Abstract
Purpose: :
To demonstrate how indices of reproducibility can improve the results of endothelial cell density (ECD), cell area (A), coefficient of variation (CV) and hexagonal percentage (%H) from corneal specular microscopes (CSM)
Methods: :
200 consecutive exams from CSO no contact CSM. The ECD, A, CV and %H will be calculated through CSO software. With these results, the software Cells Analyzer - Corneal Endothelial Statistical Lab will calculate the indices of reproducibility: calculated sample error (CSE), calculated endothelial sample size (CESS), endothelial cell counted (ECC) and sample ratio (ECC/CESS). The sample accuracy was 95% reliability degree (RD) with 0.05 relative error (RE) and 99% RD with 0.01 RE. The effect of the indices of reproducibility on the result of the exams with the cells of only one image examined (Group I) and with more cells than the sample size calculated by the Cells Analyzer (Group II) in as many images as necessary, will be demonstrated by the Confidence Interval (CI) to the ECD, A, CV and %H. The range of CI for ECD, A, CV and %H will be compared between Groups I and II with "t" Student (p<0.05 will be statistically significant).
Results: :
The reproducibility index demonstrates that 100% of specular microscopy exams can be done with 95% of RD and no one can be done with 99% of RD. The range of 95% CI was statistically significant (p<0.05) smaller when the reproducibility index demonstrated that the cell number included in the exam represented the endothelial cell population.
Conclusions: :
The reproducibility index was useful to improve the precision of the results of corneal endothelial exams in specular microscopes. The authors suggest that all brands of specular microscopes include this index in their reports.
Keywords: cornea: endothelium • imaging/image analysis: clinical • image processing