Novel algorithm for exudate extraction from fundus images of the eye for a content based image retrieval system

摘要

With technological advancements in medical electronics, and computerization of all standard medical institutions, the amount of image data being produced in these fields has been increasing constantly. Each hospital or institute provides services to thousands of patients per day, and most of the diagnosis tools are images for example x-rays, ultrasound scanners, magnetic resonance imaging etc. Since most hospitals keep records of each patient's case history, it is possible to detect any diseases in a person in its early stages itself, by studying the cases of the previous patients who exhibited similar symptoms, and direct the doctors for further tests. A possible environment for testing the aforementioned procedure is to use images of the fundus of the eye, which are used to detect and monitor the presence or progress of diseases pertaining to the eye. Extraction of features from these images that generally indicate the presence of afflictions, namely exudates, is addressed in this paper. A novel algorithm for the detection of the Optical Disk and the presence of any exudates has been described in this paper. The algorithm has been tried in MATLAB version 2011b. An accuracy of 92% was achieved.