Machine learning and deep learning based strategies for diagnosis of diabetic maculopathy from sdoct retinal scans

Abstract

One of the major complications in human eyes due to Diabetes is newlineDiabetic Retinopathy (DR). Diabetic retinopathy occurs when the blood or newlineother fluids leaks from tiny blood vessels of the light-sensitive retinal tissues newlineand accumulate. Due to this the retinal layers gets swollen, resulting in cloudy newlineor blurred vision. It can cause loss of vision if it is left undiagnosed and newlineuntreated. newlineDR includes mainly three stages background retinopathy, newlinepreproliferative stage and proliferative stage. Diabetic Maculopathy (DM) is a newlinecondition of DR and is the main cause of vision loss. In DM, the macula newlineregion, center portion of retina gets affected. DM can occur in any stage of newlineDR, but it likely increases as DR worsens. The chances of occurrence of DM newlineare high in persons with uncontrolled Type 1 or Type 2 diabetes for more than newline10 years duration. newlineRetina is a ten layered structure and the breakdown of blood retinal newlinebarrier leads to Diabetic Macular Edema (DME), which is the cause of DM. newlineInitially, edema starts occurring in the outer nuclear layer or outer plexiform newlinelayer and hence the entire retinal layer thickness increases. In the advanced newlineDM stage, DME may cause several structural changes in the retinal layers. newlineThe crucial task is automated detection of DM in the early stage and newlineidentification of the structural pattern involved in the latter stage. Therefore newlineanalysis of retinal layers is the foremost task in the prevention of vision loss. newline