Detection of Hookworm infection in the wireless capsule endoscopy images

摘要

The Wireless Capsule Endoscopy is a fast growing medical technology to diagnose the internal parts of the human. It paved the way to clearly analyse the organs of the human digestive track in an efficient way. The Wireless Capsule Endoscopy is used to identify the abnormalities like Bleeding, Ulcers, Crohn's Disease etc, in the human digestive track. There are more than thousands of images from the Capsule Endoscopy. Therefore, It is difficult for the physicians to view and analyse each and every imge. Thus, Image Processing Techniques are implemented to automatically detect the abnormalities. In this proposed work, the blood sucking parasite called as Hookworm is detected and localized. Initially, during the pre-processing the colored images are converted to gray-scale images. Then, the region of interest is segmented from the gray-scale images. Later, once the necessary informations are obtained from the segmented image, using the GLCM properties, the images are classified into Normal and Abnormal using the Back propogation neural networks. Inorder to localize the Worm in the digestive track, the Image Enhancement techniques like Adaptive Histogram and Morphological operations are used. Thus, the Hookworm Infection is detected and localized in the digestive track of the human, using the Image Processing techniques.