Automatic Segmentation of Liver Tumor Ultrasound Images Based on GGVF Snake

Abstract

In this paper, an approach based on generalized gradient vector flow (GGVF) snake model is proposed for automatic segmentation of liver tumor ultrasound images. According to it the initial contour of GGVF snake can be generated automatically in stead of artificial appointment. The preprocess including anisotropic diffusion filtering and texture classification is implemented on ultrasound images, which ensures the initial contour close to the tumor's real boundaries. The edge map function of GGVF is also modified for obtaining better segmenting performance on ultrasound images. Experimental results show the approach is suitable and effective for segmentation of liver tumor in ultrasound images.