Structure-preserving guided retinal image filtering for optic disc analysis

Abstract Retinal color fundus images have been widely used for the diagnosis of ocular diseases such as glaucoma, age-related macular degeneration, diabetic retinopathy, and pathological myopia. In recent years, many techniques have been developed for the automatic analysis of the retinal color fundus images. In this chapter, we first give a review of the recent technology for optic disc analysis including optic disc and optic cup segmentation. After that we introduce our newly developed technology that addresses an important issue in the analysis (i.e., the image quality). We introduce a method to overcome the issue of low image quality due to the disease like cataract and discuss its application in automatic disc analysis. The proposed method approximates the degradation of the retinal images as a combination of human-lens attenuation and scattering. It includes a step is a global structure transferring and a step of global edge-preserving smoothing. We apply the algorithm to process the images for subsequent analysis tasks. The experimental results show that the method enhances the contrast of images. Meanwhile, it also makes the optic analysis tasks such as optic cup segmentation and cup-to-disc ratio computation more accurate. The method has potential to improve disease detection such as glaucoma.