Efficient pancreas segmentation in computed tomography based on region-growing

Abstract

Pancreas segmentation in computed tomography data is one of difficult problems in medical area. Segmentation of pancreas tissue in computed tomography is difficult even for human, since the pancreas head is always directly connected to the small bowel and can in most cases cannot be visually distinguished. In this paper, an efficient method to extract the pancreas from such computed tomography images is proposed. Histogram equalization is used to enhance the contrast of computed tomography images. After that, regiongrowing technique is applied to label pancreas region and return the result of segmentation. The proposed method will be experimented and evaluated by using Jaccard index between an extracted pancreas and a true one. For evaluating the proposed method, we have compared the results of our proposed method with the other recent methods available in literature.