Differential Retinal Structural Damage Exhibited by Image Enhancement of Fundus Photographs

Abstract

The use of fundus photographs and fluorescein angiographs aids significantly in differential diagnosis of retinal diseases [1]. The quality of fundus photographs is, however, often poor and therefore early detection of retinal damage is not always an easy task. To restore and enhance information contained in degraded photographs such as aerial photographs or satellite images, high speed digital computers have been used since the early 1960s [2,3]. Such image processing operations have been recently applied to analyze fundus photographs for early detection of retinal structural damage [4]. Until now, most image processing algorithms such as adaptive binarization, spatial and spectral filtering, point to point mappings as well as image registration and normalization of illumination, have been performed with mainframe or minicomputers. The use of such techniques as a clinical research tool necessitates development of a user-friendly, low-cost image processing system with menu-driven specific software for easy operation and documentation. Such a fundus image processing system (FIPS) has been recently developed with specific image processing capabilities comparable to existing large scale systems. The applicability of such capabilities in quantitative estimation of retinal structural damage arising from macular degeneration and glaucoma is presented.