An Efficiency way to analyse Diabetic Retinopathy Detection and Classification using Deep Learning Techniques

Lesions develop on the retina of the eyes as a result of the progressive eye condition known as diabetic retinopathy (DR), which is brought on by type-2 diabetes. In especially for the working-age population in sustainable nations, it is thought that Diabetes Retinopathy is the main cause of blindness in diabetes patients. The aim of the treatment appears to be to maintain the patient's degree of eyesight because the issue is chronic. Diabetic retinopathy must be accurately identified in order to fully protect the patient's vision. The major issue with DR detection is manual diagnosis, which is time-consuming, expensive, and labor-intensive. A retinal scan of the patient's eyes must also be evaluated by an ophthalmologist as part of the treatment. The latter also appears to be more challenging, particularly in the early phases of the ailment when sickness indications are less obvious in the photos. Early detection of diabetic retinopathy has become easier because of deep learning algorithms, and images of the retinal fundus (DR) may now be analysed using machine learning. There are various stages of diabetic retinopathy, and the early stages are symptomless. Ophthalmologists can spot some retinal issues, but they can't always determine their root causes or stages of development. Ophthalmologists advise retina specialists to treat disease as a result. Bayesian neural networks (BNNs) had been used to benchmark the binary categorization of diabetic retinopathy as referable or non-referable in the current system. We suggest developing a Convolution neural network (CNN) and data analysis method to categorise diabetic retinopathy based on clinical data, predicting whether the patient is diabetic or not and identifying its stage with estimation, employing measurements are needed to maximise the intended performance measure with different datasets and clinical lesion images.