Abstract
In this study a computer aided screening system for the detection of bright lesions or exudates using color fundus images is proposed. The proposed screening system is used to identify the suspicious regions for bright lesions. A texture feature extraction method is also demonstrated to describe the characteristics of region of interest. In final stage the normal and abnormal images are classified using Support vector machine classifier. Our proposed system obtained the effective detection performance compared to some of the state-of-art methods.
Highlights
Diabetic retinopathy (DR) is the leading cause of blindness and it is the diabetic eye disease
International diabetes federation (IDF) declared that 52% of Indians do not know about diabetes that they are suffering with high blood sugar
In order to remove unwanted some visible spots, noise, lines, obstacles, Bilateral filter (BF) [22,23] is used. Because it smoothens flat surfaces while preserving sharp edges in image by having same pixels placed in every neighborhood Ibf
Summary
Diabetic retinopathy (DR) is the leading cause of blindness and it is the diabetic eye disease. Ravishankar et al [7] proposed localization of lesions, based on color properties, intensity variations and morphological operations They obtained sensitivity and specificity are 95.7% and 94.2% respectively. Section-2 describes pre-processing, detection of Optic disk, vessels, identification of lesion parts, feature extraction and classification. In the resultant images, each element is multiplied with the illumination corrected image elements presented in (2) and (3) Both the structures i.e., optic disk and main vessels are masked out. After the masking of the optic disk and main vessel structures, lesion parts will be segmented by thresholding based on histogram. The contours of yellow color exudates are drawn in the illumination corrected fundus image based on intensity pixel values of thresholded image
Sign-in/Register to view highlights & summary 
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call