Automated Glaucoma Screening in Retinal Fundus Images

Abstract Glaucoma is an eye disease which damages the optic nerve that carries information from the eye to the brain. Glaucoma is the second foremost reason of blindness. An efficient hardware based implementation of glaucoma screening is a significant task in the automated retinal image analysis method. This paper presents an automatic glaucoma screening using a TMS320C6416DSK DSP board. The detection procedure consists of two stages. The first stage comprises of image pre-processing and detection of optic nerve head center using circular Hough Transform .In the second stage, the optic disk diameter is calculated and cup is segmented from disk. The proposition between disk and cup is calculated for abnormal image screening. The implemented technique is tested on a publicly available retinal image data sets and the average accuracy achieved is 97.5%. Keywords: fundus image, glaucoma, optic disk, circular hough transform, TMS320c6416 1. Introduction Glaucoma is an irreversible eye syndrome. According to reports in 2010, it is second primary reason of blindness in the world. 2.3% of the peoples got affected from total population. It is predicted that this number will increase to 2.86% in 2020.Studies have been shown that increase in intraocular pressure (IOP) of the eye is one the cause for glaucoma [2]. To maintain healthy vision, eye produces a small amount fluid called aqueous humor the same amount fluid will be thrown out of eye. This balance keeps the IOP in limit. If the balance is not maintained the IOP increase and damage the optic nerve head which make irreversible vision loss. So, the early precise detection and treatment of glaucoma will control the progression of the disease. The ophthalmologist uses the ratio between optic cup and optic disk (CDR) as a parameter to screen the glaucoma patients. The CDR is <0.65 in case of normal eye and for the glaucoma eye the value is lies in between 0.65 to 0.9. Manual glaucoma screening is time consuming and prone to human error. So, An intricate algorithm is needed for mass screening of glaucoma.Figure 1 shows the normal image and glaucoma image.