Deep Learning Solution for Diabetic Retinopathy Diagnosis based on Convolutional Neural Networks and Image Processing Algorithms

Abstract

The purpose of this paper is to reflect upon the results obtained for diabetic retinopathy diagnosis through the implementation of high-performance computing algorithms, libraries and programming languages. Their subsequent implementations aim to help medical practitioners in a precise determination of symptoms related to diabetic retinopathy, with minimal errors and with the support of Machine Learning and Image Processing Algorithms. The solution was developed based on Python and the NumPy, Pandas, Tensorflow, Keras and Pillow libraries. The correctness of the diagnosis varies from 70% to 77%.