Automatic extraction of liver regions and classification of liver cirrhosis from the abdominal CT images

We propose a new technique for automatic extraction of liver region and its classification for liver cirrhosis by computer based on a set of abdominal CT images. In this paper, we extract a liver area for extraction of the region of interest employing smoothing, binarization, expansion, shrinking, and labeling process from the inputted each slice images. The region having the largest area is regarded as a tentative liver region. Furthermore, to delete surrounding tissues, we create a new liver region from the estimation of a slice having the greatest square employing Mahalanobis distance. Finally, the liver cirrhosis is classified from the liver region with respect to the abdominal CT images. Experiment is performed employing twenty abdominal CT image sets and the obtained results are shown along with a discussion.