Automated video image morphometry of the corneal endothelium.

The central corneal endothelium of 13 eyes in 13 subjects was visualized with a non-contact specular microscope. This report describes the computer-assisted morphometric analysis of enhanced digitized images, using a direct input by means of a frame grabber. The output consisted of mean cell area, cell density, frequency distribution of the individual cell area, and cell polygonality. Results showed that the mean coefficient of variation of images analyzed three times consecutively was 0.95 percent. The cell analysis of three different images of the same subject was accurate, with a mean coefficient of variation of 4.2 percent.