Classification of Normal and Fatty Liver Ultrasound Images.

Date of Submission

December 2009

Date of Award

Winter 12-12-2010

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Machine Intelligence Unit (MIU-Kolkata)


Maji, Pradipta (MIU-Kolkata; ISI)

Abstract (Summary of the Work)

Liver diseases are taken seriously because of the liver’s vital importance to the life of the patient. Fatty liver is a common liver disease caused by accumulation of fat in liver cells via the process of steatosis. It occurs when the fat content of the hepatocytes increases. It is the initial and most common histologic response to excessive alcohol ingestion. At present the global incidence rate of fatty liver is still increasing due to growth of obesity, alcoholism and diabetes, affecting estimated 10-24% of worlds population. Fatty liver could be reversible in its early stage, But if left untreated it may lead to inflammation of liver and patient’s liver may get permanantly damaged. So the early detection and treatment is crucial for control of the disease.1.1 Imaging Modalities for fatty liver Diagnosis The main diagnostic methods for fatty liver are Biopsy, B-mode ultrasound, CT and MRI. Liver biopsy, the diagnostic GOLD standard (as doctors say) is not well accepted by patients because it is invasive and it also has disadvantage that it poses a risk of cancer spreading if it cuts through a localized cancer area. MRI and CT-scan, on the other hand are quite expensive. Since B-scan ultrasound is noninvasive, inexpensive and easy to operate, it is the most commonly used modality to diagnose fatty liver. Doctors often make a presumptive diagnosis based on the B-mode ultrasound and then patient is diagnosed with a biopsy if case is found suspicious.1.2 Objective The purpose of this work was to design and implement Automatic Classifier(s), which, as accurately as possible, could classify fatty and normal livers using ultrasound images of the liver parenchyma, based on the traditional criteria used by the physicians in the diagnostic process by visual inspection of the ultrasound images.1.3 Importance of the workDiagnostic accuracy of fatty livers using only visual interpretation is currently estimated to be around 70-75%. Clinical diagnosis of B-mode ultrasound images of fatty liver, to a large extent, relies on the image quality and experience of technicians and doctors. Using the subjective judgment and non-quantitative description, doctors determine the incidence and the severity of fatty liver. However, the poor image quality, speckle noise, and use of different types of ultrasound imagers, and various physical conditions of patients obstruct a unified diagnostic standard. Therefore, a computer-aided liver ultrasound image quantitative analysis is necessary and will contribute to establishing a clinical objective fatty liver diagnosis method and standard, and improve clinical diagnostic accuracy, repeatability, and efficiency.1.4 Outline of the workIn this work, after preprocessing of B-mode ultrasound images of normal and fatty liver, ROIs of 4 different sizes 16 × 16, 32 × 32, 48 × 48 and 64 × 64 were selected as per doctors’ suggestions. 14 differnet features in total were extracted from each type of ROIs to distinguish between the two categories. Features were extracted according to the different characteristics of fatty liver and healthy liver. 12 Features were texture features and 2 other features were non-texture feature (i.e. independent of neighbourhood pixels). Some of the features were contrast, angular second moment, GLCM correlation etc.


ProQuest Collection ID:

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


This document is currently not available here.