Analysis, Processing And Three-dimensional Reconstruction Of Vessel Structure In Medical Fundus Images

With1970s’ CT （Computerized Tomography,）, MRI （Magnetic Resonance Imaging,MRI）and other medical imaging techniques applied in medicine, the doctor could analyze thelesion part of the patient according to a group of two-dimensional images of the lesion part,which made a huge improvement in the medical diagnostic accuracy. However, it was hard toestablish a three-dimensional structure only according to the two-dimensional images, and theimage data was not intuitive and dimensional. So the clinical diagnosis was mostly made bythe doctor’s clinical experience, and it was easy to get the mistaken diagnosis. In order toimprove the scientificalness and accuracy of medical diagnosis and treatment,3Dreconstruction and image processing are becoming an important breakthrough and keytechnology.This paper discusses the feature extraction, three-dimensional reconstruction and displaymethod of human fundus vessel system. High resolution skeleton and outline image of fundusvessel can be obtained by pre-processing and feature extraction of two-dimensional fundusimage, then we can obtain high-resolution retinal vascular image of skeleton diagrams andcontour maps, then create the retinal surface and retinal vascular3D data model. Based on thisthree-dimensional reconstruction model, finally, D3D was applied to add material and renderlight, thus we can finally get the three-dimensional image of the retinal vascular system. D3Dwas applied to add material and render light, after having drawn the model of fundus system.In this way, we completed the feature extraction and3D display of the retinal vascular image,All of the deep studies on fundus image information are based on obtaining HR fundusimage by applying medical equipments. During the process, due to the limitations of moderntechnological level, the original image has some shortcomings, like locality, brightness fuzzy,uncertain boundary. In order to overcome the above problems and prepare well for thefollowing reconstruction, we can not directly use the original medical images as materials, butgo through a series of pre-processing work, including filtering, enhancing, registering and soon. Comparing with the original material, the image after pre-process has clearer texture,lower noise, higher quality. Therefore, the following extraction information is easily obtained,and the result is more effective.Fundus vascellum is the only megascopic blood vessel system. And recent years, theresearch on the computer simulation of fundus vascellum model has increasingly deepened.This paper starts with the chosen feature points of fundus vascellum, and passing the feature points, drawing a blood vessel section round vertical to vascular axis. Because the chosenfeature points is separated, the obtained section round has certain interval. Then thetheoretical model of fundus vascellum will be obtained after sequentially and smoothlyconnected the obtained section round.As for the3D model’s display, this paper adopts the DX method based on the VC++torealize3D reconstruction. First, it reconstructs the3D fundus vascellum model, next endowswith the features like model material and lightning, and then displays them on computer. The3D model obtained in this way, generally speaking, reflects the situation of fundus vascellumreally and accurately. It will facilitate the medical personnel to observe blood vesselomnibearingly, and pathological changes. It will be easily applied to medical domain.At present, a lot of research institutions in the world are committed to the research in thisfield. The3D visualization technology has been applied in radiation therapy, diagnosticmedicine, plastic and prosthetic surgery, and the like. Although the applications in these areasare relatively simple and have limitations, they play an important role in medical diagnoseand techniques. It is this reason that makes the research draw great attentions from manycountries. Hence the Medical Image3D Reconstruction and Processing Technologyundoubtedly become a hot spot in the biomedical engineering research. The studied domain isso complex and mysterious and involves in wide fields and disciplines, so there are still manyaspects worth exploring and putting forward more constructive and pertinent new theories andmethods.