Schaldemose E L, Fontain F I, Karlsson P, Nyengaard J R
Danish Pain Research Center, Aarhus University Hospital, Aarhus, Denmark.
Department of Clinical Medicine - Core Center for Molecular Morphology, Section for Stereology and Microscopy, Aarhus University Hospital, Aarhus, Denmark.
J Microsc. 2017 Oct;268(1):3-12. doi: 10.1111/jmi.12581. Epub 2017 May 26.
Corneal confocal microscopy (CCM) is a noninvasive clinical method to analyse and quantify corneal nerve fibres in vivo. Although the CCM technique is in constant progress, there are methodological limitations in terms of sampling of images and objectivity of the nerve quantification. The aim of this study was to present a randomized sampling method of the CCM images and to develop an adjusted area-dependent image analysis. Furthermore, a manual nerve fibre analysis method was compared to a fully automated method.
23 idiopathic small-fibre neuropathy patients were investigated using CCM. Corneal nerve fibre length density (CNFL) and corneal nerve fibre branch density (CNBD) were determined in both a manual and automatic manner. Differences in CNFL and CNBD between (1) the randomized and the most common sampling method, (2) the adjusted and the unadjusted area and (3) the manual and automated quantification method were investigated.
The CNFL values were significantly lower when using the randomized sampling method compared to the most common method (p = 0.01). There was not a statistical significant difference in the CNBD values between the randomized and the most common sampling method (p = 0.85). CNFL and CNBD values were increased when using the adjusted area compared to the standard area. Additionally, the study found a significant increase in the CNFL and CNBD values when using the manual method compared to the automatic method (p ≤ 0.001).
The study demonstrated a significant difference in the CNFL values between the randomized and common sampling method indicating the importance of clear guidelines for the image sampling. The increase in CNFL and CNBD values when using the adjusted cornea area is not surprising. The observed increases in both CNFL and CNBD values when using the manual method of nerve quantification compared to the automatic method are consistent with earlier findings. This study underlines the importance of improving the analysis of the CCM images in order to obtain more objective corneal nerve fibre measurements.
角膜共焦显微镜检查（CCM）是一种用于在体内分析和量化角膜神经纤维的非侵入性临床方法。尽管CCM技术在不断进步，但在图像采样和神经量化的客观性方面仍存在方法学上的局限性。本研究的目的是提出一种CCM图像的随机采样方法，并开发一种调整后的面积依赖性图像分析方法。此外，还将手动神经纤维分析方法与全自动方法进行了比较。
使用CCM对23例特发性小纤维神经病变患者进行了研究。以手动和自动方式测定角膜神经纤维长度密度（CNFL）和角膜神经纤维分支密度（CNBD）。研究了（1）随机采样方法与最常用采样方法之间、（2）调整后的面积与未调整的面积之间以及（3）手动和自动量化方法之间CNFL和CNBD的差异。
与最常用的方法相比，使用随机采样方法时CNFL值显著降低（p = 0.01）。随机采样方法与最常用采样方法之间的CNBD值无统计学显著差异（p = 0.85）。与标准面积相比，使用调整后的面积时CNFL和CNBD值增加。此外，研究发现与自动方法相比，使用手动方法时CNFL和CNBD值显著增加（p≤0.001）。
该研究表明随机采样方法与常用采样方法之间的CNFL值存在显著差异，表明图像采样明确指南的重要性。使用调整后的角膜面积时CNFL和CNBD值增加并不奇怪。与自动方法相比，使用手动神经量化方法时观察到的CNFL和CNBD值增加与早期研究结果一致。本研究强调了改进CCM图像分析以获得更客观的角膜神经纤维测量结果的重要性。
