Hybrid multiscale affine and elastic image registration approach towards wireless capsule endoscope localization

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

“…Thus this curve is a graphical tool that quantifies the degree of dissimilarity between consecutive frames in a WCE video and provides the possibility, to a medical doctor, of quickly identifying a set of critical frames, that correspond to the peaks in the curve. Furthermore, as observed in [10], this dissimilarity measure is also an indicator of the speed of the capsule, since low values mean that the scene does not change (so the capsule is almost still or moves very slowly), while high values mean the scene changes (so the capsule either rotates or moves forward due to peristalis).…”

“…A concise description of the adopted approach, that includes the dissimilarity measure, is summarized next, and we refer the reader to [10] for the details.…”

“…In a previous work [10] we proposed a multiscale elastic image registration procedure, that includes an affine pre-image registration, for tracking the motion of the capsule, and an indicator of the WCE speed was also obtained. This particular strategy intends to take into account the combination of two types of movements: the movement of the capsule, which is rigid-like, and the nonrigid movement of the small intestine, which is an elastic organ, that bends and deforms due to peristalsis.…”

“…In the present paper we use precisely the image registration approach and the dissimilarity measure employed in [10], to investigate whether the curve representing this dissimilarity measure is a good identifier of frames with lesions in the WCE video. We focus on the identification of frames with bleeding, in the small bowel.…”

“…After this introduction, the next Section 2 briefly describes the methodology introduced in [10]. In Section 3, the experiments and results, obtained with the aforementioned approach, are reported.…”