Adapted Bit-Plane Probability and Wavelet-Based Ulcer Detection in Wireless Capsule Endoscopy Images

Abstract

Wireless capsule endoscopy (WCE) has been proven to be a robust technology which is able to ease diagnosing the GI tract diseases. It can be seen that a better computational algorithm is needed to analyze WCE images. Ulcer is one of the several diseases which are diagnosed using these images. Non-uniform lighting can complicate the detection process because it can change the color of tissue and make it seem darker or lighter than usual. This change of color makes the detection harder as the main feature of detecting ulcer as the color of the tissue. In this research work, adapted bit-planes are used to detect useful areas in images and then two sets of features, bit-plane probability and wavelet-based features, were extracted from the detected areas and used to classify them. Experimental results demonstrate a promising ground for further analysis of the channel-based bit-plane data and wavelet-based features.

Multi‐scale analysis of ulcer disease detection from WCE images

“…An approach was introduced in [16] for ulcer recognition in WCE images, in which a saliency detection method based on multi‐level super‐pixel representation to outline the ulcer candidates. Bit‐plane probability and wavelet‐based features are used for the classification of ulcer images in [17]. Unfortunately, these methods have been tested and discarded as their inclusion does not provide a significant performance improvement.…”