Abstract:
Diabetes is one of the major problem that modern world is facing today. The factors
that contribute to increasing rate of diabetes are obesity, physical inactivity, and aging
population. Diabetes is classified into two major categories: Type I usually diagnosed
in young adults and children in which body does not produce insulin, Type II is the
most common form of diabetes in which the body does not produce enough insulin
or the cells ignore insulin. Diabetic retinopathy (DR) is one of the disabling micro
vascular complications of diabetes mellitus that causes the loss of central vision or in
most of the cases complete vision loss if it is not recognized and cured at the earlier
stage. The epidemiological survey classifies DR amongst the four major causes of vision
impairment. According to study, the chances for patient of type I diabetes of having DR
after 20 years of diabetes increasing by 99 percent, and 60 percent with type II diabetes
have same degree of diabetic retinopathy. This disease needs a regular screening in
order to prevent vision loss as it is asymptomatic and it is unrecoverable if detected
at the later stage. The workload would increase a lot for ophthalmologists if all the
patients undergo to regular screening within quality assured framework. The manual
examine of DR have chances of human error, low accuracy and it is not a time efficient
solution. Therefore automated detection of diabetic retinopathy is required for regular
and timely screening. The problem of automatic detection of DR signs in a retinal
image (fundus image) lies in the domain of image processing and pattern recognition.
The work represents an automated technique for detection of diabetic retinopathy by
localization and identification of exudates. The work utilized Gabor wavelet filter bank
for detection of exudates. The proposed method has the ability to extract blood vessels,
localize exudate region and optic disk in retinal image.
