Diabetic retinopathy (DR) poses a large economic burden on the healthcare system with nearly 1 in 10 diabetes patients developing vision-threatening DR. Early detection, accurate evaluation and timely treatments are effective in preventing blindness. However, the ability to implement this is limited by the manual nature of examining patient’s retinal images. This is especially a problem in Asia due to rising incidence of diabetes, the absence of a CAD tool for automated detection of DR and the lack of centres specializing in fundus image grading. Clearly, there is a great need for an automated DR detection tool.
SELENA employs the latest image analysis and state-of-the-art machine learning techniques to serve as an automated, real-time detection tool that is able to match human grading. It has undergone extensive validation testing to screen DR by automatically classify the diabetic patients into those who need medical referral, and do not need further assessment or treatment.
Final optimization of SELENA expanding its eye dataset to determine the minimum requirement and specification of input image characteristics; plus a human factor study to determine user requirement, leading to a robust product and performance specification for a commercially viable SELENA.
Principal Investigator: Professor WONG Tien Yin
Institution: Singapore Eye Research Institute
NHIC Ref: NHIC-I2D-1409022