Deep Learning Analysis for detection of Morphological Features of HR & DR

Retinal disorders, which include diabetic retinopathy and hypertensive retinopathy, are some of the main causes of blindness in the world. Early detection would reduce much of the risk of vision loss among all these retinal disorders. The paper creates a new method of automated diagnosis and classification based on fundus images of these diseases. Here, deep learning is combined with morphological features comprising mean, standard deviation, variance, entropy, contrast, skewness, area, and correlation comprehensively extracted to explain retinal pathology. It shall be noted that it actually employs a three-stage classification scheme; it even follows support vector machines as the first classifier of the highlevel classification and further specific and accurate classification with deep learning. The proposed methodology has been validated on benchmark datasets from hospitals, which shows very promising accuracy in differentiating the various retinal disorders. This work puts in evidence the robustness of deep learning techniques in the morphological features regarding HR and DR. Because this work deals with the challenge of manual diagnosis, this work brings in enormous contributions to medical image analysis as it renders more efficiency in early intervention strategies for retinal diseases. As a result, this research will serve as a suitable basis for future developments in the automated diagnosis and treatment of retinal disorders, as prompt delivery of appropriate medical care to patients is guaranteed.