Emulation on Motion Tracking of Endoscopic Capsule inside Small Intestine

Video capsule endoscopy (VCE) provides a noninvasive way to examine the intestinal lumen as the capsule moves along the gastrointestinal (GI) tract. However, this technology is unable to localize itself when an abnormality is found by the video source. Knowing how the capsule moves will greatly enhance the accuracy in localizing the capsule. In this paper, we develope a novel image processing technique to track the motion of the endoscopic capsule. The proposed method is based on analyzing the displacements of feature points between consecutive images captured by the camera while traveling through the GI tract. To validate the performance of our algorithm, we created a virtual cylindrical tube that looks exactly the same as a small intestine and placed a virtual camera inside the tube to emulate the transition of the video capsule. By processing the emulated images, our proposed method was proved to be able to accurately estimate the linear transition, tilt, and rotation of the video capsule. Index Terms — Video capsule endoscopy (VCE), motion tracking, localization, feature matching.