Corneal Endothelium Evaluation With 2 Noncontact Specular Microscopes and Their Semiautomated Methods of Analysis

Abstract

Noncontact specular microscopes and semiautomated methods of endothelial analysis are widely used in the clinical practice and for research purposes. In this study, the interexaminer reproducibility between two noncontact specular microscopes using different semiautomated methods was assessed in normal corneas. Average cell size (ACS), endothelial cell density (ECD), coefficient of variation (CV), and hexagonality values were independently calculated by two examiners with the Topcon SP 2000P Image-NET retraced method and the Konan CC7000 center method in 49 normal subjects. Interexaminer reproducibility and correlation between instruments/methods were assessed through 95% limits of agreement, intra-class correlation coefficient (ICC), Pearson correlation coefficient, and Spearman rank test. Interexaminer reproducibility was good for ACS and ECD measurements (ICC > 0.85) but weak for CV and hexagonality (ICC < 0.65). Significantly higher ACS and lower ECD values (P < 0.05) were obtained with the Topcon in comparison with the Konan method by both examiners. ACS and ECD mean differences were respectively, +26 microm2 and -184 cell/mm2 for examiner 1 and +36 microm2 and -228 cell/mm2 for examiner 2. No significant difference (P > 0.05) between instruments/methods was found in CV or hexagonality values; however, the correlation for these parameters was low (r < 0.20). Either instrument or method allows reproducible and accurate ACS and ECD measurements on normal corneas, but the difference between the two instruments is systematic and significant. The difference found were not clinically meaningful however for research/longitudinal study purposes, the data collected with the two systems cannot be used interchangeably.