Edge enhancement of liver CT images using non subsampled shearlet transform based multislice fusion

Segmentation of liver from the computed tomogram (CT) images of the abdomen is a challenging task as the liver boundaries are weak in nature. Multi slice fusion of abdominal CT images based on Non Subsampled Shearlet Transform (NSST) is proposed as an edge enhancement technique. Prior to liver segmentation, two adjacent slices of liver CT images are decomposed by means of NSST in different scales and in different directions to obtain low and high frequency sub band coefficients. The information contents of the low and high frequency sub bands are fused using fusion rules based on phase congruency and directive contrast respectively. To achieve better segmentation accuracy, Sum Modified Laplacian (SML) is integrated with the contrast features measure for effective fusion of edge information. Finally the edge enhanced image is obtained using NSST reconstruction. The effectiveness of the proposed edge enhancement technique is quantified by performing segmentation with and without fusion process. From the acquired results, it is obvious that the proposed fusion framework capture the edge detail information of the source images very effectively and can provide better edge information than the one obtained through individual images, thereby achieving improvement in segmentation accuracy by 60.11%.