Abstract
Purpose :
Specular microscope (SM) is a device that was designed to recognize the borders of corneal endothelial cells (EC). Using this image, a computer program in the device calculates and estimates the corneal morphological status of the endothelium. However, SM can miscount or not recognize some images/borders. Previous studies have described three types of miscounting errors and demonstrated that these errors do not have statistical difference in three consecutive examinations of the same healthy cornea. The purpose of this study is to analyze the sample error, estimate the discarded cell area in automatic mode, and miscounting error patterns of 3 sequential examinations in patients submitted to cataract surgery.
Methods :
Three consecutive examinations using non-contact SM were performed in 30 eyes of 30 patients submitted to uneventful cataract surgery. The following data generated by SM was analyzed: counted cells (CC), endothelial cell density (ECD), average cell area (AVG), coefficient of variation (CV), percentage of hexagonal cells (HEX) and pachymetry. Manually, miscounting error patterns were quantified as: non-counted EC (type 1); number of EC groups (type 2) and number of split EC (type 3). The percentage of discarded area of the reticulum was calculated as {137500 (µm2) – [counted cells x average cell area (µm2)]}/137500 (µm2). To analyze repeatability, data generated from the 3 consecutive examinations were expressed as average and standard deviation and compared using ANOVA (p<0.05).
Results :
No statistical difference was observed between the 3 consecutive examinations in all measurements performed.
Conclusions :
All SM data of 3 consecutive examinations generated automatically and manually showed considerable repeatability in patients submitted to cataract surgery. Even the miscounting cell errors were repeated in all examinations.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.