Automatic segmentation of the liver in CT images using a model of approximate contour

Abstract

The segmentation of the liver structure from a computed tomography (CT) image is an important function of the software designed to assist liver diagnostics, because it allows for the elimination of excess information from the diagnostic process. In this paper, the task of segmentation has been implemented through first finding the contour of the liver which is made up of a finite number of joint polylines approximating individual fragments of the liver boundary in the CT image. Next, the field outside the contour is divided into two polygons and eliminated from the image. The initial reference point for the calculations is the lumbar section of the spine which is a central point of any CT image of the liver. The automatic method of segmentation is to be used in a dedicated computer system designed to diagnose liver patients. © Springer-Verlag Berlin Heidelberg 2006.