Development of a Detection System Microaneurysms in Color Fundus Images

The detection of microaneurysms (MAs) in color fundus images remains an open issue in the medical image processing due to the low availability of reliable programs. In this paper, we present a system for automated detecting of MAs based with the techniques Hessian matrix and contrast limited adaptive histogram equalization (CLAHE). Thus, features extraction as optical disc, vascular tree and other pathologies for reducing false positives ( FP). To evaluate the detection system MAs, we examined 81 color fundus images, classified by conditions and compared with fluorescence images, these are angiograms. The system provides a sensitivity of 15-24% in the phase of mild no proliferative diabetic retinopathy (DR) while in the phase of moderate no proliferative DR gave 10% sensitivity, in both phases the system detects more than 30 as candidates MAs. In healthy patients gives an average of 6 FP per image and patients with other diseases without DR less than 20 FP per image. Demonstrating that the system can work within different conditions, having a difference in results both in patients without DR as with DR, it which could be used in the early detection of disease.