Stochastic Approach for the Identification of Retinopathy of Prematurity

• Main Authors: Prabakar, S., Porkumaran, K., Karthikeyan, R., Isaac, S., Kannan, R., Nor, N.M., Elamvazuthi, I.
• Format: Conference or Workshop Item
• Published: Institute of Electrical and Electronics Engineers Inc. 2021
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85124171321&doi=10.1109%2fICIAS49414.2021.9642618&partnerID=40&md5=b00a0f51ae97b5edc0075b357e7f5930; http://eprints.utp.edu.my/29178/

Retinopathy of Prematurity (ROP) is a retinal vessel growth condition that affects premature infants who usually are under a birth weight of 1,500 grams and have a gestational age of 32 weeks or less. The back portion of the eye, retina captures light and sends signals to the brain, creating the vision of eye. Blood vessels supply oxygen and nutrients to the retina and these blood vessels begin to develop at 16-18 weeks after conception, and complete with full maturity before the normal birth. Premature birth suffers the development of the retina vasculature and leaves some of the retina without blood vessels, which causes of visual loss in childhood. Serial RetCam images are acquired from premature infants. By applying three pre-processing techniques such as green colour plane, histogram equalization, Contrast Limited Adaptive Histogram Equalization (CLAHE), we can get the best contrast between the vessels and background of retina. After segmentation, watershed transform is applied, and then six statistical features are taken out from that image. Finally the images are categorized as normal or abnormal by using ANFIS (The Adaptive Neuro-Fuzzy Inference System) classifier. © 2021 IEEE.