Intestinal Canal Reconstruction from Video Capsule Image

DESCRIPTION (provided by applicant): The major objective of this proposal is to develop a novel 3D technology to reconstruct the 3D gastrointestinal images captured by video capsule endoscopes. The reconstructed 3D shape can help physicians locate the position of an intestinal lesion and measure its volume. Video capsule endoscopy is a new technology which revolutionizes the diagnosis of small intestinal diseases. Conventional endoscopy tests, such as gastroscopy and colonoscopy, are effective for the upper or lower digestive tract, but achieve very limited success in the 20 feet of small intestine where endoscopic instruments are hard to control and manipulate. Although this new technology brings endoscopic imaging into a new era, technical improvements are still needed for further clinical applications. Discussions held with our collaborators at the Division of Gastroenterology at Cedars-Sinai Medical Center in Los Angeles indicated that the current system for viewing video capsule images needs further improvement. Examples include the lengthy viewing time to analyze the video, difficulty in identifying the location of the findings, and the lack of ways to quantitatively identify the size of the lesion. To overcome these problems, we propose to develop a novel 3D technology called SuperSD which includes three parts: 1) a quick structure from motion technology (SFM) to construct the 3D shape of the intestinal canal from video sequences captured by video capsules; 2) a set of post-processing software tools, including 3D mosaic, 3D compression and visualization which will facilitate the storage and measurement of the reconstructed 3D shape; and 3) a simulation platform which can quantitatively evaluate the accuracy of our algorithm. We will leverage our over ten years experience with 3D imaging to handle these unique challenges in video capsule images.