Elsevier

Knowledge-Based Systems

Volume 75, February 2015, Pages 66-77
Knowledge-Based Systems

Ultrasound-based tissue characterization and classification of fatty liver disease: A screening and diagnostic paradigm

https://doi.org/10.1016/j.knosys.2014.11.021Get rights and content

Highlights

  • We review the modalities that detect Fatty Liver Disease (FLD).

  • We review the ultrasound-based Computer Aided Diagnostic techniques for FLD detection.

  • We conclude that there is need for more clinical trials to evaluate these techniques.

Abstract

Fatty Liver Disease (FLD) is a progressively prevalent disease that is present in about 15% of the world population. Normally benign and reversible if detected at an early stage, FLD, if left undetected and untreated, can progress to an irreversible advanced liver disease, such as fibrosis, cirrhosis, liver cancer and liver failure, which can cause death. Ultrasound (US) is the most widely used modality to detect FLD. However, the accuracy of US-based diagnosis depends on both the training and expertise of the radiologist. US-based Computer Aided Diagnosis (CAD) techniques for FLD detection can improve accuracy, speed and objectiveness of the diagnosis, and thereby, reduce operator dependability. In this paper, we first review the advantages and limitations of different diagnostic methods which are currently available to detect FLD. We then review the state-of-the-art US-based CAD techniques that utilize a range of image texture based features like entropy, Local Binary Pattern (LBP), Haralick textures and run length matrix in several automated decision making algorithms. These classification algorithms are trained using the features extracted from the patient data in order for them to learn the relationship between the features and the end-result (FLD present or absent). Subsequently, features from a new patient are input to these trained classifiers to determine if he/she has FLD. Due to the use of such automated systems, the inter-observer variability and the subjectivity of associated with reading images by radiologists are eliminated, resulting in a more accurate and quick diagnosis for the patient and time and cost savings for both the patient and the hospital.

Introduction

Fatty Liver (Steatosis) or Fatty Liver Disease (FLD) indicates accumulation of triglycerides (fat) in the liver. Fatty liver can occur with or without the intake of alcohol. In 1980, Ludwig et al. [76] named it as Non-Alcoholic Fatty Liver Disease (NAFLD) when the fatty liver condition is independent of alcohol intake. FLD has been associated to metabolic syndrome [127], and hence it leads to several diffuse and prevalent pathologies, such as diabetes mellitus, insulin resistance, hypertension, and dyslipidemia. The accumulation of fat in the liver may eventually lead to inflammation, condition called as alcoholic or non-alcoholic steatohepatisis (ASH or NASH) and finally to cirrhosis (which describes large scale liver degeneration associated with an increased risk of hepatocellular carcinoma [95], [116], [19]. Studies revealed that the prevalence of FLD depends on sex [99], ethnicity [37], and age [52]. Overall, FLD affects about 15% of the world population [37], [98]), and it is the most common reason for elevated liver enzymes and chronic liver disease in developed countries [18]. Early diagnosis of FLD is of paramount importance to prevent its degeneration into irreversible liver diseases, such as liver cancer [117] and acute liver failure [146]. FLD is also a major risk factor for heart attacks and stroke [138], [139]. Furthermore, advanced liver diseases result in higher health care utilization, which implies higher cost for the health care provider [14].

Even though detection of FLD is easy, the differential diagnosis of FLD is difficult [149]. In fact, FLD might be linked to different factors, such as infections, inflammations, and drug or toxin-related injuries. Hepatic steatosis is usually categorized as macrovesicular or microvesicular [149]. Macrovesicular steatosis is a common occurrence in ambulatory patients, and microvesicular steatosis is associated with severe mitochondrial injury and acute hepatic dysfunction [149]. Therefore, liver biopsy is the preferred diagnostic technique for FLD detection [100]. However, biopsy is invasive, and it causes anxiety and discomfort to patients due to pain and the possibility of bleeding/hemorrhage [80]. These complications occur in at least 1.3% of all cases and the mortality ranges from 0.1% to 0.5% [17]. Given the relatively high prevalence of FLD in the general population, minimal invasive procedures have been developed for FLD diagnosis and the assessment of its degree of severity. Among all noninvasive techniques, Ultrasound (US) is the most common and widely used imaging modality for FLD diagnosis, because it is (a) inexpensive, (b) emits no harmful radiation, (c) is widely available and (d) has high sensitivity. A major downside of this imaging modality is the operator dependability [135]. Computer Aided Diagnosis (CAD) systems have been and are being developed as adjunct techniques to reduce operator dependability and to get reproducible results [30], [39], [113], [125], [49], [144]. Therefore, developing CAD systems that detect early stage FLD is of utmost importance to: (a) save patients from unwanted anxiety, (b) increase the chance of recovery and (c) reduce the cost associated with providing treatments for advanced liver diseases [7].

