Determination of the average size, in variation in size, and regularity of shape of the cells of the innermost layer of the cornea is extremely important in the clinical evaluation of corneal tissue. Aging, injury, and disease increase both the average cell size and the deviation of cell shape from perfect hexagonality. If cell densities below critical thresholds are measured, surgical procedures like cataract removal are considered poor risks. Because current techniques for estimating these parameters are cumbersome and computationally involved, statistical estimates of cell density, coefficient of variation of cell size, and hexagonality (percent hexagons) are often based on the measurement of a mere 50-100 endothelial cells. This number is considered too small in most clinically interesting cases, where there is typically wide variation in cell size and shape. The purpose of the proposed research is to conduct preliminary investigations of a promising high-speed optical/digital electronic method for measuring statistically meaningful and clinically significant size and shape-related parameters of corneal endothelial cells. Parameters are obtained by information-reducing measurements of the two-dimensional Fourier transform of cell boundary patterns. Of particular importance are radial and angular projections of polar-coordinate representations of the transform distributions. These projections allow statistical characterizations of the cell boarder patterns that are free, respectively, of the effects of cell shape and size variations. The key to high-speed implementation is the virtually instantaneous calculation of the two-dimensional Fourier transform of the cell patterns by a diffraction-based optical system. Preliminary studies, based on both optically- and digitally-computed fourier transforms of human endothelial cell patterns, indicate that average cell size--the clinically most significant parameter--can be estimated both quickly and accurately over patterns of cell ranging room fifty to several thousand in number. Shape-related measurements, which for individual cells allow differentiation between, e.g., hexagons and pentagons, can also be easily made; however, the relationship of measurable quantities to currently- measured morphological characteristics (e.g., degree of hexagonality) is not clear at this time. The proposed program of research includes (1) assembly of an optical system incorporating a segmented wedge-ring photodetector in the Fourier transform plane for real-time reduction of data to radial and angular projections; (2) evaluation of different procedures (e.g., coarse-grained scan of large areas followed by fine-grained scans of smaller regions) for scanning endothelial cell patterns to extract statistical parameters; (3) refinement of methods (e.g., curve-fitting, half-width estimation, etc.) for extracting statistically reliable measures of cell characteristics from raw data curves; (4) comparison of statistical parameters obtained by this method with those obtained from computer-aided cell morphometric techniques; (5) comparison of regional and global morphological statistics.
