Comparison Of Two Specular Microscopes Devices In Endothelial Clinical Parameters

Published 2023 - 41st Congress of the ESCRS

Reference: PO0247 | DOI: 10.82333/0634-bt19

Authors: Ruiling Xi ¹ , Yinglin Zhang ² , Lingxi Zeng ³ , Lingxi Hu ⁴ , Risa Higashita ¹ , Zheng Chang ⁵ , Chen Lin ⁶ , Jiang Liu* ¹

¹Research Institute of Trustworthy Autonomous Systems and Department of Computer Science and Engineering,Southern University of Science and Technology,Shenzhen,China, ²School of Computer Science, University of Nottingham Ningbo China,Ningbo,China, ³Computer Science and Engineering,Southern University of Science and Technology,Shenzhen,China, ⁴School of Computer Science,University of birmingham,Birmingham,United Kingdom, ⁵Department of Cornea,Shenzhen Aier Eye Hospital Affiliated to Jinan University,Shenzhen,China, ⁶ophthalmology department,Shenzhen People's Hospital,Shenzhen,China

To evaluate the intra-device repeatability and inter-device agreement of endothelial clinical parameters obtained with two specular microscopes.

Shenzhen Aier Eye Hospital, Shenzhen, China.

110 eyes of 55 healthy subjects were included in this cross-sectional study. Average age was 29.9 ± 5.86, and 33 subjects were female. Each eye was captured twice by EM-4000 (Tomey, Japan) and SP-1(Topcon, Japan) on the same visit separately. Endothelial Cell Density (ECD) was recorded by their built-in analysis software. To avoid software-dependent bias, we also introduced a deep learning network with our developed parameter analyzer on captured specular microscope images. Intra-device repeatability and inter-devcie agreements were evaluated using Bland-Altman plots and intraclass correlation (ICC), respectively.

The ECD of EM4000 and SP-1 by the built-in analysis are 2786.89 ± 230.68 and 2977.90 ± 336.87 respectively, while by ourself-developed analyzer are 2996.06 ± 276.72 and 3370.90 ± 317.58. The Bland-Altman plot method demonstrates that the zero difference fell within 95% limits of agreement (LoA) for all experiments. EM-4000 has narrower LoA compared with SP-1, considering both captured image and build-in software analysis. The inter-device measurement agreement between two devices is poor on captured image, with ICC = 0.02, P = 0.89, and r = 0.01. The inter-device measurement agreement between two devices is moderate through build-in analysis software, where ICC = 0.54, P < 0.01, and r = 0.39.

Continuous and proper monitoring of ECD is crucial after cataract refractive surgery to ensure surgical outcomes and detect potential complications. Therefore, a reliable and consistent method for measuring ECD is critical. In this study, we compared two specular microscope devices for ECD estimation. Both EM-4000 and SP-1 showed intra-device repeatability. EM-4000 demonstrated consistently smaller variance in both captured image and built-in software analysis, which might suggest higher stability. Other factors, such as SP-1's higher ECD estimation mean value, may also contribute to its higher variance. The agreement between the two instruments is moderate. Therefore, we do not recommend interchangeable use of EM-4000 and SP-1 microscopes.