Glaucoma stage classification using UNET-based segmentation with multiple feature extraction technique.

According to the World Health Organization, glaucoma is the second biggest cause of blindness globally, with roughly 60 million cases documented worldwide in 2010. Glaucoma is a disease that, if left untreated, may cause irreparable damage to the optic nerve, ultimately resulting in blindness. Examining the optic nerve head, which includes the assessment of the cup-to-disc ratio, is regarded as one of the most important ways of structural diagnosis of the illness in its early stages. Optic Disc (OD) segmentation is a critical stage in analyzing the colour fundus image. In this work, the DRISHTI-GS and LAG images are resized and normalized as part of the preprocessing process. The optic disc is segmented using the trained UNET. Cropping is done to the optic disc following segmentation. Take segmented images and extract the statistical and edge characteristics. The optic image is then classified as normal or glaucoma using the trained KNN classifier. This work achieves an accuracy of 0.997, a sensitivity of 0.986, a specificity of 0.982 for the Drishti-GS, an accuracy of 0.987, and a sensitivity of 0.972 and a specificity of 0.992 for the LAG databases, respectively. [ABSTRACT FROM AUTHOR]