In this paper, we first review the advantages and limitations of current modalities that are used for FLD detection (Section 2). Subsequently, we discuss the structure of an US-based CAD system and briefly describe the features that are extracted from the US images and the commonly used classification algorithms (Section 3). We then review the methodology and evaluation results of several CAD systems proposed in the literature (Section 4). In these techniques, first informative features are extracted from the US images. The features are used as input to train automated decision making systems. Coupling feature extraction with automated classification provides a way to evaluate the features in a practical setting, i.e. it is a way to find out how useful these features are for a working radiologist. After careful analysis of the literature, we found that the US-based CAD techniques for FLD can improve accuracy, speed and objectiveness of the diagnosis, and thereby, reduce operator dependability. We conclude the paper in Section 5.

Section snippets

Liver

The liver is the heaviest and the largest glandular organ in the human body and it is absolutely crucial to life [12]. The liver performs vital functions: synthesis of proteins, fats and fatty acids, metabolism and storage of carbohydrates, and bile production and excretion. It maintains both volume and quality of blood by filtering potentially harmful biochemical products from the blood. One of these harmful products is bilirubin, which forms during the breakdown of old blood cells [134].

Computer Aided Diagnostic (CAD) technique workflow

In this section, we present a few of the commonly used processing methods which extract informative features from US images. These features can be used in automated classification algorithms for FLD detection. Fig. 4 shows a typical system block diagram which depicts both data flow and algorithms which act on the data. The following sections provide a short discussion of these algorithms.

Review of published cad studies for FLD diagnosis

In this section, we review selected CAD based studies for FLD diagnosis.

Conclusion

Generally, US-based diagnosis of FLD is subjective in nature because it depends to a large extend on the skills and experience of the physicians. In this review, we established the need for computer based FLD diagnosis systems by reviewing the advantages and disadvantages of several currently used diagnostic modalities. Subsequently, we described techniques and algorithms used in the individual stages of a CAD system, namely feature extraction, feature selection, and classification. We also

References (153)

  • A.M. Hashem et al.

    Single stage and multistage classification models for the prediction of liver fibrosis degree in patients with chronic hepatitis C infection

    Comput. Methods Programs Biomed.

    (2012 Mar)
  • K.A. Jain et al.

    Spectrum of CT and sonographic appearance of fatty infiltration of the liver

    Clin. Imaging

    (1993)
  • A.E. Joseph et al.

    Comparison of liver histology with ultrasonography in assessing diffuse parenchymal liver disease

    Clin. Radiol.

    (1991)
  • M. Karcaaltincaba et al.

    Imaging of hepatic steatosis and fatty sparing

    Eur. J. Radiol.

    (2007)
  • D. Lebrec et al.

    Transvenous liver biopsy: an experience based on 1000 hepatic tissue samplings with this procedure

    Gastroenterology

    (1982)
  • R.H. Lin

    An intelligent model for liver disease diagnosis

    Artif. Intell. Med.

    (2009)
  • R.H. Lin et al.

    A hybrid diagnosis model for determining the types of the liver disease

    Comput. Biol. Med.

    (2010)
  • U.L. Mathiesen et al.

    Increasedliverechogenicityatultrasoundexaminationreflectsdegree of steatosis but not of fibrosis in asymptomatic patients with mild/moderate abnormalities of liver transaminases

    Dig. Liver Dis.

    (2002)
  • A.J. McCullough

    The clinical features, diagnosis and natural history of nonalcoholic fatty liver disease

    Clin. Liver Dis.

    (2004)
  • M.H. Mendler et al.

    Dual-energy CT in the diagnosis and quantification of fatty liver: limited clinical value in comparison to ultrasound scan and single-energy CT, with special reference to iron overload

    J. Hepatol.

    (1998)
  • P. Mills et al.

