Optic Disc Detection using Fine Tuned Convolutional Neural Networks

The detection of the Optic Disc (OD) is an significant step in retinal fundus images analysis; it allows to extract relevant information that proved to be useful for the prevention of several pathologies, such as glaucoma, hypertension, diabetes and other cardiovascular diseases, which manifest their effects in the retina. In this work we present a supervised method for automatically detecting the position of the Optic Disc in retinal fundus digital images; the goal has been achieved by means of a proper reuse of previous knowledge from a pre-trained Convolutional Neural Network (CNN), already able to detect faces in an image. Experimental analyses showed high level of accuracy in the detection of the optic disc on the DRIVE, STARE and DRIONS databases.