Ultrasound Liver Surface and Textural Characterization for the Detection of Liver Cirrhosis
This chapter addresses the problem of liver cirrhosis classification via ultrasound imaging. For this classification problem, a liver semiautomatic contour segmentation algorithm to characterize the morphology and a textural feature extraction scheme for
- PDF / 617,742 Bytes
- 24 Pages / 439.37 x 666.142 pts Page_size
- 54 Downloads / 191 Views
Abstract This chapter addresses the problem of liver cirrhosis classification via ultrasound imaging. For this classification problem, a liver semiautomatic contour segmentation algorithm to characterize the morphology and a textural feature extraction scheme for the characterization of liver parenchyma are proposed. Phase congruency is used to enhance liver contour and help medical doctor in the inspection of liver surface. The regularity of the enhanced liver contour is characterized from geometrical features that are used together with US textural features in the classification process. The classification of the proposed method is tested by using support vector machine, Bayesian, Parzen and k-nearest neighbor classifiers and their performance are compared. The Bayes classifier outperformed the compared classifiers, attaining an overall accuracy of 87.22 %, with a detection rate of 88.52 % and 86.11 % for the non-cirrhotic and cirrhotic class, respectively.
R. Ribeiro (*) Institute for Systems and Robotics/Instituto Superior Te´cnico, Escola Superior de Tecnologia da Sau´de de Lisboa, Lisboa, Portugal e-mail: [email protected] R.T. Marinho Liver Unit, Department of Gastroenterology and Hepatology, Hospital de Santa Maria, Medical School of Lisbon, Lisbon, Portugal e-mail: [email protected] J. Suri Point of Care Division-Healthcare Solutions Global Biomedical Technologies, Inc., Roseville, CA, USA e-mail: [email protected] J.M. Sanches Instituto de Sistemas e Robo´tica, Instituto Superior Te´cnico, Torre Norte, 6 Piso, Av. Rovisco Pais, 1049-001 Lisboa, Portugal e-mail: [email protected] A.S. El-Baz et al. (eds.), Abdomen and Thoracic Imaging: An Engineering & Clinical Perspective, DOI 10.1007/978-1-4614-8498-1_6, © Springer Science+Business Media New York 2014
145
146
R. Ribeiro et al.
Introduction Chronic liver disease (CLD) is an oncogenic disease where most likely developments, if not treated, are hepatocellular carcinoma (HCC) or death. Major progress in the knowledge and management of liver disease has been observed in the past 30 years; however, still 29 million people in the European Union suffer from a chronic liver condition [1]. Cirrhosis is the end-stage of every CLD [2], defined as the histological development of regenerative nodules surrounded by fibrous bands in response to chronic liver injury [3] and has a variety of clinical manifestations and complications [4]. The most probable outcome of cirrhosis is HCC, which is the fifth most common cause of cancer in Europe [1]. There are 14–26 new cirrhosis cases per 100,000 inhabitants per year or an estimated 170,000 deaths per year [1]. Four leading causes of cirrhosis have been identified, namely chronic alcohol consumption, chronic viral hepatitis B, chronic viral hepatitis C, and non-alcoholic fatty liver disease (NAFLD). If detected in time, each of these causes is responsive to treatment. The problem is, in part, related to the fact that CLD is characterized by a silent and asymptomatic phase. Thus, the solution relies on
Data Loading...
