Computer-Aided Bleeding Detection in WCE Video

Abstract

Wireless capsule endoscopy (WCE) can directly take digital images in the gastrointestinal tract of a patient. It has opened a new chapter in small intestine examination. However, a major problem associated with this technology is that too many images need to be manually examined by clinicians. Currently, there is no standard for capsule endoscopy image interpretation and classification. Most state-of-the-art CAD methods often suffer from poor performance, high computational cost, or multiple empirical thresholds. In this paper, a new method for rapid bleeding detection in the WCE video is proposed. We group pixels through superpixel segmentation to reduce the computational complexity while maintaining high diagnostic accuracy. Feature of each superpixel is extracted using the red ratio in RGB space and fed into support vector machine for classification. Also, the influence of edge pixels has been removed in this paper. Comparative experiments show that our algorithm is superior to the existing methods in terms of sensitivity, specificity, and accuracy.

Software for enhanced video capsule endoscopy: challenges for essential progress

“…Current methods to detect haemorrhage mainly include variations of this approach with respect to colour representation and the size and shape of the image regions, for example, so-called Key points ■ Computational software can enhance the diagnostic yield of video capsule endoscopy (VCE), both in terms of efficiency and accuracy ■ Despite increasing activity in information technology (IT) research worldwide, the translation of this information to clinical practice has been limited ■ The development of intelligent software systems requires close collaboration between medical and IT scientists at a laboratory level ■ Public sharing of anonymized and annotated VCE image and video data is essential pyramid histograms extracted from subimages iteratively sampled at various resolutions. 24 The latest approaches consider regions of arbitrary shape that are automatically selected by image segmentation algorithms, such as region-growing (the algorithm begins from an initial set of scattered pixels, iteratively selects neighbouring pixels and eventually contiguous regions within the image) 25 or super-pixel (the algorithm groups pixels into perceptually meaningful image regions, which can be used to replace the rigid structure of the pixel grid) 26 for segmentation of the blood regions in the VCE images. In addition, promising results have been obtained by methods that evaluate colour at the pixel level rather than at the region level.…”

Saliency Aware Locality Constrained Linear Coding for WCE Multi Abnormality Detection based on SURF

“…Many works have been made in the literature for automatic abnormal image detection from the WCE videos [8]- [14]. The common diseases seen in the GI tract are typically bleeding, polyp, and ulcer, as shown in Fig.…”

“…They segmented images through super pixel segmentation and feature of each pixel is extracted using the red color ratio in RGB color space Yeh et al proposed [8] a novel method for detecting bleeding and ulcers in WCE images by using color features to determine the status of the small intestine. Charisis et al [11] introduces a Discrete Curvelet Transform (DCT), method which calculates the lacunarity index of DCT sub bands of the WCE images for acquiring the textural information's to detect the ulcer images. Yuan and Meng [10] proposed a novel texture feature that integrates the advantages of both Gabor filter and monogenic local binary pattern (LBP) methods.…”

A Survey on WCE Abnormality Detection Methods

“…Ulcer mainly corresponds to the saliency region, so saliency max-pooling method integrated with the Localityconstrained Linear Coding (LLC) method is used for characterization of the images. Yanan Fu et al [3] proposed a new method which is able to detect bleeding regions from WCE video more effectively and efficiently. Because edge pixels and bleeding pixels share similar hue, traditional algorithms often mistake edge pixels for bleeding pixels.…”