Using adaptive edge technique for detecting microaneurysms in fluorescein angiograms of the ocular fundus

This paper describes a new digital image-processing algorithm for detecting microaneurysms in fluorescein angiograms of the ocular fundus for diabetic retinopathy patients. Firstly, edges in the image have been detected using the Canny edge detector. Edges which represent vessel segments have been removed from the image. Sobel edge detection is then used for the neighborhoods of the remaining objects. The analysis of size, shape, and energy characteristics of the resultant objects is used to finally segment the microaneurysm from other retinal features. The proposed scheme outperforms currently available techniques in both the sensitivity of microaneurysm detection and the speed of performance.