*Corresponding author: Shantala Girradi
B.V.B. College of Engineering and Technology, Hubbali, Karnataka, India
ISSN: 0976-3031
RESEARCH ARTICLE
QUALITY ASSESSMENT AND GRADING OF DIABETIC RETINOPATHY
Shantala Girradi, Anoop Betigeri and Chaitra Besthar
B.V.B. College of Engineering and Technology, Hubbali, Karnataka, India
ARTICLE INFO ABSTRACT
Article History:
Received 5
th
, March, 2015
Received in revised form 12
th
,
March, 2015
Accepted 6
th
, April, 2015
Published online 28
th
,
April, 2015
Diabetes, caused due to lack of insulin in human blood. This affects some parts of human body one such
important organ is eye leading to Diabetic Retinopathy. Diabetic Retinopathy, a complication in diabetic
patients, causes damage to retina which in later stages may lead to blindness. The disorder can be detected
by regular scanning of the retina of diabetic patients. However, Indian Ophthalmologists to Indian
population ratio is 1:90000 which is not proportional necessitating to automation of the system. The
proposed system, using Image processing techniques, requires retinal fundus image of an eye taken from
fundus camera. But sometimes it is difficult for the system to process due to the dark images leading to
misinterpretation in the results. Hence system performs the quality assessment of image by dividing Image
into blocks, extracting first order histogram features from retinal image, detecting exudates using AND
logic and finally grading the Diabetic Retinopathy as mild, moderate or severe. This paper mainly focuses
its content on Image Quality assessment and Retinopathy Grading System. System has achieved 90.9%
and 90% of sensitivity and specificity in quality assessment of retinal image.
Key words:
Diabetic Retinopthy(DR),
exudates, macula, retina, quality
analysis, AND logic, grading of
DR.
Copyright © Shantala Girradi, Anoop Betigeri, Chaitra Besthar., This is an open-access article distributed under the terms of the Creative
Commons Attribution License, which permits unrestricted use, distribution and reproduction in any medium, provided the original work
is properly cited.
INTRODUCTION
The most challenging disorder in today’s generation is
Diabetes. Though Diabetes is preventive, it is seen as almost
inevitable for every individual with their lifestyles. Among
80% of patients who have Diabetes for 10 years or more are
likely to be suffer from Diabetic retinopathy, a retinal disorder,
leading to vision loss. Since the vision of any individual is
precious it needs to be cared the most. During early stage of
diabetic retinopathy, NPDR, blood vessels are bulged resulting
in soft exudates. In later stages, PDR, abnormal blood vessels
are formed. These vessels burst and bleed since those are weak.
This forms another type of exudates known as hard exudates.
In any of these stages the vision of eye will be reduced and
even may result to complete vision loss because of these
exudates in macula region of retina. Hence conclusion can be
drawn that presence of exudates are key in detecting Diabetic
Retinopathy and hence grading it.
Method followed to detect Diabetic Retinopathy is regular
screening of retina using fundus camera. A research shows that
regular screening for diabetic retinopathy could reduces the
numbers of people at least by 90% in new cases who develop
vision-threatening retinopathy. The estimate of the actual
number of diabetics in India is around 40 million and is
expected to reach 70 million by 2025. During their lifetime,
nearly half of the nation’s estimated 16 million people with
diabetes will develop some degree of diabetic retinopathy, and
as many as 25,000 people go blind annually. However, it is
estimated that the ratio of ophthalmologist to population is
1:90000 in India. This ratio seems not to be proportional as DR
needs aggressive screening. Thus it necessitates the need of
automated system.
This paper presents an automated system which aims to
achieve two purposes. First, analyze the quality of the images.
Second, grade the diabetic retinopathy by detecting the
maculae and exudates region and then grade the DR as mild,
moderate and severe.
LITERATURE SURVEY
With respect to the Quality Analysis, Radhe et al [1], the input
image accepted is converted to gray and then using the bit plan
separation, contrast enhancement and morphological
processing are applied to extract the blood vessels. By using
the digital wavelet transforms and the energy feature
coefficients are used to get the features and then trained to
PNN (probabilistic neural network) and Classification is done
with the help of segmentation like K-means Clustering method
thereby extracting the exudates determining whether the retina
is normal or abnormal and then Morphological operations are
applied on segmented image for smoothening the exudates part.
But the image without the exudates cannot be used to analyze
the Image quality. In the paper described by the Ramon Pires
[2] et al, In this methodology, a reference image with assured
quality is assumed to be known and quantitative measures of
quality for any image are extracted by comparisons with the
Available Online at http://www.recentscientific.com
International Journal
of Recent Scientific
Research
International Journal of Recent Scientific Research
Vol. 6, Issue, 4, pp.3259-3264, April, 2015