Uplifted Tissue Characterization and Classification of Fatty Liver Disease from Ultrasound Images

Abstract

A fatty liver is the result of the extra fat in liver cells and it causes deadly diseases such as liver cancer, fibrosis, cirrhosis, tumor and so on. For detecting fatty liver disease, we have developed a new system called ultrasound based Computer-Aided Diagnosis system. We have used Thershold with Median Filter to remove speckle noise from images. We use image texture based features like Entropy and Local Binary Pattern (LBP). To classify the liver image, we have used a built-in classifier: the Support Vector Machine (SVM), and a Polynomial Kernel function. It has been chosen for its accuracy. We have also used K-means Cluster rule to provides best product, once data set are well separated from one another. It can achieve better classification between the fatty liver tissues and normal tissues. Therefore, our proposed system is very efficient into hospital workflows and has been used in a clinical setting to obtain more reproducible results with high accuracy, reliability, speed and objectiveness of the diagnosis of fatty liver disease.