Mathematical morphology based fovea center detection using retinal fundus images

Abstract

Exudative diabetic maculopathy is a frequent cause of visual deterioration in patients with diabetic retinopathy and represents a form of diabetic macular edema (DME), which is derived from leaking retinal vessels. The detection of the fovea center is a prerequisite for diagnosis of exudative diabetic maculopathy. In this work, a novel method for fovea center detection from color retinal fundus images is presented. With the prior knowledge of relative location of the optic disc, mathematical morphology is used to detect fovea center. The proposed method is robust to inconveniences caused by diabetic retinopathy lesions like microaneurysms, hemorrhages and exudates. Experiments were performed on local and public databases that yielded success rate of 91.38 % and 91.75 %, respectively. © Springer International Publishing Switzerland 2014.