Optical Disc Segmentation from Color Fundus Image using Contrast Limited Adaptive Histogram Equalization and Morphological Operations

Various ocular diseases, like Cataract followed by Glaucoma and Diabetic Retinopathy, are the world’s leading causes of blindness and affect millions of people every year. Rapid and automated testing is therefore essential for assisting in early discernment and analysis. The traditional method of screening for such diseases involves manual analysis of the fundus image obtained by ophthalmologists whose versions of determining the region-of-interest vary. Segmenting the disc region serves as the first and crucial step before feature extraction, for the diagnosis of a variety of pathologies and abnormalities related to the retina. For automated analysis, various features, such as cup to disc ratio (CDR), area cup-disc ratio (ACDR) etc., are estimated to foretell the onset of such diseases. Our paper proposed and explored a new segmentation technique based on the Contrast Limited Adaptive Histogram Equalization (CLAHE) method and mathematical morphology for optic disc discernment. The analysis was done on a freely available database, DRISTI-GSI. The segmentation results of the proposed algorithm showed an average accuracy of 94.7% after testing and verification.