Pancreas Segmentation in Medical Images Based on Gaussian Mixture Model in Bandelet Domain

Abstract

The pancreas is one of important organs in human body. The pancreas segmentation from medical images is a difficult task. Because adjacent organs have the similar intensity values. In this paper, we proposed a new method for pancreas segmentation in bandelet domain. Our method includes two stages. Firstly, applying the Gaussian mixture model for foreground and background is estimated using marked pixels by the user in bandelet domain. Finally, that is pancreas segmentation in medical images in the first stage by objcut. Our dataset is pancreas images that have been collected in many hospitals. We undervested the proposed method by calculating the Jaccard index values. The results of the proposed method are better than the other recent methods.