Evaluation of preprocessing methods for microaneurysm detection

In this paper, we provide a comparative evaluation of preprocessing methods for microaneurysm detection in color fundus images. Our aim is to achieve high sensitivity values in the candidate extraction phase. This requirement can be fulfilled if we select an appropriate preprocessing method for the candidate extractor. Our experiments showed that Contrast Limited Adaptive Histogram Equalization achieved the best performance among the investigated preprocessing methods. Moreover, we show that certain relationship also exists between the specificity and the information content of the images. This observation can help us to select preprocessing methods for which we do not have prior information regarding microaneurysm detection.