Abstract
Current computer-aided diagnosis (CAD) systems tend to neglect the detection and grading of Diabetic Macular Edema (DME) signs. This chapter introduces a new computer based scheme for detecting and grading DME signs using color eye fundus images. The grading scheme integrates methods for: (a) detecting retinal structures (e.g. optic disk and fovea); (b) detecting lesions in the retina (e.g. exudates); (c) analyzing the spatial distribution of DME signs in the retina; and (d) grading the severity of a DME case as absent, mild, moderate or severe. In a preliminary experimental evaluation of our DME grading scheme using publicly available eye fundus images (i.e., DIARETDB1 image database), an accuracy of 94.29 % was obtained with respect to the mode of the evaluations of the same DME cases by four experts. This is encouraging, since a similar DME grading performance is achieved by a DME expert. In order to calculate the clinicians grading performance, we assumed the mode of all experts DME gradings as the reference evaluation for each case. Thus, if an expert assigned each DME case to the class identified as the mode of the experts severity gradings, that expert achieved an accuracy of 100 %.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call