Automatic boundary tumor segmentation of a liver

Abstract

This paper proposes automatic boundary tumor segmentation for the computer aided liver diagnosis system. As pre-processing, the liver structure is first segmented using histogram transformation, multi-modal threshold, C-class maximum a posteriori decision, and binary morphological filtering. After binary transformation of the liver structure, the image based bounding box is created and convex deficiencies are segmented. Large convex deficiencies are selected by pixel area estimation and selected deficiencies are transformed to gray-level deficiencies. The boundary tumor is selected by estimating the variance of deficiencies. In order to test the proposed algorithm, 225 slices from nine patients were selected. Experimental results show that the proposed algorithm is very useful for diagnosis of the abnormal liver with the boundary tumor. © Springer-Verlag Berlin Heidelberg 2005.