DETECTION OF DIABETIC RETINOPATHY USING RADIAL BASIS FUNCTION

[attachment=15967]
I. INTRODUCTION
Diabetic Retinopathy (DR) cause blindness
[1]. The prevalence of retinopathy varies with the age
of onset of diabetes and the duration of the disease .
Color fundus images are used by ophthalmologists to
study eye diseases like diabetic retinopathy [2]. Big
blood clots called hemorrhages are found. Hard
exudates are yellow lipid deposits which appear as
bright yellow lesions. The bright circular region from
where the blood vessels emanate is called the optic
disk. The fovea defines the center of the retina, and
is the region of highest visual acuity. The spatial
distribution of exudates and microaneurysms and
hemorrhages[3], especially in relation to the fovea
can be used to determine the severity of diabetic
retinopathy


II. MATERIALS AND METHODS
This research work proposes contextual
clustering (CC) and Radial Basis Function (RBF)
network. CC is used for feature extraction. The
extracted features are input to the RBF network. In
order to achieve maximum percentage of
identification of the exudates, proper data input for
RBF, optimum topology of RBF and correct training
of RBF with suitable parameters is a must.
A large amount of exudates and non
exudates images are collected. Features are
extracted from the images using contextual clustering
segmentation. The features are input to the RBF and
labeling is given in the output layer of RBF. The
labeling indicates the exudates. The final weights
obtained after training the RBF is used to identify the
exudates. Figure 1 explains the overall sequence of
proposed methodology.

III.CONTEXTUAL CLUSTERING
Image segmentation is a subjective and
context-dependent cognitive process. It implicitly
includes not only the detection and localization but
also the delineation of the activated region. In
medical imaging field, the precise and computerized
delineation of anatomic structures from image data
sequences is still an open problem. Countless
methods have been developed, but as a rule, user
interaction cannot be negated or the method is said
to be robust only for unique kinds of images.