Research Of Image Processing Techniques In Optical Coherence Tomography Retinal Images

Changes in the structure of the retina can reflect a variety of pathological,physiological changes and the damage of end-organ.Study on the segmentation algorithm of retinal layer and analyzing the specific layer structure have important role in the clinical diagnosis and assessment of treating diseases such as diabetes,high blood pressure,and arteriosclerosis.Based on the retinal layer segmentation,research and analysis on the retinal vascular networks,can effectively help ophthalmologists locate the lesion area.It will be helpful to clinical diagnosis and treatment of retinal diseases.In this thesis,we present a novel algorithm for segmentation of optical coherence tomography retinal images.On the basis of this,we extracted the retinal tissue in the specified location.The vascular bifurcations were detected by using optical coherence tomography retinal images and corresponding fundus images.This thesis presents a retinal segmentation method based on the statistical information of retinal thickness and the shortest path algorithm.The statistical information of retinal thickness was used to limit the search region of the shortest path algorithm.The retinal layer boundaries were detected by iterated using shortest path algorithm.In addition,vascular bifurcations detection method based on transfer learning model were presented in this paper.The feature point detection problem was converted to the corresponding classification problem.By using the transfer learning model Adaptive Support Vector Machine(A-SVM),the vascular bifurcations were detected by optical coherence tomography retinal images and corresponding fundus images.