An ultrasonic image evaluation system for assessing the severity of chronic liver disease

doi:10.1016/j.compmedimag.2007.05.001

Computerized Medical Imaging and Graphics

Volume 31, Issue 7, October 2007, Pages 485-491

https://doi.org/10.1016/j.compmedimag.2007.05.001 Get rights and content

Abstract

A quantitative ultrasonic image evaluation system that generates a numerical severity measurement to assess the progression of chronic liver disease and assist clinical diagnosis is proposed in this paper. The progression of chronic liver disease is closely related to the amount of fibrosis of the liver parenchyma under microscopic examination. The powerful index, computer morphometry (CM) score developed in Sun et al. [Sun YN, Horng MH, Lin XZ. Automatic computer morphometry system techniques and applications in medical diagnosis. In: Cornelius TL, editor. Computational methods in biophysics, biomaterials, biotechnology and medical systems. Algorithm development, mathematical analysis and diagnostics, vol. 4. Boston/Dordrecht/London: Kluwer Academic Publishers; 2003. p. 33–50], accurately measures the fibrosis ratio of liver parenchyma from a microscopy of human liver specimens. Therefore, the results of the CM score of patients serves as an assessment basis for developing the disease measurement of the B-mode liver sonogram under echo-texture feature analysis methods. The radial basis function (RBF) network is used to establish the correlates between texture features of ultrasonic liver image and the corresponding CM score. The output of the RBF network is called the ultrasonic disease severity (UDS) score. The correct classification rate of 120 test images by using the UDS score is 92.5%. These promising results reveal that the UDS is capable of providing an important reference to diagnose chronic liver disease.

Introduction

In clinical diagnosis, liver disease is taken seriously because liver is a vital organ in the human body. Liver diseases can be divided into two main categories, focal disease (e.g. hepatoma, hemangioma) and diffused diseases (e.g. hepatitis, cirrhosis). The focal diseases are where the abnormality is concentrated in a small area of liver tissue, whereas diffused disease is where the abnormality is distributed all over the liver volume. Chronic liver disease is frequent among diffused liver diseases. The severity of the disease on inflicted patients may range from a healthy carrier to hepatitis, and more serious to cirrhosis. Liver biopsy is the standard clinical routine for diagnosing chronic liver disease and for guiding and monitoring treatment [1], [2]. However they are associated with morbidity (3%) and mortality (0.03%). Therefore, developing a reliable, non-invasive and quantitative disease scoring system using a B-mode liver sonogram for evaluating histological changes is highly promising in diagnosing and monitoring chronic liver diseases.

The texture-based measures have been applied to ultrasound images over a decade. Richard and Keen extracted the micro-texture features using the Laws’ 5-by-5 feature masks and then applied the probabilistic relaxation algorithm to the segmentation of ultrasound images of the prostate [3]. Nicholas et al. used textural features of B-scan images to classify livers and spleens of normal humans [4]. Wu et al. applied a multi-resolution fractal feature vector and compared it with other texture measures to distinguish hepatoma, cirrhosis, and a normal liver [5]. Pavlopoulos et al. evaluated the performance of different texture analysis techniques for a quantitative characterization of ultrasonic liver images [6]. Bleck et al. used the autoregressive periodic random field model to classify normal and fatty liver, hepatitis and cirrhosis [7]. Mojsilivis et al. adopted the nonseparable wavelet transform to discriminate between the states of normal, steatosis and cirrhosis [8]. Gebbinck et al. and Kadah et al. applied discriminate analysis with a neural network to discriminate between various types of diffused liver diseases, such as disease of a fatty, cirrhotic, and normal liver [9], [10]. Pavlopoulos et al. applied the fuzzy neural network to classify liver states by using features of different texture analysis methods including gray-level difference statistics, gray-level run-length statistics, a spatial gray-level dependence matrix, and the fractal dimension [11]. This study classified the diseases of fatty, cirrhotic, and normal liver. Horng et al. compared the effectiveness of texture descriptors that include the gray-level co-occurrence matrix, the fractal dimension, the texture spectrum, the statistical feature matrix, and the texture feature coding method in classifying chronic liver diseases that are normal liver, hepatitis and cirrhosis [12]. The results of our past work [12] reveal that the features generated from the gray-level co-occurrence matrix (GLCM) [18] and texture feature coding method (TFCM) were effective for classifying the three liver states. As a result, two methods are used to select texture features for assessing the severity of chronic liver disease and discussed in this paper.

