Automated detection of diabetic retinopathy on digital fundus images

Abstract

Fully automated computer algorithms were able to detect hard exudates and HMA. This paper presents encouraging results in automatic identification of important features of NPDR.

Diabetic retinopathy

“…Further possibilities offered by digital imaging include stereo imaging [67] and telemedicine [68]. Algorithms for automated photographic reading could possibly reach acceptable sensitivity (78-88%) and specificity (84-100%) in detecting different lesions of retinopathy, as compared to an ophthalmologist [69,70], opening the way to computer-assisted grading. As a main problem with medical image interpretation is the rapid deterioration of graders' concentration with increasing workloads, this prospect will become a viable option in mass screening for STDR.…”

Automated quality assessment of retinal fundus photos

“…In particular in ophthalmology photos of the eye background are used by medical experts to diagnose and document diseases like glaucoma or diabetic retinopathy. In addition the images are commonly further evaluated by automatic software tools to support the diagnosis [1][2][3].…”

Segmentation of retinal blood vessels using a novel clustering algorithm (RACAL) with a partial supervision strategy

“…It is an important step in screening programs for early detection of diabetic retinopathy [1], registration of retinal images for treatment evaluation [2] (to follow the evaluation of some lesions over time or to compare images obtained under different conditions), generating retinal map for diagnosis and treatment of age-related macular degeneration [3], or locating the optic disc and the fovea [4].…”

“…Unsupervised methods in the literature comprise the matched filter responses, edge detectors, grouping of edge pixels, model based locally adaptive thresholding, vessel tracking, topology adaptive snakes, and morphology-based techniques [5]. Supervised methods, which require feature vector for each pixel and manually labelled images for training, are the most recent approaches in vessel segmentation and use the neural networks [1], or the K-nearest neighbour classifier [5,6] for classifying image pixels as blood vessel or nonblood vessel pixels. These methods depend on generating a feature vector for every pixel in the image and then using training samples (with known classes) to design a classifier to classify these training samples into their corresponding classes.…”