The computer analysis of stereo photographs of the optic disk has been shown to be useful in quantitating changes in optic disk topography. Digital photogrammetry, in which a computer processes the stereo images and extracts elevation data automatically, is currently being used in a number of institutions, and shows promise as a practical, low-cost technique for quantitative analysis of the optic disk. A common problem in digital photogrammetry is that of accurately determining the stereobase associated with a given pair of images. Control of stereobase is particularly difficult when a standard monoscopic fundus camera is used to collect stereo images by shifting the camera slightly between pictures. Camera motion, patient head motion, and eye rotation all affect the stereobase. Eye rotation is particularly difficult to control, and the errors in stereobase which it causes limit the repeatability of elevation measurements made with current instruments. This Phase I effort is devoted to development of photographic and image processing techniques which can measure the eye rotation associated with a pair of images and make corrections for it in the elevation calculations. The research plan includes both theoretical photogrammetric studies to model the photographic process, and experimental studies with a model eye to test the mathematical model. Finally, the techniques will be applied to sequential stereo fundus photographs for which simultaneous stereo photographs are also available as a standard. These new techniques will permit the collection of accurate, repeatable elevation data from the optic disk using sequential stereo images from standard cameras. Phase II would consist of development of a low-cost computer-based instrument which could be easily integrated into any fundus photography service to provide the capability to measure optic disk topology.
