Reconstruction of vascular structures in retinal images

Abstract

Vessels in retinal fundus images are useful in revealing the severity of eye-related diseases. In addition, they can act as landmarks for localizing lesions and the central vision area, and guide laser treatment of neovascularization. In this paper, we propose a two-stage method to identify and extract vascular structure. First, the vessels are enhanced by mathematical morphology with respect to their spatial properties. These vessels are differentiated from background patterns through the use of curvature evaluation and linear filtering. However, this may result in missing out some important features of bifurcation and intersection points. To overcome this, a reconstruction process is performed using dynamic local region growing to recover the complete vascular structure. Experiment results indicate that the proposed method is able to detect the vascular structure in 97% of 35 retinal images with acceptable quality.