Unlike the above-mentioned literature on the classification of ultrasonic liver images, the goal of this paper is to develop a scoring system for measuring the progression of chronic liver diseases. The key to establishing the scoring system is to find the texture features that are highly correlated with the progression of liver disease. The progression of liver disease, from normal to liver hepatitis and more serious to cirrhosis, depends on the amount of fibrosis of the liver specimens from the histological view. In clinical diagnosis, Knodell's score [13] was broadly used to measure liver fibrosis when physicians observed the patient's specimens in a microscopy in the last two decades. It recorded only five discrete numerical scores for staging liver fibrosis, based on the physician's observation. Obviously, it was not enough and nor suitable for developing a quantitative progression index for the assessment of the liver disease. Alternatively, as reported in [14], [15], we proposed a quantitative index, called a computer morphometry (CM) score, which was more reliable and effective than the conventional Knodell's score for evaluating the amount of liver fibrosis. Therefore, the CM scores are used herein as the criteria for selecting powerful texture features.

The powerful ultrasonic scoring system should generate a disease severity score that matches the corresponding CM score as closely as possible. The correlates between the selected texture features and the corresponding CM score is intrinsic a multivariate regression problem. The radial basis function network is herein used to solve this multivariate regression [16]. The output of the radial basis function network is denoted to the ultrasonic disease severity (UDS) scores. Besides, the intervals of UDS scores in different liver states are also determined as standards for classification. Experiments with 40 training images demonstrate that the UDS scores generated from this system are significantly correlated to the CM scores of corresponding biopsy specimens. In addition, 120 test ultrasonic liver images were used to analyse the classification capability. The resulting correct classification rate reaches as high as 92.5%. These results reveal the proposed system of a UDS score has promising potentiality to provide valuable satisfactory reference for the diagnosis of chronic liver disease.

Section snippets

Ultrasonic liver image acquisition

A Toshiba Sonolayer SSA 250A sector scanner ultrasonic machine with the transducer PVE375A 3.75 MHz obtained the ultrasound images used in this study. A VFG frame grabber card on an IBM-PC was used to capture image frames with 512 × 512 spatial resolutions and 8 bits/pixel intensity quantification. Physicians through specially designed programs could manipulate the image acquisition. A number of software tools were developed on this system to allow physicians to define the ROI in the image and

Experimental results and discussion

In this study, we have developed an ultrasonic scoring system to assess the severity of chronic liver disease. The system integrates the techniques of texture analysis of a liver sonogram with the corresponding pathological CM score measurement. In this system, all programs are encoded by Visual C++ version 6.0 with a Pentium personal computer in a Microsoft Windows 2000 environment. The system provides user-friendly interfaces and efficient computation for real-time clinical evaluation. In the

Conclusion

This paper proposes a quantitative ultrasonic scoring system for assessing the chronic liver disease by analysing the characteristics of liver echo-texture. It not only generates quantitative indices to assess liver disease progression and also classify the disease states of ultrasonic liver images. This new quantitative index is called the ultrasonic disease severity (UDS) score. In clinical diagnosis, the amount of liver fibrosis is one of the key factors used to assess the progression of

Acknowledgment

The authors would like to thank the National Science Council, ROC NSC 93-2213-E-251-001 for support of this work.

