Purchase this article with an account.
Curry Bucht, Per Söderberg; A Large Population Study Assessing Fully Automated Frequency Domain Based Estimation Of Cell Densities At Low Cell Count Images Of The Corneal Endothelium. Invest. Ophthalmol. Vis. Sci. 2011;52(14):6444.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Pathological conditions as well as physical trauma to the cornea may impact negatively on the corneal endothelial cell density (ECD), threatening the optical properties of the cornea. Analysis of the corneal endothelium is a common procedure in several clinical applications. A common method for in vivo corneal endothelial morphometry is by semi-automated analysis of endothelial images, captured by Clinical Specular Microscopy (CSM). Due to the need of operator involvement, current means of ECD estimation can be time consuming, having a negative impact on sampling size. The purpose of the study presented here was to compare a newly developed fully automated method for estimating ECD of images captured by clinical specular microscopy in vivo, to a semi-automated method provided by the IMAGENET-640, using a large population of low cell count images.
The analysis software was developed in MATLAB. Pictures of corneal endothelia captured by CSM were read into the software. The pictures were automatically enhanced. Next, the image data was transformed from the spatial domain to the frequency domain using a numerical Fourier transform. Tools for extraction of relevant image characteristics and analysis of frequency domain data were applied to the transformed images. Data obtained from each transformed image was used to estimate the ECD of the endothelium images. A population of 250 unique in vivo images was used in the study, spanning a large range of ECD. The ECD estimated using the new, fully automated analysis method and the ECD estimated by the semi-automated IMAGENET-640 barrier tracing method were compared. Ill conditioned images were excluded from the study.
The average ECD of the 250 images was 2678 cells/mm2, spanning [1199; 4750] cells/mm2. A 95% CI over the residual standard deviation of the fully automated ECD estimate in relationship to the semi-automated estimate was found at [183; 218] cells/mm2, corresponding to a coefficient of variation of [0.068; 0.081]. Analysis time was in the order of 15-20 seconds.
Using frequency analysis for cell count estimation is not new. However, so far the method has predominately been applied to high cell count structures such as models or donor buttons due to resolution issues with low cell count data. The relatively high correlation in estimated ECD between the new and significantly faster method and the semi-automated method is encouraging.
This PDF is available to Subscribers Only