Automated corneal endothelial cell analysis

Abstract

Corneal endothelial cell images taken by a noncontact specular microscope usually have a strong gray-scale gradient and flare that make automated analysis very difficult. The development is reported of an automated analysis system of the noncontact specular image, using a specially variant contrast enhancement to remove these obstacles from the original image. The results obtained by the system are compared with the traditional method that uses a hand sketch of boundaries of the cell.<>