Purchase this article with an account.
Obidulho Sakassegawa Naves, Fábio Ursulino Reis Carvalho, MARCO Otavio Santos de ALMEIDA, Andre Ferreira da Silva, Ricardo Holzchuh, Richard Y Hida, Fernando Cesar Abib; Possible sampling error in donor corneal specular microscopy images: a pilot study.. Invest. Ophthalmol. Vis. Sci. 2017;58(8):1454.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The Cells Analyzer software® (Corneal Endothelial Lab, Technicall, Brazil) can calculate the sample error in specular microscopy (SM) and suggest the need to increase counted endothelial cells (EC) to acquire reliability. This study aims to analyze the sample error in donor corneal SM images and its variations between images of the same cornea.
Seven donor corneas were submitted to SM (KeratoAnalyzer EKA-10, KONAN®). Four images were obtained for each cornea on the day of arrival in the Eye Bank (Group 1) and four images on the day of the distribution (Group 2). The corneal tissue was preserved in Eusol-C® at a temperature of 39,2°F (4°C) and it was incubated at 80,6°F (27°C) for 30 minutes before examination. EC were counted with the semi-automatic method for each image and the manufacturer’s software (KSS-EB10) performed a multi-sample average (Image 1). Then, the Cells Analyzer® software was used to calculate the sample error for each image and to indicate the number of EC that should be counted to obtain a representative sample (NCR). The variables studied were: Calculated sample error (CSE), counted cells (CC), NCR and the difference between the NCR among the images of the same cornea (Δn). The planned error (PE) was 5% in all samples. Descriptive statistics was used for statistical analysis.
All images obtained showed CSE higher than the PE, in both groups. 57,1% of the images in group 1 and 100% in group 2 had a Δn higher than 150 cells. The obtained data is shown in tables 1 and 2.
All images showed CSE higher than 5%. It seems that SM tries to compensate this fact by performing a multi-sample analysis. However, the great variability of NCR among the images of the same cornea makes it inapropriate to calculate an average between these images.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.
This PDF is available to Subscribers Only