Abstract:
Nowadays, glaucoma is a progressive, persistent, and chronic eye condition that
causes irreparable blindness or visual loss. Globally, around 80 million people will have
glaucoma in 2020, and this number is projected to rise to nearly 111 million by 2040. In
their early stages, glaucoma is characterized by a lack of symptoms, although it
ultimately causes damage to the optic nerve. Review of the scientific literature
demonstrates that glaucoma can be managed prior to vision loss or total blindness if it is
diagnosed early. In past investigations, the most often employed scanning procedures for
glaucoma identification were RNFL analysis and the cup-to-disk ratio. In glaucoma, there
is a growing body of evidence supporting the use of statistical analysis optical coherence
tomography as a supplemental tool for clinical evaluation and research. In order to
diagnose glaucoma early in this study, we apply two main analyses: statistical analysis
and identification of retinal layers. We created an image categorization system employing
OCT images and discrete wavelet transform. After denoising high-resolution OCT
pictures, DWT decomposition is employed to derive statistical characteristics. Statistical
characteristics are extracted using the DarkNet-53 neural network. During the second
phase, all retinal layers are identified by locating the ROI and using a segmentation
technique for each layer. The macular thickness is then computed using the layer borders;
for better results, the retinal layers are also identified manually by licensed
ophthalmologists. The macular thickness is a statistical characteristic. In the final stage,
the outputs of the several blocks are correlated, allowing for the diagnosis of glaucoma
stage. In addition, we wanted to introduce the platform for future measurement of
macular thickness and the potential utility of macular spectral-domain optical coherence
tomography in glaucoma in this study. These procedures will aid experts in the diagnosis
and observation of eye diseases by providing them with accurate and precise information
regarding the structure of the optic nerve head
