Academic Journals Database
Disseminating quality controlled scientific knowledge

Integration of Morphology and Graph-based Techniques for Fully Automatic Liver Segmentation

ADD TO MY LIST
 
Author(s): Wan Nural Jawahir Hj Wan Yussof | Hans Burkhardt

Journal: Majlesi Journal of Electrical Engineering
ISSN 2008-1413

Volume: 4;
Issue: 3;
Start page: 59;
Date: 2010;
Original page

Keywords: Computed Tomography | 3D Liver segmentation Graph-cuts | Morphological Operations.

ABSTRACT
Here a fully 3D algorithm for automatic liver segmentation from CT volumetric datasets is presented. The algorithm starts by smoothing the original volume using anisotropic diffusion. The coarse liver region is obtained from the threshold process that is based on a priori knowledge. Then, several morphological operations is performed such as operating the liver to detach the unwanted region connected to the liver and finding the largest component using the connected component labeling (CCL) algorithm. At this stage, both 3D and 2D CCL is done subsequently. However, in 2D CCL, the adjacent slices are also affected from current slice changes. Finally, the boundary of the liver is refined using graph-cuts solver. Our algorithm does not require any user interaction or training datasets to be used. The algorithm has been evaluated on 10 CT scans of the liver and the results are encouraging to poor quality of images.
Affiliate Program      Why do you need a reservation system?