Elsevier

Survey of Ophthalmology

Volume 58, Issue 5, September–October 2013, Pages 466-475
Survey of Ophthalmology

Diagnostic and Surgical Techniques
Quantitative analysis of in vivo confocal microscopy images: A review

https://doi.org/10.1016/j.survophthal.2012.12.003Get rights and content

Abstract

In vivo confocal microscopy (IVCM) is a non-invasive method of examining the living human cornea. The recent trend towards quantitative studies using IVCM has led to the development of a variety of methods for quantifying image parameters. When selecting IVCM images for quantitative analysis, it is important to be consistent regarding the location, depth, and quality of images. All images should be de-identified, randomized, and calibrated prior to analysis. Numerous image analysis software are available, each with their own advantages and disadvantages.

Criteria for analyzing corneal epithelium, sub-basal nerves, keratocytes, endothelium, and immune/inflammatory cells have been developed, although there is inconsistency among research groups regarding parameter definition. The quantification of stromal nerve parameters, however, remains a challenge. Most studies report lower inter-observer repeatability compared with intra-observer repeatability, and observer experience is known to be an important factor. Standardization of IVCM image analysis through the use of a reading center would be crucial for any future large, multi-centre clinical trials using IVCM.

Section snippets

Image selection and analysis

When selecting IVCM images for quantitative analysis, it is important to be consistent regarding the location (central vs peripheral cornea) and depth of images. Consistency in corneal location may be maximized by the use of fixation targets--internal for slit-scanning IVCM17 or external for laser scanning IVCM.47 Accuracy in determining section depth can be maximized by using fixed landmarks (e.g. measuring keratocyte density immediately posterior to Bowman's layer) or by using devices such as

Nidek Advanced Vision Information System software

Nidek Advanced Vision Information System (NAVIS) Endothelial Analysis Software (Fig. 2), available for use with the ConfoScan IVCM software (Nidek Technologies, Fremont, CA) enables quantitative analysis of in vivo confocal images. The region of interest is easily defined in terms of area, the dimensions of which may be adjusted as required prior to analysis. For corneal endothelial images, analysis may be performed manually, automatically, or using a combination of both techniques. Manual

Corneal epithelium

The corneal epithelium typically consists of five to seven layers of cells, including superficial epithelial cells, wing cells, and basal epithelial cells. Superficial cells have most commonly been quantified in terms of cell diameter (μm) and cell density (cells/mm2); images obtained by IVCM may not be sufficient to demonstrate full detail, however.39 Mocan et al39 have objectively shown that topical fluorescein application prior to IVCM enhances the visualization of the superficial

Corneal sub-basal nerves

Corneal sub-basal nerves are easily and reproducibly imaged because of their location and orientation on Bowman’s layer. Nonetheless, the manner in which nerve density is defined has been somewhat inconsistent in the literature. The majority of studies have defined sub-basal nerve density as the total length of nerves visible within a defined area (μm/mm2 or mm/mm2),44, 48 but some investigators have only included nerve branches longer than 50 μm in their measurements.9 Others have analyzed the

Stromal nerves

Quantitative analysis of stromal nerves imaged by IVCM remains controversial. A wide range of values for stromal nerve diameter have been reported, and this variation is due to a number of factors. First, stromal nerves commonly traverse obliquely relative to the en face section of IVCM images. These nerves will therefore appear shorter than one whose path is parallel to the plane of the image (Fig. 4). The visible nerve length per frame area will also depend on the axial resolution of the

Corneal endothelium

Manual analysis of endothelial cell density simply involves counting the number of cells within a predefined frame. This traditional form of analysis only provides information regarding the cell density and number of cells counted and does not supply any morphometric data. In contrast, contemporary software applications such as the NAVIS software of the ConfoScan IVCM (see the Image Analysis Software section) provide data regarding the number of cells counted, cell density, age-matched normal

Corneal reflectivity

Measurement of corneal reflectivity is an objective method of assessing corneal haze.40 Most investigators have expressed corneal haze in terms of the specific units of image intensity from the instrument used for measurement; however, the image “brightness” of light back-scattered from corneal haze can only be compared with brightness measured at a different time in longitudinal studies or across laboratories if the IVCM instruments are standardized so that units that express haze intensity

Corneal thickness

The slit-scanning IVCM has poor repeatability for corneal thickness measurements, exhibiting the widest 95% limits of agreement both within and between sessions when compared with ultrasound, optical coherence tomography, and Orbscan.67

This is because the position of the cornea relative to the objective lens varies throughout the scan acquisition, and the error is compounded by the several second length of scanning time during which involuntary axial motion of the eye is inevitable.33

The

Immune/inflammatory cells

Zhivov et al68 were first to report in vivo evaluation of Langerhans cells within the human corneal epithelium. Quantification of Langerhans cell density is achieved by counting the number of these cells per image frame.29, 66, 68 Keratic precipitates have also been quantitatively evaluated in Fuchs heterochromic cyclitis. Characteristics such as density, diameter, area, the ratio between the total size and the body size, and number of pseudopodia are analyzed.13, 27

Reading centers

The inter-observer variability of many of the quantitative parameters discussed in this review raises the question whether there should be reading centers for IVCM image analysis. A reading center is a central facility specializing in the standard evaluation of images. Reading centers are commonly used in multicenter clinical trials. Usually, each study center requires certification by the reading center and must use the center's imaging protocol.65 In the case of IVCM, a minimum level of

Conclusion

Quantitative analysis is an increasingly common feature of studies using IVCM. It is clear that appropriate image selection and randomization are crucial prior to analysis. There are several software applications available for quantifying IVCM images, each with specific advantages and disadvantages. There are also multiple ways of defining morphological parameters and there is currently no consensus regarding “gold standard” definitions of parameters such as sub-basal nerve density, making

Method of literature search

Searches were performed using PubMed and Medline, and all years were searched. The following search terms were used: In vivo confocal microscop, AND cornea, corneal endothelium, corneal epithelium, keratocyte, sub-basal nerves, corneal haze, corneal reflectivity, corneal thickness, repeatability, and reproducibility.

