Liver Segmentation using Bidirectional Region Growing with Edge Enhancement in NSCT Domain

Abstract

Segmentation of liver from abdominal CT image seems to be a challenging task in the computer aided diagnosis of liver pathology. This is due to the fact that liver and the adjacent organs in the CT image possess similar intensity attributes. In this work, a semi-automatic segmentation of liver is carried out through bidirectional region based segmentation. Prior to segmentation, the edge regions of the liver are enhanced by means of unsharp masking in Non Sub-sampled Contourlet Transform (NSCT) domain. This ensures that there is a precise differentiation between the liver region and the adjacent organs and other structures like ribs. From the acquired results, it is clear that the combination of bidirectional region growing segmentation and edge enhancement in NSCT domain outperforms the segmentation by region growing algorithm and bidirectional region growing algorithm without edge enhancement.