PURPOSE: Several automated systems have become available to determine endothelial cell density (ECD) of donor corneas. The purpose of this study was the comparison of 2 systems, the Endothelial Analysis System EAT V1.4 (Rhinetec, Duesseldorf, Germany) and the NAVIS Cell Count Version 3.4.1 (Nidek Technologies, Erlangen, Germany), against manual ECD counts. METHODS: During organ culture, the endothelium of 50 human corneas was photographed on both conventional film and digitally. Manual fixed-frame counting was performed on 2 photographs by 3 experienced ophthalmologists. Three digital pictures of each cornea were analyzed semiautomatically on the Rhinetec EAT system V1.4. Images were imported and analyzed by the fully automatic mode of the NAVIS system V3.4.1 (which also offers a semiautomatic counting mode). In an additional experiment on 11 corneas, the NAVIS counts on the microscopy system previously used only for manual cell counts were directly compared with manual counts. RESULTS: Mean endothelial cell density obtained by manual counting of all observers was 2679 +/- 313 (SD) cells/mm. Values obtained by digital analysis with EAT were a mean of 1.3% higher (2713 +/- 286 cells/mm), which is not a significant difference. In contrast, analysis of the digital pictures with the fully automated NAVIS software mode showed a mean of 7.0% higher measurements (2867 +/- 111 cells/ mm), a significant difference (P = 0.0001). Altman-Bland analysis revealed that the fully automated NAVIS counts gave too low results for high ECDs and too high results for low ECDs. This result was confirmed by the direct comparison with manual counting. CONCLUSIONS: Fully automated analysis of ECD remains problematic. Validation of any computer-assisted system against manual counting is of importance.
PURPOSE: Several automated systems have become available to determine endothelial cell density (ECD) of donor corneas. The purpose of this study was the comparison of 2 systems, the Endothelial Analysis System EAT V1.4 (Rhinetec, Duesseldorf, Germany) and the NAVIS Cell Count Version 3.4.1 (Nidek Technologies, Erlangen, Germany), against manual ECD counts. METHODS: During organ culture, the endothelium of 50 human corneas was photographed on both conventional film and digitally. Manual fixed-frame counting was performed on 2 photographs by 3 experienced ophthalmologists. Three digital pictures of each cornea were analyzed semiautomatically on the Rhinetec EAT system V1.4. Images were imported and analyzed by the fully automatic mode of the NAVIS system V3.4.1 (which also offers a semiautomatic counting mode). In an additional experiment on 11 corneas, the NAVIS counts on the microscopy system previously used only for manual cell counts were directly compared with manual counts. RESULTS: Mean endothelial cell density obtained by manual counting of all observers was 2679 +/- 313 (SD) cells/mm. Values obtained by digital analysis with EAT were a mean of 1.3% higher (2713 +/- 286 cells/mm), which is not a significant difference. In contrast, analysis of the digital pictures with the fully automated NAVIS software mode showed a mean of 7.0% higher measurements (2867 +/- 111 cells/ mm), a significant difference (P = 0.0001). Altman-Bland analysis revealed that the fully automated NAVIS counts gave too low results for high ECDs and too high results for low ECDs. This result was confirmed by the direct comparison with manual counting. CONCLUSIONS: Fully automated analysis of ECD remains problematic. Validation of any computer-assisted system against manual counting is of importance.
Authors: Juan P Vigueras-Guillén; Busra Sari; Stanley F Goes; Hans G Lemij; Jeroen van Rooij; Koenraad A Vermeer; Lucas J van Vliet Journal: BMC Biomed Eng Date: 2019-01-30
Authors: Mukhtar Bizrah; Steve C Dakin; Li Guo; Farzana Rahman; Miles Parnell; Eduardo Normando; Shereen Nizari; Benjamin Davis; Ahmed Younis; M Francesca Cordeiro Journal: BMC Bioinformatics Date: 2014-06-05 Impact factor: 3.169
Authors: Jianyan Huang; Jyotsna Maram; Tudor C Tepelus; Cristina Modak; Ken Marion; SriniVas R Sadda; Vikas Chopra; Olivia L Lee Journal: J Optom Date: 2017-08-07