Abstract
Automating screening and diagnosis in the medical field saves time and reduces the chances of misdiagnosis while saving on labor and cost for physicians. With the feasibility and development of deep learning methods, machines are now able to interpret complex features in medical data, which leads to rapid advancements in automation. Such efforts have been made in ophthalmology to analyze retinal images and build frameworks based on analysis for the identification of retinopathy and the assessment of its severity. This paper reviews recent state-of-the-art works utilizing the color fundus image taken from one of the imaging modalities used in ophthalmology. Specifically, the deep learning methods of automated screening and diagnosis for diabetic retinopathy (DR), age-related macular degeneration (AMD), and glaucoma are investigated. In addition, the machine learning techniques applied to the retinal vasculature extraction from the fundus image are covered. The challenges in developing these systems are also discussed.
Highlights
Machine learning methods have been developed and exploited for process automations in many fields and have recently taken a leap forward thanks to the feasibility of deep learning and storage of massive amounts of data
Unlike the extraction of retinal landmarks, building an automated diagnosis and screening system is much more complex work because it should accompany the analysis of the corresponding lesions: exudate, hemorrhage, MA for diabetic retinopathy (DR); drusen, depigmentation, and geographic atrophy (GA) for age-related macular degeneration (AMD); and glaucomatous optic neuropathy (GON) for glaucoma
Its symptoms are commonly referred to as glaucomatous optic neuropathy (GON) and includes many features that indicate structural damage, such as thinning or notching of the neuro-retinal rim, retinal nerve fiber layer (RNFL) thinning, disc hemorrhage, parapapillary atrophy (PPA), increased or asymmetric cup-to-disc ratio (CDR), the nasalization of the central optic nerve head (ONH) vessels, the baring of circumlinear vessels, excavation of the optic cup, and laminar dots [119,120]
Summary
Machine learning methods have been developed and exploited for process automations in many fields and have recently taken a leap forward thanks to the feasibility of deep learning and storage of massive amounts of data. The machine learning applications for images in the medical field include lesion detection, automatic diagnosis, medical image segmentation, and medical image generation. Fundus imaging devices use an intricate microscope with an installed image sensor that records the reflected light from the interior surface of the eye This optical system allows physicians to observe the major biological landmarks inside the eye, as well as the complex background patterns that are created by the inner retinal structures. The fundus image is the reflection of the interior surface of the eye, and it is normally recorded by image sensors, usually in three colors It includes information about the observable biological structures, such as the surface of the retina, retinal vasculature, the macula, and the optic disc. The red spectrum is only related to the choroidal layer beneath the pigmented epithelium, and contains content about the choroidal ruptures, choroidal nevi, choroidal melanomas, and pigmentary disturbances [25]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