Ming-Huwi Horng received the B.S. degree in Mathematics from National Cheng Kung University, Taiwan, in 1990 and the M.S. and Ph.D. degrees in Information Engineering form National Cheng Kung University in 1992 and 1997, respectively. He is currently associate professor at the Department of Information Technology, National PingTung Institute of Commerce. His research interests focus on medical image processing, evolutionary algorithm, pattern recognition and bioinformatics. He is a member of

References (21)

W.D. Richard et al.
Automated texture-based segmentation of ultrasound images of the prostate
Comput Med Imag Graph
(1996)
D. Nicholas et al.
Tissue characterization from ultrasound B-scan data
Ultrasound Med Biol
(1986)
M.H. Horng et al.
Texture feature coding method for classification of liver sonography
Comput Med Imag Graph
(2002)
Y.N. Sun et al.
Automatic computer morphometry system techniques and applications in medical diagnosis
G.F. Vawter et al.
Cystic fibrosis in adults: an autopsy study
Pathol Annu
(1979)
C.M. Wu et al.
Texture features for classification of ultrasonic liver images
IEEE Trans Med Imag
(1992)
S. Pavlopoulos et al.
Evaluation of texture analysis techniques for quantitative characterization of ultrasonic images
J.S. Bleck et al.
Random field models in the textural analysis of ultrasonic images of the liver
IEEE Trans Med Imag
(1996)
A. Mojsilivis et al.
Characterization of visually similar diffuse diseases from B-scan liver images using non-separable wavelet transform
IEEE Trans Med Imag
(1998)
Y.M. Kadah et al.
Classification algorithms for quantitative tissue characterization of diffuse liver disease from ultrasound image
IEEE Trans Med Imag
(1996)

There are more references available in the full text version of this article.

Cited by (21)

