ABSTRACT
Diabetic retinopathy is a well-known disease related to diabetes that can cause blindness in its later stages. Each stage of this disease is characterized by a different kind of retinal lesion. Microaneurysms are small red dots that appear in the first stages of diabetic retinopathy. For this reason, the detection of these structures from retinal images is a key issue for early diagnosis. Moreover, several studies have shown that microaneurysm turnover, that is, the formation rate of this kind of lesion, is a sign of worsening of the patient’s condition. In this chapter, we propose an automatic procedure for the identification of microaneurysms from retinal images as well as a methodology for lesion registration in order to compute the rate of microaneurysm turnover. We also include an automatic procedure for evaluating the significance of the microaneurysms related to their position with respect to the fovea, which is responsible for sharp central vision. Moreover, we describe the integration of these algorithms in a web-based framework for retinal analysis.
