Performance of Deep Learning Based Classification of Retinal Image and Analysis of Effective Diagnosis of Glaucoma

Glaucoma eye condition detected by computer-aided diagnostic systems continues to be a difficult task. During eye assessment, ophthalmologists commonly detect glaucoma by looking for structural defects in the optic disc, nerve fibre loss and peripapillary region atrophy (reffered to as APR). This analysis offers a unique strategy termed Glaucoma-Deep, in contrast to robotic CADx systems that employ segmentation-based hand-crafted features in retinal images for disease assessment. In this method, multilayered characteristics are extracted from actual images using a convolutional neural network. The most unique deep characteristics are subsequently extracted from the labelled dataset for training using a deep network of beliefs (DBN) model. The final conclusion is made using a softmax linear classifier, which can discriminate between retinal pictures with and without glaucoma When evaluating the Glaucoma-Deep system. In summary, the Glaucoma-Deep system is a cutting-edge method for accurately detecting glaucoma eye illness from unprocessed retinal pictures. Glaucoma-Deep proposes a viable approach for overcoming professional specialist issues in massive amounts examination situations by outperforming current methods and providing outstanding sensitivity, specificity, accuracy, and precision metric.