Overestimation of Corneal Endothelial Cell Density in Smaller Frame Sizes in In Vivo Confocal Microscopy

Purpose: To evaluate the effect of frame size on the calculated corneal endothelial cell density (CECD) in images of laser scanning in vivo confocal microscopy (IVCM). Methods: Forty-nine corneal endothelial images acquired by laser scanning IVCM (Heidelberg Retina Tomograph 3 with Rostock Corneal Module) with different endothelial cell densities were analyzed. In each image (160,000 mu m(2)), the CECD was calculated using the fixed-frame method by counting cells in the following frame sizes: 80,000 mu m(2), 40,000 mu m(2), 20,000 mu m(2), 10,000 mu m(2), 5000 mu m(2), and 2500 mu m(2). The calculated CECD was then compared with that of the variable-frame method as the reference value. Results: There was no significant difference in the calculated CECD between the variable-frame method (2004 +/- 832 cells/mm(2)), and the fixed-frame method using a 40,000-mu m(2) frame (2023 +/- 810 cells/mm(2)). On the other hand, the calculated CECD showed significant overestimations in frame sizes of 20,000 mu m(2) (2066 +/- 820 cells/mm(2)), 10,000 mu m(2) (2156 +/- 785 cells/mm(2)), 5000 mu m(2) (2352 +/- 783 cells/mm(2)), and 2500 mu m(2) (2715 +/- 754 cells/mm(2)), with P < 0.001 in all. This resulted in overestimations of 4.8 +/- 9.8%, 11.9 +/- 16.2%, 24.9 +/- 23.1%, and 49.1 +/- 38.8% for these frame sizes, respectively. Images with lower CECD demonstrated higher overestimations of cell density in smaller frame sizes. Conclusions: In laser scanning IVCM images, there is significant overestimation of CECD if the cells are counted in frames smaller than 25% of the image. Similar frame sizes should be used when monitoring CECD over time.