Liver Segmentation in CT Images Using Chan-Vese Model

Liver segmentation in computerized tomography (CT) images has been widely studied in recent years, which generally focuses on segmentation accuracy and processing speed. Of which active models demonstrate a great potential in this field. This paper presents an approach based on Chan-Vese model and other techniques. Firstly, the basic theory on Chan-Vese model is introduced. Secondly, a pre-processing method using Gaussian function is employed to get liver likelihood images for segmentation. Thirdly, an improved Chan-Vese model is proposed to optimize the contour evolution when segmenting each CT slice. Finally, the superior liver region is extracted by applying morphologic operation together with priori knowledge. Experiments on a variety of CT images show its effectiveness and efficiency.