An Automated Algorithm For Photoreceptors Counting In Adaptive Optics And Its Application

Eyes are important organs of humans that detect light and form spatial and colorvision. Knowing the exact number of cones in retinal image has great importance inhelping us understand the mechanism of eyes’ function and the pathology of some eyedisease. In order to analyze data in real time and process large-scale data, an automatedalgorithm is designed to label cone photoreceptors in adaptive optics retinal images.Images acquired by the flood-illuminated AO system are taken to test the efficiency ofthis algorithm. We labeled these images both automatically and manually, and comparedthe results of the two methods. A minimum94.9%match rate and93.0%agreement ratebetween the two methods are achieved in this experiment, which demonstrated thereliability and efficiency of the algorithm. Besides, we use Bennett’s method to computethe density of cones in the fovea area of retina and developed the cone countingsoftware using MFC.