Cirrhosis recognition of liver ultrasound images based on SVM and uniform LBP feature

Abstract

Liver disease is one of the main causes of human healthy problem. Many clinical cases are still influenced by the subjectivity of physicians in some degree. Then, the subjectivity will affect the accuracy of diagnosis and the treatment of the patients. In this paper we proposes a new Computer Aided Diagnosis(CAD) system for the cirrhosis recognition in liver ultrasound(US) images using uniform LBP(u-LBP) features. This system seemed suitable for applying computer to recognize normal or cirrhotic liver, then the cirrhotic nidus will be earlier detected. We extract u-LBP features for each sample on the limited training and test datasets, and make a classification between the normal liver and cirrhotic liver through SVM, and we get a considerable recognition accuracy of 87.00%. Moreover, we have also made a comparison among the results of u-LBP-SVM, PCA-SVM and GLCM-SVM. And we got the conclusion that the proposed method which combined SVM and u-LBP features is relatively effective.