Vertical and horizontal cup-disc ratio (CDR) is an important constraint, which is utilised clinically for detecting glaucoma or to scrutinise its progression and is evaluated manually from fundus images. Owing to the scarcity of glaucoma specialists and owing to the rise in glaucoma's affected persons, an automated approach can be much helpful for screening glaucoma. Here, the study intends to create a new glaucoma detection scheme with 4 phases. Initially, preprocessing is done via median filtering of the input image. Then, optic cup segmentation is done via an enhanced U-net method. Moreover, features like cup to disc ratio (CDR), inferior superior nasal temporal (ISNT) features, fractal features and proposed local Gabor features are determined. These characteristics are categorised using an improved version of the LSTM, and the weights are adjusted using the shark smell algorithm with sigmoid function-based new position update (SSA-SFNPU). Further, the examination is done regarding wide-ranging metrics.
Singh et al. (Thu,) studied this question.