ABSTRACT
Advanced developments in the Internet of Health Things (IoHT) offers a new class of applications and higher competence for present healthcare services. Globally, diabetic retinopathy (DR) is the major cause of loss of eyesight in diabetic patients. Various types of complexities are met by people who have diabetes and DR could be reduced by balancing blood glucose and having regular treatment. Since the DR comes with distinct stages and varying difficulties, it is tedious and time consuming to detect DR. In this approach, it has been deployed with an automatic segmentation-relied classification method for DR. Recently, contrast-limited adaptive histogram equalization (CLAHE) has been employed for preprocessing and the watershed method is used for segmentation of images. Then, the Hough transform (HT)–based feature extraction task is carried out using adaptive neuro-fuzzy inference system (ANFIS) approach to classify the images into diverse grades of DR. In the experimental investigation, a data set was retrieved from the Kaggle website that is assumed to be an open-source environment that aims to develop a DR prediction technique. The maximum classification function has been accomplished by the proposed system along with a higher accuracy of 94.65%, a sensitivity of 80.18%, and a specificity of 97.39% when compared with alternate models.
