Computer-aided Hepatopathy Detection Based On MRI Perfusion Images

Since the liver perfusion computed tomography (CT) imaging was described. It has been generally used with the development of imaging techiques and poes-processing software. Just like the Magnetic Resonance (MR) Imaging has become the research focus of liver perfusion. In this paper we describe an automated method to detect the hepatopathy based on MR perfusion imaging with computer.This article (i)introduces the basic principles of magnetic resonance perfusion imaging of liver and summarized the currently available literature. Classic algorithms for liver perfusion (Non-deconvolution or Deconvolution) are reviewed systematically. Furthermore, experiments of all the methods are conducted and evaluated in terms of the complexity and the accuracy, (ii)studies the classification based on computer-aided of liver perfusion, focusing on exploring the application of support vector machine classification in liver lesions determination. Explore through experiments in computer-aided classification of liver lesions detection performance.In this article, (i)we review and evaluate the computer-aided liver perfusion based on MRI for the first time, including the complexity and the accuracy of each algorithm. (ii)we use the pattern classification method for computer-aided liver perfusion image sequence to detect liver lesions.Above all, we find the complexity and the accuracy of Dual-input One Compartment Model Deconvolution method are the best, but the BF, BV of Singular Value Decomposition method is closer to the actual value. When adding a pattern classification method, the false-positive rate is significantly improved. Whether it is single-core support vector machine(SVM) classification, or multi-core SVM classification, the accuracy rate also is significantly improved.