Abstract:Wireless capsule endoscopy (WCE) is a video technology to inspect abnormalities, like bleeding in the gastrointestinal tract. In order to avoid a complex and long duration manual review process, automatic bleeding detection schemes are developed that mainly utilize features extracted from WCE images. In feature-based bleeding detection schemes, either global features are used which produce averaged characteristics ignoring the effect of smaller bleeding regions or local features are utilized that cause large feature dimension. In this paper, pixels of interest (POI) in a given WCE image are determined using a linear separation scheme, local spatial features are then extracted from the POI and finally, a suitable characteristic probability density function (PDF) is fitted over the resulting feature space. The proposed PDF model fitting based approach not only reduces the computational complexity but also offers more consistent representation of a class. Details analysis are carried out to find the best suitable PDF and it is found that fitting of Rayleigh PDF model to the local spatial features is best suited for bleeding detection. For the purpose of classification, the fitted PDF parameters are used as features in the supervised support vector machine classifier. Pixels residing in the close vicinity of the POI are further classified with the help of an unsupervised clustering-based scheme to extract more precise bleeding regions. A large number of WCE images obtained from 30 publicly available WCE videos are used for performance evaluation of the proposed scheme and the effects on classification performance due to the changes in PDF models, block statistics, color spaces, and classifiers are experimentally analyzed. The proposed scheme shows satisfactory performance in terms of sensitivity (97.55%), specificity (96.59%) and accuracy (96.77%) and the results obtained by the proposed method outperforms the results reported for some state-of-the-art methods.

Bleeding Detection in Wireless Capsule Endoscopy Images Based on Color Invariants and Spatial Pyramids Using Support Vector Machines

Bleeding Detection in Wireless Capsule Endoscopy Image Video Using Superpixel-Color Histogram and a Subspace KNN Classifier.

Computer-aided Bleeding Detection in WCE Video.

Bleeding detection in wireless capsule endoscopy based on MST clustering and SVM

Bleeding Detection from Wireless Capsule Endoscopy Images Using Improved Euler Distance in CIELab

Detection of Bleeding Image in Wireless Capsule Endoscope Based on Superpixel Segmentation

[A Method for Bleeding Detection in Endoscopy Images Using SVM].

Design of the Color Vector Similarity Coefficient Classifier for Bleeding Detection in Wireless Capsule Endoscopy Images

Saliency-Based Bleeding Localization for Wireless Capsule Endoscopy Diagnosis

Detection of Bleeding in Wireless Capsule Endoscopy Images Using Range Ratio Color

A Deep Convolutional Neural Network for Bleeding Detection in Wireless Capsule Endoscopy Images

Blood detection in wireless capsule endoscope images based on salient superpixels

A Study on Automated Segmentation of Blood Regions in Wireless Capsule Endoscopy Images Using Fully Convolutional Networks

Detection of Protruding Lesion in Wireless Capsule Endoscopy Videos of Small Intestine.

Feature selection and classification for Wireless Capsule Endoscopic frames

Segmentation of Bleeding Regions in Wireless Capsule Endoscopy Images an Approach for inside Capsule Video Summarization

Probability density function based modeling of spatial feature variation in capsule endoscopy data for automatic bleeding detection

Gastrointestinal Bleeding Detection In Wireless Capsule Endoscopy Images Using Handcrafted And Cnn Features

Research on a New Feature Detection Algorithm for Wireless Capsule Endoscope Bleeding Images Based on Super-pixel Segmentation.

Bleeding Detection in Wireless Capsule Endoscopy Based on Probabilistic Neural Network.

GI Bleeding Detection in Wireless Capsule Endoscopy Images Based on Pattern Recognition and a MapReduce Framework