3D segmentation of MRI of the liver using support vector machine

Abstract

In this paper, we propose a semi-supervised method for segmentation of the liver in three-dimensional (3D) magnetic resonance images (MRI), based on a Support Vector Machine (SVM) classifier. For segmentation, two classes have been considered: 'Liver' and 'Background'. Firstly, an anisotropic diffusion filter is applied to eliminate noise in the image and generate a multi-band image. Then a method based on an edge detector is used to select the training set. This method minimizes the user intervention during the training process of the SVM. Finally, the 3D volume of the image is segmented by the SVM classifier. The experiments on real MRI have shown that the proposed method allows segmenting the liver with high accuracy. © Springer International Publishing Switzerland 2014.