Optic Disc Localization from Retinal Fundus Image Using Discrete Cosine and Hough Transforms

Diabetic retinopathy, resulting from diabetes mellitus, is one of the main causes of global blindness and visual impairment. The integration of artificial intelligence in the analysis of retinal images helps healthcare professionals, with the precise location of the optic disc being essential for detecting vital components of the retina. In this paper, we present a method that uses the cosine transform in conjunction with the circular Hough function. This method also involves converting retinal images from the RGB color space to an adapted grayscale, focusing only on the Red and Green channels. The dataset used for the experiment was RIGA, consisting of the subsets Messidor, Magrabia, and BinRushed. The results showed 100% accuracy for Messidor, 97.87% for Magrabia, and 98.46% for BinRushed, achieving an overall accuracy of 99.33%.