Diabetic Retinopathy (DR)
is an diabetic disease which affects the vision of human
eye due to high sugar level or insulin in blood
and it can also leads to
blindness .Since it does not have any particular notable symptoms in early
stage it causes vision loss. In order to treat it at early stage automatic
lesions detections or screening of eye were done which include Microaneurysms (MAs) and
hemorrhages(HEMs) and bright lesions, such as exudates (EXs) Akram,
Khalid 2.The screening of DR is
done by automatic computer-aided
diagnosis (CAD) system which accurately identifies the lesions.
Microaneurysm is the first
stage of DR in which the tiny blood
vessels starts swelling and they appear as small, red dots in the superficial retinal layers
and they are oval are spherical in shape ,some of them are very thin,
microaneurysms are called bright lesions Zhang, Wu 2.
Hemorrages were bright
lesions or red
lesions caused by the rupture of the small blood vessels in the deeper layers of
the retina and also by retinal ischemia.
Exudates are final stage of Diabetic
Retinopathy (DR),they are yellowish-white lesions caused due to plasma leakage
from the capillaries. Since exudates
affects macular region they are also known as exudative maculopathy.
The CAD screening results will
sometimes become more complicated when
the blood vessels and lesions are
similar in shapes, non-uniform illumination,color
Detection of DR’s microaneurysms is based on the principle
of analyzing rotating cross-sectional profile of the pre-processed image Lazar and
Hajdu 3, Splat Feature
detection is used for Hemorrages detection and classification Tang, Niemeijer4 ,Exudate Detection is done by
Exudate Segmentation in fundus image using optimization approach is the basics
of exudates detection and classification 1.Microaneurysms in some cases were
too big and close to blood vessels and fails for clear detection and it can be
overcome by Cheriet, and. Langlois 6
.Since Hemorrages are rupture of
blood vessels which were clotted during the first stage .In this proposed
method the mutual work of matched filter (MF) and Laplacian of Gaussian (LoG) were used for the
operation of lesion detection .Matched Filter method not only gives approximate
response to microaneurysms
but also for Hemorrages. Matched Filter
along with LoG filter is used to separate exudates and hemorrages .
main contributions of this method are:
Optimal wideband bandpass filter.
make the system adaptive , Differential Evolution is used to set gain and