Abstract
Purpose: :
In vivo confocal microscopy yields images that are useful for evaluating corneal health, but the images contain substantial information that is difficult to analyze. This project was designed to develop robust, rapid, and objective image processing techniques that would facilitate the quantitation of changes in corneal microstructure.
Methods: :
From an image database of normal subjects who had consented to scans with a HRT Rostock Corneal Module (Heidelberg), the investigators selected 10 representative scans to develop ARIES. Each scan had at least 40 images, which were 80 µm deep and in 2 µm increments through the central cornea. ARIES was designed to automatically extract and quantify wing and dendritic immune cells and compare to manual analyses. The software also identifies the most prominent nerve section and automatically registers the surrounding sections and reconstructs and displays the nerve plexus in the 3D space. Additional scans (>390) from the database were used for software validation.
Results: :
Within 15 seconds, ARIES provides more reproducible, accurate, and comprehensive results than manual assessments. Automated analyses offer several additional parameters that are not practical with manual analysis, including average cell size, proximity to nerves, and shape factors (surface area, perimeter, circularity, diameter, etc). In the case of complex images, a post-processing capability is available to refine the automated results. Unlike manual analyses, ARIES generates a 3D image of the nerve plexus and provides quantitative data including the length, orientation, and number of nerve branches. All data is easily exported for statistical analysis.
Conclusions: :
ARIES provides researchers and clinicians with tools to objectively and rapidly quantify changes in corneal health parameters related to epithelial cells, immune cells, and nerves. These advances may facilitate evaluations of the efficacy of a corneal treatment or the progression of a corneal disorder.
Keywords: image processing • cornea: clinical science • imaging/image analysis: clinical