To check and classify if microaneurysms are present in fundus eye images, which would then help in analysis of Diabetic Retinopathy.
Most of the currently available methods divide MA (microaneurysm) detection into two consequent stages: candidate extraction and classification. Usually, the first step of candidate extraction is image pre-processing to reduce noise and improve contrast. After pre-processing, specific image segmentation is used to extract as many regions as possible that probably correspond to MAs. In the second step, the resulting candidates are labelled as true or false using a supervised learning based method.