Rapid image evaluation system for corneal in vivo confocal microscopy

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Image Analysis

Background:

• Corneal in vivo confocal microscopy (IVCM) provides detailed microstructural imaging.
• Objective analysis of IVCM images is crucial for diagnosing and monitoring ocular conditions.
• Current analysis methods can be time-consuming and subjective.

Purpose of the Study:

• To develop and validate rapid image processing techniques for objective analysis of corneal IVCM.
• To automate the detection and quantification of key corneal cellular structures.

Main Methods:

• Utilized laser IVCM to obtain central corneal volume scans from healthy volunteers.
• Developed software employing image feature analysis (size, shape, contrast) to detect nerve plexus layers.
• Implemented algorithms for dendritic immune cell detection based on size, brightness, and nerve association.
• Applied regional minima and watershed transformations for epithelial wing cell identification.

Main Results:

• Software successfully identified nerve layers in all 15 scans analyzed.
• High agreement between manual and automated analysis: 81.8% for immune cells and 94.4% for wing cells.
• Rapid analysis (under 10 seconds per scan) provided quantitative data on cell counts, dimensions, and density.
• Analysis included parameters for nerve and cell shape and position.

Conclusions:

• The developed software offers a rapid and accurate method for analyzing corneal IVCM.
• Quantitative data on corneal nerves, immune cells, and wing cells can be efficiently obtained.
• This technique has potential for objective assessment in various ocular surface diseases.