Automated system for OD detection in funds Images (P-0086) (MFN 4659)

Abstract

Optic disc is an important part of retina and it will get affected in case of different eye diseases. So it is important to detect the optic disc accurately to detect that disease. The optic disc changes its size in the retinal diseases such as glaucoma and retinopathy so it is important to find it out automatically by designing automated system. I have designed computer aided system that detect optic disc from colour retinal fundus images. The proposed system has four parts i.epreprocessing, vessel extraction, bright region extraction and measurement of vessel density in bright regions. In preprocessing, the background has been extracted to find out the fundus part of image.In second step, bright regions are detected because optic disc has the brightest region. The next step is extraction of vessels in the image because optic disc has maximum vessels density and the last step is to check which bright spot has maximum vessel density and it is marked as optic disc. The system is tested using five different publicly available databases i.eSTARE, AFIO dataset, MESSIDOR, DRIVE and DIARETDB and the average accuracy was found to be 94%.