Machine learning based liver disease diagnosis: A systematic review
2022, Neurocomputing
Citation Excerpt :
It is envisioned that estimating tumor size and segments of the liver are novel tasks that could help to predict the stage of cancer and reoccurrence rate with procurement steps such as chemotherapy, surgery, etc., [3,13,191]. Furthermore, multimodal image input data with multi-scale, multi-resolution, multi-contrast and multi-phase can assist in predicting more reliable outcomes [172,241]. In this study, we investigate the potential of CAD system for malignant liver detection and treatment.
The computer-based approach is required for the non-invasive detection of chronic liver diseases that are asymptomatic, progressive, and potentially fatal in nature. In this study, we review the computer-aided diagnosis of hepatic lesions in view of diffuse- and focal liver disorders. This survey mainly focuses on three image acquisition modalities: ultrasonography, computed tomography, and magnetic resonance imaging. We present the insightful analysis with pros and cons for each preliminary step, particularly preprocessing, attribute analysis, and classification techniques to accomplish clinical diagnostic tasks. In preprocessing, we explore and compare commonly used denoising, deblurring and segmentation methods. Denoising is mainly performed with nonlinear models. In contrast, deep neural networks are frequently applied for deblurring and automatic segmentation of region-of-interest. In attribute analysis, the most common approach comprises texture properties. For classification, the support vector machine is mainly utilized across three image acquisition modalities. However, comparative analysis shows the best performance is obtained by deep learning-based convolutional neural networks. Considering biopsy samples or pathological factors such as overall stage, margin, and differentiation can be helpful for improving the prediction performance. In addition, technique breakthrough is expected soon with advances in machine learning models to address data limitation problems and improve the prediction performance.
Hybrid feature selection-based feature fusion for liver disease classification on ultrasound images
2020, Advances in Computational Techniques for Biomedical Image Analysis: Methods and Applications
This paper presents a hybrid feature selection (HFS)-based feature fusion system that selects the best features among multiple feature sets to classify liver ultrasound images into four classes: normal, chronic, cirrhosis, and heptocellular carcinomas evolved over cirrhosis. After extracting features by gray-level co-occurrence matrices, gray-level difference matrix and ranklet transform, the system utilizes HFS to select features. Here, HFS method is proposed by combining filter (ReliefF) and wrapper [sequential forward selection (SFS)] methods. Firstly, ReliefF method rank features and preselection are done by discarding low ranked features. Secondly, SFS method finds the optimal feature set. The advantage of proposed method is to make feature selection faster since filter method rapidly reduces the effective number of features under consideration. Thereafter, to take advantage of complementary information from different feature sets, feature fusion schemes are implemented: serial feature combination, serial feature fusion and, hierarchical feature fusion. Experiments are conducted to evaluate the (1) effectiveness of extracted features and proposed HFS method, (2) effectiveness of feature fusion schemes, and (3) performance based on the number of selected features, computational time and accuracy of ReliefF, SFS, sequential backward selection, and proposed method. Finally, the HFS-based hierarchical fusion set obtained accuracy of 95.2% with k-nearest neighbor.
Quantification of liver fat: A comprehensive review
2016, Computers in Biology and Medicine
Citation Excerpt :
Table 6 presents automated approaches that were proposed for classification of normal and fatty liver patterns without staging fatty liver diseases or grading of steatosis from US images [195, 208–219]. In the literature, there are also automated methods [194,200,222–238] that were proposed for diagnosis and evaluation of fatty liver using US images. Table 6 gives these methods, which were published in the last 15 years, with their conclusions.
Fat accumulation in the liver causes metabolic diseases such as obesity, hypertension, diabetes or dyslipidemia by affecting insulin resistance, and increasing the risk of cardiac complications and cardiovascular disease mortality. Fatty liver diseases are often reversible in their early stage; therefore, there is a recognized need to detect their presence and to assess its severity to recognize fat-related functional abnormalities in the liver. This is crucial in evaluating living liver donors prior to transplantation because fat content in the liver can change liver regeneration in the recipient and donor. There are several methods to diagnose fatty liver, measure the amount of fat, and to classify and stage liver diseases (e.g. hepatic steatosis, steatohepatitis, fibrosis and cirrhosis): biopsy (the gold-standard procedure), clinical (medical physics based) and image analysis (semi or fully automated approaches). Liver biopsy has many drawbacks: it is invasive, inappropriate for monitoring (i.e., repeated evaluation), and assessment of steatosis is somewhat subjective. Qualitative biomarkers are mostly insufficient for accurate detection since fat has to be quantified by a varying threshold to measure disease severity. Therefore, a quantitative biomarker is required for detection of steatosis, accurate measurement of severity of diseases, clinical decision-making, prognosis and longitudinal monitoring of therapy. This study presents a comprehensive review of both clinical and automated image analysis based approaches to quantify liver fat and evaluate fatty liver diseases from different medical imaging modalities.
Case-based reasoning support for liver disease diagnosis
2011, Artificial Intelligence in Medicine
Citation Excerpt :
Liver diseases are usually caused by inflammation or damaged hepatocytes infected with viruses, bacteria, or fungi, and exposure to toxins like alcohol and autoimmunity. The severity of the disease on infected patients may range from a healthy carrier to hepatitis, and more seriously to cirrhosis [2]. According to the statistics from the Department of Health in Taiwan [3], liver disease has been consistently listed as one of the top ten fatal diseases in Taiwan, costing more than 10,000 lives every year.
