A Novel Integrated Approach Using Dynamic Thresholding and Edge Detection (IDTED) for Automatic Detection of Exudates in Digital Fundus Retinal Images

Abstract

The automatic screening of patients for early detection and prevention of diabetic retinopathy (DR) has been the prime focus in recent times due to the large ratio of patients to medical ophthalmologists. Exudate detection is one of the main steps of DR. A reliable method for detection of exudates is presented in this paper. Optic disc (OD) is localized by the principle component analysis (PCA). Active contour based approach is used for accurate segmentation of boundary of OD. In our IDTED method, pre-processing techniques such as histogram specification and local contrast enhancement are integrated with dynamic thresholding (DT) and edge detection for exudate detection. The IDTED algorithm, when tested on 25 digital fundus retinal images and compared with the performance of a human grader, has shown a mean sensitivity of 99% and a mean predictivity of 93%