Liver Segmentation from 3D Abdominal CT Images

Segmentation of the whole liver region from computed tomography (CT) image is the first step in the computer-aided diagnosis for liver disease. In this paper, we propose a new method for segmenting liver region from 3D CT images of abdomen using enhanced Otsu method. Our algorithm uses Otsu method with some improvements to construct intensity model and shape model for liver. First, the 3D CT image of abdomen is cropped, preserving only the part that contains liver. Second, the liver region in each slice of preserved part is segmented using intensity model. Finally, the segmented liver regions from all slices are post-processed to increase performance and merged into a single liver region.