WITHDRAWN: Hardware-based identification of bright lesions to diagnose diabetic retinopathy

Abstract

Diabetic Retinopathy is a major problem for diabetic patients and is caused by changes in the blood vessels and abnormalities in the macular region. This disease will lead to vision loss if the diabetes is not controlled. This disease can be encountered by the primary signs of Microaneuryusm, Haemorrhages and Exudates. In this paper, a methodology is proposed for Hardware-based Detection of Exudates using RAPIDS CUDA Segmentation algorithm. The Input retinal image is pre-processed by using the Gray scale conversion method and Contrast Limited Adaptive Histogram technique. The pre-processed retinal image is further segmented by using K-mean Segmentation technique and the exudates are detected from the segmented output region. In this proposed approach RAPIDS CUDA Machine Learning algorithm runs completely on the GPU architecture which has increased the speed of the execution with less execution time.