In Taiwan, as well as in the other countries around the world, liver disease has reigned over the list of leading causes of mortality, and its resistance to early detection renders the disease even more threatening. It is therefore crucial to develop an auxiliary system for diagnosing liver disease so as to enhance the efficiency of medical diagnosis and to expedite the delivery of proper medical treatment.
The study accordingly integrated the case-based reasoning (CBR) model into several common classification methods of data mining techniques, including back-propagation neural network (BPN), classification and regression tree, logistic regression, and discriminatory analysis, in an attempt to develop a more efficient model for early diagnosis of liver disease and to enhance classification accuracy. To minimize possible bias, this study used a ten-fold cross-validation to select a best model for more precise diagnosis results and to reduce problems caused by false diagnosis.
Through a comparison of five single models, BPN and CBR emerged to be the top two methods in terms of overall performance. For enhancing diagnosis performance, CBR was integrated with other methods, and the results indicated that the accuracy and sensitivity of each CBR-added hybrid model were higher than those of each single model. Of all the CBR-added hybrid models, the BPN–CBR method took the lead in terms of diagnosis capacity with an accuracy rate of 95%, a sensitivity of 98%, and a specificity of 94%.
After comparing the five single and hybrid models, the study found BPN–CBR the best model capable of helping physicians to determine the existence of liver disease, achieve an accurate diagnosis, diminish the possibility of a false diagnosis being given to sick people, and avoid the delay of clinical treatment.
Performance evaluation of multiple classification of the ultrasonic supraspinatus images by using ML, RBFNN and SVM classifiers
2010, Expert Systems with Applications
This paper proposed an effort to apply the several multi-class classifiers that are the maximum likelihood classifier, the radial basis function neural network, the fuzzy support vector machine and the error correcting output codes method to classify the ultrasonic supraspinatus images. The maximum mutual information criterion is adopted to search for the powerful features generating from the first order histogram statistics, gray-level co-occurrence matrix and texture feature coding method. In experiments, the most commonly used performance measures including the accuracy, sensitivity, accuracy and F_score are applied to evaluate the classification of the four classifiers. In addition, the Youden’s index, the discriminant power and the area of receiver operating characteristics curve are also used to analyze the classification capability. The experimental results demonstrate that the implementation of radial bass function neural network can achieve 94.1% classification accuracy and performance measures are significantly superior to the others.
Multi-class support vector machine for classification of the ultrasonic images of supraspinatus
2009, Expert Systems with Applications
Citation Excerpt :
Chen, Chang, Kuo, Chen, and Huang (2002) utilized the wavelet transform and neural networks to classify the breast tumors. Horng (2007) applied the texture feature coding method and gray-level co-occurrence matrix to extract correlated features for classification of ultrasonic liver images. Kakkos et al. (2007) used spatial gray level dependence matrices and gray level run-length statistics to discriminate the diseases of symptomatic carotid plaques.
This article proposes an effort to apply the multi-class support vector machine classifiers to classify the supraspinatus image into different disease groups that are normal, tendon inflammation, calcific tendonitis and supraspinatus tear. The supraspinatus tendon is often involved in the above-mentioned disease groups. Four different texture analysis methods – texture feature coding method, gray-level co-occurrence matrix, fractal dimension evaluation and texture spectrum – are used to extract features of tissue characteristic in the ultrasonic supraspinatus images. The mutual information criterion is adopted to select the powerful features from ones generated from the above-mentioned four texture analysis methods in the training stage, meanwhile, the five implementations of multi-class support vector machine classifiers are also designed to discriminate each image into one of the four disease groups in the classification stage. In experiments, the most commonly used performance measures including sensitivity, specificity, classification accuracy and false-negative rate are applied to evaluate the classification of the five implantations of multi-class support vector machines. In addition, the receiver operating characteristics analysis is also used to analyze the classification capability. The present results demonstrate that the implementation of multi-class fuzzy support vector machine can achieve 90% classification accuracy, and performance measures of this implementation are significantly superior to the others.

View all citing articles on Scopus

View full text

An ultrasonic image evaluation system for assessing the severity of chronic liver disease

Abstract

Introduction

Section snippets

Ultrasonic liver image acquisition

Experimental results and discussion

Conclusion

Acknowledgment

Comput Med Imag Graph

Ultrasound Med Biol

Comput Med Imag Graph

Automatic computer morphometry system techniques and applications in medical diagnosis

Cystic fibrosis in adults: an autopsy study

Pathol Annu

Texture features for classification of ultrasonic liver images

IEEE Trans Med Imag

Evaluation of texture analysis techniques for quantitative characterization of ultrasonic images

Random field models in the textural analysis of ultrasonic images of the liver

IEEE Trans Med Imag

Characterization of visually similar diffuse diseases from B-scan liver images using non-separable wavelet transform

IEEE Trans Med Imag

Classification algorithms for quantitative tissue characterization of diffuse liver disease from ultrasound image

IEEE Trans Med Imag