Establishing the macular grading grid by means of fovea centre detection using anatomical-based and visual-based features

Abstract

This paper presents a methodology for establishing the macular grading grid in digital retinal images by means of fovea centre detection. Usually, visual and anatomical feature based criteria have been used separately for fovea segmentation. However, the combination of the benefits of both techniques constitutes the core methodology presented throughout this paper. Firstly, an acceptable fovea centre estimation is obtained by using a priori-known anatomical feature with respect to the optic disc and the vascular tree. Secondly, a type of morphological processing attempts to improve the obtained fovea centre estimation when the fovea is detectable in the image; otherwise it is declared indistinguishable and the first result is retained. The methodology was tested on the MESSIDOR and DIARETDB database using a distance criterion between the obtained and the real fovea centre. Fovea centres in the brackets between Excellent-Fair (fovea centres primarily accepted as valid in the literature) made up 97.42% and 95.51% of all cases in the MESSIDOR and DIARETDB, respectively.