Could the coefficient of variation (COV) of the corneal endothelium be overestimated when a centre‐dot method is used?

Abstract

Background:  Little has been published on the reliability of estimates of the coefficient of variation (COV) in cell area for human corneal endothelia. The present study compares two methods.

Methods:  A non‐contact specular micrograph (Topcon SP‐2000P) was obtained from the central region of the corneal endothelium of 20 healthy myopic white European subjects, aged from 32 to 53 years, half of whom were successful long‐term soft contact lens wearers. The captured image file was either assessed using a machine‐based algorithm, in which 25 cells in the middle of the image were marked and their areas reported (designated as ‘centre‐dot’ method) or by a manual method, by which all the cells in the image were outlined on very high magnification prints of the endothelia and the cell areas measured by a manual digitiser in stream mode. The average cell area was used to calculate the endothelial cell density (ECD), while the COV was calculated from the standard deviation (SD) of the cell area measures.

Results:  Identical mean cell area values were found (392 µm²) with the two methods, a marginally higher ECD estimate (2,594 versus 2,569) with the centre‐dot method (p = NS) but a much higher COV with the centre‐dot method (43.8 versus 29.0 per cent). This highly statistically significant difference in COV (p < 0.001) was seen in both contact lens wearers and non‐contact lens wearers. A Bland‐Altman analysis reveals a bias in the centre‐dot method, especially for the COV estimates, that appears to be linked to erroneous definition of a single large cell domain on any individual image.

Conclusions:  A centre‐dot method can be reliably used to generate useful data on cell area and ECD but it should be used cautiously for estimates of polymegethism (COV).

Key words:

• cell area
• coefficient of variation
• cornea
• endothelium
• morphometry
• specular microscopy

ACKNOWLEDGEMENTS

The author would like to thank Mr Sven Jonuscheit and Dr Emil Oblak for assistance with the specular microscopy image acquisition.