Quality Control in the Corneal Bank with Artificial Intelligence: Comparison of a New Deep Learning-based Approach with Conventional Endothelial Cell Counting by the “Rhine-Tec Endothelial Analysis System”

Abstract

Endothelial cell density (ECD) is a crucial parameter for the release of corneal grafts for transplantation. The Lions Eye Bank of Baden-Württemberg uses the “Rhine-Tec Endothelial Analysis System” for ECD quantification, which is based on a fixed counting frame method considering only a small sample of 15 to 40 endothelial cells. The measurement result therefore depends on the frame placement and manual correction of the cells counted within the frame. To increase the sample size and create higher objectivity, we developed a new method based on “deep learning” that automatically detects all visible endothelial cells in the image. This study aims to compare this new method with the conventional Rhine-Tec system. 9375 archived phase-contrast microscopic images of consecutive grafts from the Lions Eye Bank were evaluated with the deep learning method and compared with the corresponding archived analyses of the Rhine-Tec system. Means, Bland-Altman and correlation analyses were compared. Comparable results were obtained for both methods. The mean difference between the Rhine-Tec system and the deep learning method was only − 23 cells/mm² (95% confidence interval − 29 to − 17). There was a statistically significant positive correlation between the two methods, with a correlation coefficient of 0.748. What was striking in the Bland-Altman analysis were clustered deviations in the cell density range between 2000 and 2500 cells/mm² – with higher values in the Rhine-Tec system. The comparable results for cell density measurement values underline the validity of the deep learning-based method. The deviations around the formal threshold for graft release of 2000 cells/mm² are most likely explained by the higher objectivity of the deep learning method and the fact that measurement frames and manual corrections were specifically selected to reach the formal threshold of 2000 cells/mm² when the full area endothelial quality was good. This full area assessment of the graft endothelium cannot currently be replaced by deep learning methods and remains the most important basis for graft release for keratoplasty.

Keywords

corneal transplantation - deep learning - corneal endothelium - corneal bank - fully automated

• References/Literatur

• 1 Nishimura JK, Hodge DO, Bourne WM. Initial endothelial cell density and chronic endothelial cell loss rate in corneal transplants with late endothelial failure. Ophthalmology 1999; 106: 1962-1965
• 2 Armitage WJ, Dick AD, Bourne WM. Predicting endothelial cell loss and long-term corneal graft survival. Invest Ophthalmol Vis Sci 2003; 44: 3326-3331
• 3 Laule A, Cable MK, Hoffman CE. et al. Endothelial cell population changes of human cornea during life. Arch Ophthalmol 1978; 96: 2031-2035
• 4 Al-Fahdawi S, Qahwaji R, Al-Waisy AS. et al. A fully automated cell segmentation and morphometric parameter system for quantifying corneal endothelial cell morphology. Comput Methods Programs Biomed 2018; 160: 11-23
• 5 Selig B, Vermeer KA, Rieger B. et al. Fully automatic evaluation of the corneal endothelium from in vivo confocal microscopy. BMC Med Imaging 2015; 15: 13
• 6 Gavet Y, Pinoli JC. Comparison and supervised learning of segmentation methods dedicated to specular microscope images of corneal endothelium. Int J Biomed Imaging 2014; 2014: 704791
• 7 Reinhard T, Spelsberg H, Holzwarth D. et al. Wissensbasierte Bildanalyse des Endothels von Hornhauttransplantaten. Klin Monbl Augenheilkd 1999; 214: 407-411
• 8 Deb-Joardar N, Thuret G, Gavet Y. et al. Reproducibility of endothelial assessment during corneal organ culture: comparison of a computer-assisted analyzer with manual methods. Invest Ophthalmol Vis Sci 2007; 48: 2062-2067
• 9 Hirneiss C, Schumann RG, Grüterich M. et al. Endothelial cell density in donor corneas: a comparison of automatic software programs with manual counting. Cornea 2007; 26: 80-83
• 10 Abib FC, Holzchuh R, Schaefer A. et al. The endothelial sample size analysis in corneal specular microscopy clinical examinations. Cornea 2012; 31: 546-550
• 11 Fabijańska A. Segmentation of corneal endothelium images using a U-Net-based convolutional neural network. Artif Intell Med 2018; 88: 1-13
• 12 Daniel MC, Atzrodt L, Bucher F. et al. Automated segmentation of the corneal endothelium in a large set of ‘real-world’ specular microscopy images using the U-Net architecture. Sci Rep 2019; 9: 4752
• 13 Heinzelmann S, Daniel MC, Maier PC. et al. Automatisierte Zellzählung in Spenderhornhäuten aus Organkultur mittels „Deep Learning“ erreicht hohe Präzision und Genauigkeit. Klin Monbl Augenheilkd 2019; 236: 1407-1412
• 14 Ronneberger O, Fischer P, Brox T. UNet: Convolutional Networks for biomedical Image Segmentation. In: Navab N, Hornegger J, Wells WM, Frangi AF. eds. Medical Image Computing and Computer-assisted Intervention – MICCAI 2015. Cham: Springer International Publishing; 2015: 234-241
• 15 Akeret J, Chang C, Lucchi A. et al. Radio frequency interference mitigation using deep convolutional neural networks. Astron Comput 2017; 18: 35-39
• 16 Schroeter J, Rieck P. Endothelial evaluation in the cornea bank. Dev Ophthalmol 2009; 43: 47-62
• 17 Schroeter J, Maier P, Bednarz J. et al. Arbeitsrichtlinien. Gute Fachliche Praxis für Hornhautbanken. Ophthalmologe 2009; 106: 265-274 276
• 18 Amann J, Holley GP, Lee SB. et al. Increased endothelial cell density in the paracentral and peripheral regions of the human cornea. Am J Ophthalmol 2003; 135: 584-590