All articles judged to be of relevance to quantitative analysis of IVCM images were included, and case reports were excluded. English articles and non-English language articles with

Disclosure

The authors reported no proprietary or commercial interest in any product mentioned or concept discussed in this article.

References (69)

  • M.J. Doughty et al.

    Assessment of the reliability of human corneal endothelial cell-density estimates using a noncontact specular microscope

    Cornea

    (2000)
  • A. Eckard et al.

    In vivo investigations of the corneal epithelium with the confocal rostock laser scanning microscope (rlsm)

    Cornea

    (2006)
  • N. Efron et al.

    Repeatability of measuring corneal subbasal nerve fiber length in individuals with type 2 diabetes

    Eye Contact Lens

    (2010)
  • J.C. Erie et al.

    The effect of age on the corneal subbasal nerve plexus

    Cornea

    (2005)
  • C.N. Grupcheva et al.

    Assessing the sub-basal nerve plexus of the living healthy human cornea by in vivo confocal microscopy

    Clin Experiment Ophthalmol

    (2002)
  • D.A. Harrison et al.

    Morphology of corneal basal epithelial cells by in vivo slit-scanning confocal microscopy

    Cornea

    (2003)
  • S. Hasler et al.

    In vivo confocal microscopy of keratic precipitates in fuchs heterochromic uveitis syndrome

    Klin Monbl Augenheilkd

    (2009)
  • P. Hertz et al.

    Reproducibility of in vivo corneal confocal microscopy as a novel screening test for early diabetic sensorimotor polyneuropathy

    Diabet Med

    (2011)
  • T. Hillenaar et al.

    Normative database for corneal backscatter analysis by in vivo confocal microscopy

    Invest Ophthalmol Vis Sci

    (2011)
  • T. Hillenaar et al.

    Wide-range calibration of corneal backscatter analysis by in vivo confocal microscopy

    Invest Ophthalmol Vis Sci

    (2011)
  • T. Hillenaar et al.

    Endothelial involvement in herpes simplex virus keratitis: an in vivo confocal microscopy study

    Ophthalmology

    (2009)
  • T.J. Holmes et al.

    Automated software analysis of corneal micrographs for peripheral neuropathy

    Invest Ophthalmol Vis Sci

    (2010)
  • V. Hurmeric et al.

    The relationship between corneal biomechanical properties and confocal microscopy findings in normal and keratoconic eyes

    Cornea

    (2010)
  • L. Imre et al.

    Reliability and reproducibility of corneal endothelial image analysis by in vivo confocal microscopy

    Graefes Arch Clin Exp Ophthalmol

    (2001)
  • S. Jonuscheit et al.

    In vivo confocal microscopy of the corneal endothelium: comparison of three morphometry methods after corneal transplantation

    Eye

    (2011)
  • P. Kallinikos et al.

    Corneal nerve tortuosity in diabetic patients with neuropathy

    Invest Ophthalmol Vis Sci

    (2004)
  • A.S. Kitzmann et al.

    Comparison of corneal endothelial cell images from a noncontact specular microscope and a scanning confocal microscope

    Cornea

    (2005)
  • C.M.C. Klais et al.

    Comparison of endothelial cell count using confocal and contact specular microscopy

    Ophthalmologica

    (2003)
  • T. Kojima et al.

    The application of in vivo laser scanning confocal microscopy as a tool of conjunctival in vivo cytology in the diagnosis of dry eye ocular surface disease

    Mol Vis

    (2010)
  • A. Labbe et al.

    Evaluation of keratic precipitates and corneal endothelium in fuchs' heterochromic cyclitis by in vivo confocal microscopy

    Br J Ophthalmol

    (2009)
  • E.C. Ledbetter et al.

    In vivo confocal microscopy of the normal equine cornea and limbus

    Vet Ophthalmol

    (2009)
  • H. Lin et al.

    Changes in corneal epithelial layer inflammatory cells in aqueous tear-deficient dry eye

    Invest Ophthalmol Vis Sci

    (2010)
  • J.W. McLaren et al.

    Automated assessment of keratocyte density in stromal images from the confoscan 4 confocal microscope

    Invest Ophthalmol Vis Sci

    (2010)
  • J.W. McLaren et al.

    Standardization of corneal haze measurement in confocal microscopy

    Invest Ophthalmol Vis Sci

    (2010)
  • Cited by (82)

    • A Review On digital image processing techniques for in-Vivo confocal images of the cornea

      2021, Medical Image Analysis
      Citation Excerpt :

      It is well known that corneal clarity, which is related to corneal quality because it permits to focus the incident light onto the retina, is correlated to the endothelial cell density and morphology. However, accurate determination of endothelial cell density using IVCM built-in software (e.g., ConfoScan proprietary image analysis program) may require the manual correction of cells boundaries Kitzmann S et al. (2005); Patel and McGhee (2013) (i.e., a semi-automatic image analysis Patel and McGhee (2013)). It is therefore tedious and time consuming Patel and McGhee (2013); Selig et al. (2015); Scarpa and Ruggeri (2016b); Fabijańska (2017); Daniel et al. (2019), and usually impractical in a clinical setting Patel and McGhee (2013); Al-Fahdawi et al. (2018).

    • Corneal fibroblasts: Function and markers

      2020, Experimental Eye Research
    View all citing articles on Scopus
    View full text