Implementation of Diabetic Retinopathy screening using realtime data

Abstract

Based on data obtained from the Ministry of Indonesia Health in 2012, Diabetes mellitus (DM) was ranked as the sixth leading cause of death. In a higher phase, it can cause complications in other organs such in retina. A disease that attacks the retina is Diabetic Retinopathy (DR). Although it is not a deadly disease, but patient can suffer permanent blindness, which can affect mental and social condition. Therefore built a system that can detect early. So the initial inspection can be faster. This Research uses data from Dr. Cipto Mangunkusumo Hospital (RSCM) Jakarta, Indonesia and DiaretDB0 as comparative data. This research uses knowledge, Fast Adaptive Histogram Equalization as pre-processing method, and Extreme Learning Machine as classification methods from previous research. Feature used is the area of blood vessels. In the previous research, DiaretDB0 reached 97,5%. After that we use these knowledge for RSCM data. But in the experiment, it reached only 52% for testing accuracy, lower than DiaretDB0. The contribution from this research is analysis why the result from both data is different and suggestion for the next research.