Liver disease, a serious threat to people’s health, is the most common and frequently-occurring disease, the number of patients increasing year by year. The diagnosis and treatment of hepatic disease has also been the hotspot and focus of medical. At present, the liver lesions clinical diagnosis should mainly rely on the doctor’s experience, with a lot of subjectivity and misdiagnosis phenomenon. This paper the second part aimed at the recognition method’s research and development of liver neoplasm multimodal images. Using the boundary contour’s difference between benign tumor and malignant, to extract the boundary feature. Using the advantage of support vector machine (SVM) in small sample data classification and complete theory support, to roundness, standardization radial length standard deviation, standardization radial length entropy and area ratio the four outline feature parameters, we use support vector machine to train and test, to obtain a classifier, getting the testing accuracy92.31%, providing a computer-aided diagnosis method for objective and accurate analysis of benign and malignant liver tumor. |