Grading of Fatty Liver Disease and Classification of Focal Liver Lesions Using Ultrasound Images

Abstract

In the present study, a comprehensive image database of 41 B-Mode ultrasound images, including 14 mild, 14 moderate and 13 severe fatty liver disease and 42 B-Mode ultrasound images of focal liver lesions including, (1) typical and atypical cases of Cyst, HEM and MET lesions, (2) small and large HCC lesions, are used. The proposed interactive system for grading of fatty liver disease and classification of focal liver lesions using B-Mode US images consist of four modules. Module 1: SVM based CAC system for grading of fatty liver disease (without feature selection). Module 2: SVM based CAC system for grading of fatty liver disease (with feature selection). Module 3: SVM based CAC system for classification of focal liver lesions (without feature selection). Module 4: SVM based CAC system for classification of focal liver lesions (with feature selection). It was observed that (a) highest overall classification accuracy of 66.6 % was obtained using ratio of GLCM mean feature for grading of fatty liver disease (b) highest overall classification accuracy of 78.3 % was obtained using ratio of GLCM mean and GLCM range feature for classification of focal liver lesions.