analysis of mri images of the liver, using a combination of wavelet and principle component analysis (pca) and support vector machine (svm) for the diagnosis and classification of benign and malignant tumors

The accurate detection of abnormal liver tissues, using an automatic classification system with accurate results in medicine is a critical issue, for which so many methods have been proposed so far. in this study, first we analyzed the liver images prepared by mri device, using wavelet in the frequency domain, differentiated them at different levels regarding resolution, extracted the features of the images. to increase algorithm speed we reduced features vector through a method called pca, then the selected features were classified, using a method called svm. in cross-validation stage, we used k-fold technique for generalization of the algorithm and four different kernels were implemented and then the results were compared. ultimately, this hybrid algorithm showed the best results with gaussian kernel. this method was compared with some of the previous methods, showing that it could produce good results in the classification of liver images and diagnosis of benign and malignant tumors, when there are few training data available, which can be used in medical diagnoses.