Cataracts are the leading cause of blindness worldwide. Research into the natural history of different types of cataract and factors that may prevent or promote cataractogenesis is a high priority. Such research demands a valid, reproducible system for determining the type and severity of lens opacities that is consistent over time. Furthermore, because cataracts develop slowly over a period of time, a system with sufficient sensitivity to assess small changes is highly desirable. The purpose of this proposal is to develop and validate an automated, objective classification system, based on currently available digitized standard lens photographs. In addition we propose to implement digital image acquisition so that the entire acquisition, archiving, classification, and grading procedure can be fully automated. For nuclear lens images, detailed densitometric analyses will be carried out based on relative optical density to control for photographic non-linearity. The use of neural net type classifiers is described. For cortical and posterior subcapsular (PSC) opacities, we are developing radial projection algorithms based both on densitometric and geometric criteria to yield elements of the area of opacities and regional location. The automated systems will be developed using a set of 50 test photographs of varying degrees of nuclear cataract severity, 50 photographs of cortical cataract, and 50 PSC. The validation of the systems will be done against a set of 100 photographs collected in a clinic series, and a longitudinal set of 540 photographs collected over 6 years in a population based study. The direct digital system will be implemented with a charge-coupled device camera and validated with an additional 100 photographs. The resulting classification system will have wide applicability in prospective clinical trials of anti-cataract agents and in epidemiologic studies of risk factors and preventative measures in cataractogenesis.
| Friedman, D S; Duncan, D D; Munoz, B et al. (1999) Digital image capture and automated analysis of posterior capsular opacification. Invest Ophthalmol Vis Sci 40:1715-26 |
| Duncan, D D; Shukla, O B; West, S K et al. (1997) New objective classification system for nuclear opacification. J Opt Soc Am A Opt Image Sci Vis 14:1197-204 |
