Review of features and metafeatures allowing recognition of abnormalities in the images of GIT

Abstract

The wireless capsule endoscopy is a new method of the gastrointestinal tract examination that is now in great demand. It provides fast and comfortable for the patient analysis of the entire digestive tract. Special software for wireless endoscopic capsule allows automated processing of data indicating images with pathologies or suspected abnormality. The software is based on recognition of features denoting the disease. This article provides the description of features and metafeatures allowing recognition of pathologies and structural changes in the images of gastrointestinal tract such as histograms of channels, gastrointestinal tract sections, histograms of texture, form, normalized histogram, histogram quantiles and value calculation for different image parts. Authors also provide the effectiveness testing of software using proposed features and metafeatures.