Morphometry of cells and guttae in subjects with normal or guttate endothelium with a contour detection algorithm

To develop a semiautomatic method to analyze morphology of cells and guttae in corneal endothelium.

Specular endothelial pictures from 42 and 21 subjects with healthy and guttate corneas, respectively, were analyzed independently by two observers with cell contour-extracting routines. One observer also analyzed healthy endothelia with the Corner method (Bambi). Differences between observers and between methods in mean cell area (MCA), coefficient of variation (CV), and percentage of cells with five, six, or seven sides were tested for significance with paired t tests. The Contour analysis of pictures with guttae included their mean area.

There were no significant differences in MCA, CV, or the percentage of cells with five, six, or seven sides between the measurements obtained on repeated analysis by the same observer or on a second analysis performed by a different observer with the Contour method. However, the differences between the Contour and Bambi methods were statistically significant for MCA (337.5 +/- 37.7 vs. 327.7 +/- 36.5), CV (0.32 +/- 0.05 vs. 0.30 +/- 0.05), and percentage of cells with six and seven sides, but not for the percentage of five-sided cells. In subjects with guttata, the MCA was 561 +/- 170 microm, and the mean area of guttae was 1,538 +/- 849 microm.

This detection algorithm is repeatable and reproducible, and it generates a cell border overlay useful in analyzing the morphology of cells and guttae. The analysis of corneal guttae could become a useful follow-up procedure to discriminate between patients with corneal guttata and Fuchs dystrophy.