The diabetic retina is a serious and common complication of long-term diabetes in individuals. This complication can damage the retina and ultimately lead to blindness. Since, the emergence of red lesions (microaneurysms) on the retina is one of the first symptoms of a diabetic retina, extensive research has been done on the effective detection of these abnormalities. In this project, we have implemented a method [1] based on the multi-scale correlation filtering and dynamic thresholding. This method involves two steps: (1) detection of candidate red lesions, and (2) precise detection of red lesions. The effectiveness of the method is evaluated on ROC dataset.
[1] Zhang, Bob, et al. "Detection of microaneurysms using multi-scale correlation coefficients." Pattern Recognition 43.6 (2010): 2237-2248.