Segmentation of Small Bowel Tumors in Wireless Capsule Endoscopy Using Level Set Method

Abstract

In this paper, we proposed an algorithm to segment small bowel tumors. In order to increase effectiveness of Level Set Method (LSM) we applied adaptive gamma correction method (AGCM) that is based on prior information of illumination of images. We applied this method on 10 small bowel tumor images captured by Wireless Capsule Endoscopy (WCE). The performance measurements (i.e. sensitivity, specificity, and accuracy) by using hand ground method are computed for different parameters of a (0.05, 0.07, 0.09, 0.11, and 0.13) in AGCM, and then compared with traditional LSM and Snake method. The proposed method shows increased sensitivity up to 0.87 in a=0.13 while other performance measurements decrease by increasing value of a. the sensitivity of the other methods are 0.2 and 0.22, respectively. The optimal value of these measurements is 0.73 that takes place in a=0.1.