Dissimilarity Measure of Consecutive Frames in Wireless Capsule Endoscopy Videos: A Way of Searching for Abnormalities

Abstract

In a previous work we have shown that the curve representing the dissimilarity measure between consecutive frames of a wireless capsule endoscopic video of the small bowel, obtained by means of an image registration method, can be regarded as a rough indicator of the speed of the capsule, and simultaneously, it is also a valuable auxiliary medical tool. In effect, this curve enables a global and fast interpretation of the video, in the sense that it clearly divides the video frames into two main categories: consecutive frames with similar content, which correspond to low values in the curve, and consecutive frames displaying abrupt changes in the image content, which are depicted by peaks, i.e. high values, in the curve. As the main goal of a wireless capsule video examination consists in searching for abnormal features in the images, the purpose of the present work is to analyse whether this curve can also be used to search, quickly, for abnormalities. The experiments performed focus on bleeding identification in small bowel images.