Detection of Microaneurysms in Color Fundus Images

Diabetic Retinopathy is an eye disease which occurs due to damage in the retina as a result of long term illness of diabetic mellitus. Microaneurysm detection is the first step in automated screening of diabetic retinopathy. To perform this detection, several methods have been proposed in the diagnosis of diabetic retinopathy. In this paper, initially the pre-processing techniques like gamma correction, green channel extraction, location based contrast enhancement process such as Adaptive Histogram Equalization (AHE) are applied to improve the visibility of microaneurysms in the color fundus images. Microaneurysms are segmented using fuzzy c-means clustering algorithm on the pre-processed image. Then features are extracted using Gray Level Cooccurrence Matrix (GLCM), wavelet and first order statistics. All these features constitute the feature set which is used in knn classifier to remove spurious candidates. The remaining candidates can be thresholded further for a binary output which gives the final candidates.