A new measure of echogenicity of ultrasound images for liver classification

Abstract

This paper presents a new quantitative metric which can be used as indirect method for characterization of Liver Ultrasound images. This new metric is based upon the visual criterion considered by the radiologists using texture parameters of the Liver image and a measure of echogenicity. The proposed metric is a single parameter extracted from 7 texture features on the basis of Homogeneity, Roughness, Contrast, Granular-size and Orientation of the liver surface for classification into Fatty or Normal liver.

Staging of Fatty Liver Diseases Based on Hierarchical Classification and Feature Fusion for Back-Scan–Converted Ultrasound Images

“…As proved in Kadah et al and Singh et al, 9,16 ROI must have the size of at least 800 pixels to provide proper sampling distribution for estimating reliable statistics. A variety of ROI sizes, from 10 × 10 to 100 × 100 1,[7][8][9][10]16,18,20,21 have been used for classification of liver diseases in different studies. In the present work, according to the images, the ROI size of 64 × 64 pixels was considered the most appropriate size.…”

“…Second, several approaches have been proposed for feature extraction in liver tissue mostly based on texture analysis. Texture analysis approaches can be divided into two groups: one based on the relation of neighbor pixels in spatial domain such as gray-level co-occurrence matrix (GLCM), 2,16,17 gray-level difference statistical (GLDS), 16 gray-level run-length matrix (GLRLM), 2 and fractal parameters (k) 2,16 ; and one based on the analysis of transform coefficients such as Fourier power spectrum (FPS), 16 discrete wavelet transform (DWT), 17 and wavelet packet transform (WPT). 7,18 The transform domain texture analyzers 7,18,19 have better performance than spatial domain ones, 2,16,20,21 as texture characteristics are extracted more efficiently in the transform domain.…”

SVM-Based Characterization of Liver Ultrasound Images Using Wavelet Packet Texture Descriptors

“…As the texture measurements are sensitive to the selection of ROI size, it should be chosen so as to provide a good statistical population. In other studies, it has been demonstrated that ROI size must be at least 800 pixels to provide good sampling distribution for estimating reliable statistics [6][7][8], whereas in few other related researches, a sample size of at least 1,000 pixels is suggested to estimate reliable statistics [9][10][11]. However, different ROI sizes ranging from 10×10 [12][13][14][15], 25×25 [16], 30×30 [3], 32×32 [2,[17][18][19], 40×40 [20][21][22], 50×50 [1,17,23,24], 60×60 [6] and 64× 64 pixels [1,4,5,25] have been used for classification of liver diseases.…”

An improved method for liver diseases detection by ultrasound image analysis

“…[ 13 14 ] Texture analysis approaches can be divided into two groups: one based on the relation of neighbor pixels in spatial domain such as the gray level concurrence matrix (GLCM), the gray level difference statistical (GLDS), the gray level run length matrix (GLRLM) and fractal parameters (k); and one based on the analysis of transform coefficients such as Fourier power spectrum (FPS), discrete wavelet transform (DWT) and wavelet packet transform (WPT). The transform domain texture analyzers[ 7 15 16 ] have better performance than spatial domain ones[ 6 17 18 19 ] as texture characteristics are extracted more efficiently in transform domain. Among the explained texture analyzers in transform domain, WPT results in more independent and robust features since it performs a multi-resolution analysis in all frequency bands.…”

“…Singh et al . [ 19 ] used some texture models including GLCM, FPS, and fractals and proposed a new metric based on the features to classify fatty and normal liver.…”