Glaucoma Diagnosis Based on Both Hidden Features and Domain Knowledge through Deep Learning Models

Glaucoma is a chronic eye disease that results in loss of vision. Since the disease can not be cured, it is necessary to identify the disease in time. This work proposes segmentation of the optic disks and optic cups using superpixel screening. This uses frames from the 2D fundus. In the segmentation of the optical disk, clustering algorithms are used to classify each superpixel as a disk or non-disk. In addition to clustering for CNN algorithms, the gabor filter is included in the feature space for optical cup segmentation. The segmented optical disk and optical cup are then used to calculate the cup-to-disk ratio in glaucoma screening. The color retinal lesion camera CDR is the primary marker for confirming glaucoma in a particular person's healthy or glaucoma phase.