Performance Analysis of Automated Detection of Diabetic Retinopathy Using Machine Learning and Deep Learning Techniques

Abstract

Diabetic Retinopathy is related to a combination of eye disorder due to difficulty of mellitus. This disorder leads to complete blindness or vision loss. Automated methods for detecting and classifying the type of disease into normal or abnormal have important medical application. Here, deep learning and machine learning techniques are used to classify a given set of images into normal or abnormal classes. In machine learning section, local binary pattern (LBP) technique is used for feature extraction. Random forest (RF) and Support Vector Machine (SVM) are the two best machine learning algorithms taken for classification purpose. AlexNet, VGG16 and Long Short Term Memory (LSTM) are used as the deep learning techniques. Single algorithms are optimise with respect to their parameters, and are compare the parameters in terms of their accuracy, sensitivity, specificity, precision and F1-score. The accuracy of SVM, RF, AlexNet, VGG16 and LSTM were found to be 88.33%, 94.16%, 98.35%, 99.17% and 97.5%. Also, the performance evaluation table of machine learning and deep learning algorithms were tabulated using these parameters.