    Ultrasound in the diagnosis of granulomatous liver disease

    Clin. Radiol.

    (1990)
  • M. Muthu Rama Krishnan et al.

    Statistical analysis of mammographic features and its classification using support vector machine

    Expert Syst. Appl.

    (2010)
  • B.A. Neuschwander-Tetri et al.

    Nonalcoholic steatohepatitis: summary of an AASLD single topic conference

    Hepatology

    (2003)
  • U.R. Acharya et al.

    Application of higher order spectra for the identification of diabetes retinopathy stages

    J. Med. Syst.

    (2008)
  • U.R. Acharya et al.

    Analysis and automatic identification of sleep stages using higher order spectra

    Int. J. Neural Syst.

    (2010)
  • U.R. Acharya et al.

    Automated diagnosis of glaucoma using texture and higher order spectra features

    IEEE Trans. Inf. Technol. Biomed.

    (2011)
  • U.R. Acharya et al.

    Data mining framework for fatty liver disease classification in ultrasound: a hybrid feature extraction paradigm

    Med. Phys.

    (2012)
  • L.A. Adams et al.

    Nonalcoholic fatty liver disease

    CMAJ

    (2005)
  • L.A. Adams et al.

    Diagnostic evaluation of nonalcoholic fatty liver disease

    J. Clin. Gastroenterol.

    (2006)
  • V. Angelis, G. Felici, G. Mancinelli, Feature selection for data mining, in: Data Mining and Knowledge Discovery...
  • P. Angulo

    Nonalcoholic fatty liver disease

    N. Engl. J. Med.

    (2002)
  • I. Arias et al.

    The Liver: Biology and Pathobiology

    (2009)
  • B.A. Birnbaum et al.

    Multidetector row CT attenuation measurements: Assessment of intra- and interscanner variability with an anthropomorphic body CT phantom

    Radiology

    (2007)
  • A.A. Bravo et al.

    Liver biopsy

    N. Engl. J. Med.

    (2001)
  • J.D. Browning et al.

    Prevalence of hepatic steatosis in an urban population in the united states: impact of ethnicity

    Hepatology

    (2004)
  • V. Chandran et al.

    Pattern recognition using invariants defined from higher order spectra: 2-d image inputs

    IEEE Trans. Image Proc.

    (1997)
  • J. Choudhury et al.

    Clinical aspects of fatty liver disease

    Semin. Liver Dis.

    (2004)
  • K.C. Chua et al.

    Cardiac state diagnosis using higher order spectra of heart rate variability

    J. Med. Eng. Technol.

    (2008)
  • K.C. Chua et al.

    Analysis of epileptic eeg signals using higher order spectra

    J. Med. Eng. Technol.

    (2009)
  • D.W. Crabb

    Pathogenesis of alcoholic liver disease: newer mechanisms of injury

    Keio J. Med.

    (1999)
  • O. Dekel, J. Keshet, Y. Singer, An efficient online algorithm for hierarchical phoneme classification, in: Proceedings...
  • O. Dekel, J. Keshet, Y. Singer, Large margin hierarchical classification, in: Proceedings of the 21st International...
  • V.E. de Meijer et al.

    Systematic review and meta-analysis of steatosis as a risk factor in major hepatic resection

    Br. J. Surg.

    (2010)
  • A. de Moura Almeida et al.

    Fatty liver disease in severe obese patients: diagnostic value of abdominal ultrasound

    World J. Gastroenterol.

    (2008)
  • K. Doi, Current status and future potential of computer-aided diagnosis in medical imaging, Br. J. Radiol. (2005) 78...
  • P. Domingos et al.

    On the optimality of the simple Bayesian classifier under zero-one loss

    Mach. Learning

    (1997)
  • W. Duch Studies in fuzziness and soft computing, in: Filter Methods, vol. 207, 2006, pp....
  • J.C. Ducommun et al.

    The relation of liver fat to computed tomography numbers: a preliminary experimental study in rabbits

    Radiology

    (1979)
  • R.O. Duda et al.

    Pattern Classification

    (2012)
  • D.G. Duman et al.

    Computed tomography in nonalcoholic fatty liver disease: a useful tool for hepatosteatosis assessment?

    Dig. Dis. Sci.

    (2006)
  • Cited by (0)

    View full text