Liver Metastasis Segmentation by Anisotropic Diffusion Filtering, Chan-Vese Algorithm, and Kmeans Clustering

Computed Tomography (CT) imaging is used extensively for many purposes such as diagnosing liver diseases and measuring the volume for surgery. In order to perform computer aided diagnosis for the liver and lesion, segmentation procedures have to be performed as a fundamental primary step in order to derive quantitative biomarkers for computer-aided decision support systems. Segmentation of the lesion is not an easy automatic task because the surrounding normal tissues around the liver have close intensity values to the liver which can lead to over segmentation. In order to solve that problem and segment the lesions accurately, the liver has to be segmented first. This paper demonstrates a method to segment liver lesions automatically in CT abdomen images by using Anisotropic Diffusion filtering, Chan-Vese algorithm, K-means clustering, and morphological binary operations.