Research On Key-technique Of Virtual Endoscope In Small Intestine

The small intestine located in the peritoneal cavity, is the longest gastrointestinal tract and it is difficult to do its examination and diagnosis. The better method of examination will have positive effects on assisting doctors on the examination and diagnosis of small intestine. Meanwhile, it can alleviate the burden of the doctor and reduce the suffering of patients.The traditional methods of examination have some problems such as the poor positioning and qualitative effect, and the high false positive results. No matter the capsule endoscope or the optical endoscope also has its restriction in the practical application. However, the virtual endoscope hews out a new road for the diagnosis of small intestine.The aim of this paper is to provide the virtual endoscope technology in allusion to small intestine and allow the doctor to navigate automatically and interactively in the MRI images of small intestine. Firstly, the proposed method exacts the distinct part from the original data and reconstructs the three-dimensional image. Then the method extracts the navigation path in the images, thereby achieves to navigate. For the automatic navigate, firstly, the paper applies region growing to segment the part of small intestine and processes the segmentation based on mathematic morphology; secondly, the proposed method finds the centerline points from the segmented region based on the map of distance from boundary; finally, the path of navigation is determined according to the fixed centerline points. However, the path of interactive navigation is planed with the algorithm of the centerline extraction based on segmentation. Here the region growing is used in the segmentation as well, and in the process of segmentation, we record the growing points orderly thereby determine the path of navigation according these points. The growing consists of inter-slice growing and intra-slice growing and both growing pattern work alternatively and update the same set of growing points, so as to keep the growing points sequent.The path planning algorithm for the automatic navigate proposed in this paper is based on parallel computing of GPU, consequently it supports to navigate in the large segmented region and the algorithm has good reference for virtual endoscope in other organs. Meanwhile the algorithm of the centerline exaction based on segmentation can satisfy the doctors’requirement further and has some practical future.