features and to improve the classification process. We assessed the classification performance using the old and the newly resulted textural features. For the experiments, we considered the hepatocellular carcinoma (HCC), the most frequent malignant liver tumor, versus the cirrhotic parenchyma on which it evolved, as well as the colorectal tumors versus the Inflammatory Bowel Diseases (IBD).
4 citations
01 Jan 2009
TL;DR: The purpose of this study is to elaborate a reliable method in order to characterize and differentiate the hepatocellular carcinoma in a non-invasive way, based only on information obtained from ultrasound images, by refining the processes of feature selection and classification.
Abstract: The purpose of our study is to elaborate a reliable method in order to characterize and differentiate the hepatocellular carcinoma in a non-invasive way, based only on information obtained from ultrasound images. Texture is a very important feature, which can reveal subtle characteristics of the tissue in ultrasound images, the computerized methods for texture characterization being able to overpass the limits of the subjective human eye. The textural features are analyzed using specific methods from the field of statistical pattern classification, the final objective being that of performing computer-aided and automatic diagnosis of HCC. Thus, we build the imagistic textural model of HCC, consisting in the exhaustive set of relevant textural features, which best characterize HCC, and their specific values in the case of HCC. In order to obtain the imagistic textural model, the following steps are due: 1.) an image analysis phase, consisting in the computation of the textural features; 2.) a learning step, involving the selection of the relevant textural features and the estimation of their specific values; 3) a validation phase, consisting in the evaluation of the imagistic textural model. In this paper we aim to build an improved imagistic textural model of HCC, by refining the processes of feature selection and classification. The feature selection will be realized by comparing the results obtained with efficient feature selection methods, applied individually, or in combination. The classification will be performed using the most appropriate classifiers, as well as metaclassifiers. HCC will be compared with other visually similar tissues: the cirrhotic parenchyma on which it evolves and the benign tumors.
4 citations
TL;DR: This study conducted on extracted premolars from people who required orthodontic treatment to demonstrate the usefulness of innovating data assessment techniques in the dental field.
Abstract: Esthetic dentistry imposes several demands on the artistic abilities of the dentist, and knowledge of the underlying scientific principles of tooth color is considered to be essential by Sikri. The supervised classification methods, such as the artificial neural networks, the support vector machines, and also the Bayesian classifier, and the feature selection methods, such as decision trees, genetic algorithms and neural networks, as well as independent component analysis combined with least square support vector machines, were applied successfully in the medical field but were less implemented in the dental analysis domain. This study was conducted on extracted premolars from people who required orthodontic treatment. Data gathering was done using spectrophotometric recordings of tooth color parameters before and after accelerated bleaching, staining, and control procedures on extracted teeth on which was simulated orthodontic treatment. Comparison between data mining techniques and classical statistical...
3 citations
02 Jul 2013
TL;DR: This work aims to discover the cirrhosis grades in a noninvasive manner, using computerized methods, using texture-based methods and statistics from gray levels, edge-based statistics, statistics of the textural microstructures, and also textural features computed at multiple resolutions, after applying the Wavelet transform.
Abstract: Cirrhosis characterization and grading is an important issue nowadays in the medical domain, as this disease can lead to death. We aim to discover the cirrhosis grades in a noninvasive manner, using computerized methods. Concerning the feature computation, we chose the texture-based methods, as they revealed subtle aspects of the tissue, not detectable by the human eye. For this purpose, we used first, second and third order statistics of the gray levels, edge-based statistics, statistics of the textural microstructures, and also textural features computed at multiple resolutions, after applying the Wavelet transform. All these features were inputs to clustering methods, such as k-means clustering and expectation maximization (EM), implemented for the determination of the cirrhosis grades, each grade corresponding to a certain cluster. The relevant textural features, for each discovered grade, were also identified, by computing a specific score, for each feature, based on the result of the clustering methods.
3 citations
References
More filters
01 Nov 1973
TL;DR: These results indicate that the easily computable textural features based on gray-tone spatial dependancies probably have a general applicability for a wide variety of image-classification applications.
Abstract: Texture is one of the important characteristics used in identifying objects or regions of interest in an image, whether the image be a photomicrograph, an aerial photograph, or a satellite image. This paper describes some easily computable textural features based on gray-tone spatial dependancies, and illustrates their application in category-identification tasks of three different kinds of image data: photomicrographs of five kinds of sandstones, 1:20 000 panchromatic aerial photographs of eight land-use categories, and Earth Resources Technology Satellite (ERTS) multispecial imagery containing seven land-use categories. We use two kinds of decision rules: one for which the decision regions are convex polyhedra (a piecewise linear decision rule), and one for which the decision regions are rectangular parallelpipeds (a min-max decision rule). In each experiment the data set was divided into two parts, a training set and a test set. Test set identification accuracy is 89 percent for the photomicrographs, 82 percent for the aerial photographic imagery, and 83 percent for the satellite imagery. These results indicate that the easily computable textural features probably have a general applicability for a wide variety of image-classification applications.
20,442 citations
TL;DR: This paper provides an introduction to the field of artificial neural nets by reviewing six important neural net models that can be used for pattern classification and exploring how some existing classification and clustering algorithms can be performed using simple neuron-like components.
Abstract: Artificial neural net models have been studied for many years in the hope of achieving human-like performance in the fields of speech and image recognition. These models are composed of many nonlinear computational elements operating in parallel and arranged in patterns reminiscent of biological neural nets. Computational elements or nodes are connected via weights that are typically adapted during use to improve performance. There has been a recent resurgence in the field of artificial neural nets caused by new net topologies and algorithms, analog VLSI implementation techniques, and the belief that massive parallelism is essential for high performance speech and image recognition. This paper provides an introduction to the field of artificial neural nets by reviewing six important neural net models that can be used for pattern classification. These nets are highly parallel building blocks that illustrate neural net components and design principles and can be used to construct more complex systems. In addition to describing these nets, a major emphasis is placed on exploring how some existing classification and clustering algorithms can be performed using simple neuron-like components. Single-layer nets can implement algorithms required by Gaussian maximum-likelihood classifiers and optimum minimum-error classifiers for binary patterns corrupted by noise. More generally, the decision regions required by any classification algorithm can be generated in a straightforward manner by three-layer feed-forward nets.
7,798 citations
Book•
01 Jan 1974
TL;DR: The objective is to establish an experimental procedure and show direct AFM progression from EMT to EMT using a simple, straightforward, and reproducible procedure.
Abstract: Pathologic basis of disease , Pathologic basis of disease , کتابخانه دیجیتالی دانشگاه علوم پزشکی و خدمات درمانی شهید بهشتی
5,162 citations
01 Apr 1976
TL;DR: In this paper, three standard approaches to automatic texture classification make use of features based on the Fourier power spectrum, on second-order gray level statistics, and on first-order statistics of gray level differences, respectively.
Abstract: Three standard approaches to automatic texture classification make use of features based on the Fourier power spectrum, on second-order gray level statistics, and on first-order statistics of gray level differences, respectively. Feature sets of these types, all designed analogously, were used to classify two sets of terrain samples. It was found that the Fourier features generally performed more poorly, while the other feature sets all performned comparably.
1,379 citations
TL;DR: The gray value distribution of the runs is proposed to be used to define two new features, viz., low gray level run emphasis ( LGRE) and high gray levelrun emphasis ( HGRE).
443 citations