Automatic 3D Segmentation of Liver Tissue from CT Datasets

Abstract. Manual segmentation of liver tissue from computerised tomography (CT) datasets can provide useful information to clinicians, such as an estimation of the volume of the liver and the quantification of abnormalities. However, manual segmentation is a slow, laborious process, and an automatic segmentation method has potential to assist in both the diagnosis of disease and in treatment planning. This paper presents initial results from work that extends on previous 2D segmentation methods by implenting full 3D liver segmentation, using a selfreparameterising active contour model. 3D liver segmentation has the advantage over 2D techiques, as the whole liver dataset is analysed at once, rather than as a series of individual slices. Initial results are presented showing volumetric analysis of four liver datasets, assembled from over 500 CT image slices. Further work on improving the segmentation technique, and methods to validate the results, are then discussed in detail.