Automatic detection of the macula in the retinal fundus image by detecting regions with low pixel intensity

Abstract

This paper proposes a method to detect the macula in the retinal fundus image automatically. The method makes use of the optic disc height obtained from the ARGALI to define the region of interest. Regions of dark spots are then detected by finding the coordinates with the lowest pixel intensity and determining the average pixel neighbourhood intensities. These regions are ranked to determine the region containing the macula. This algorithm was tested on 162 images, and an accuracy of 98.8% was achieved. The results are promising for further development and use of this method in AMD studies and physiology localization.