Abstract
In this paper, we present a complex approach to improve microaneurysm detection in color fundus images. Microaneurysms are early signs of diabetic retinopathy, so it is essential to detect these lesions accurately in an automatic screening system. The recommended detection of microaneurysms is realized through several levels. First, a specific combination of different preprocessing methods for candidate extractors is found. Then, we select candidates voted by a certain number of the candidate extractor algorithms. At all these levels, optimal adjustments are determined by corresponding simulated annealing algorithms. Finally, we classify the candidates with a machine-learning based approach considering an optimal feature vector selection determined by a feature subset selection algorithm. Our framework improves the positive likelihood ratio for the microaneurysms and outperforms both the state-of-the-art individual candidate extractors and microaneurysm detectors in these measures.