Abstract
Purpose: :
to develop automatic nerve fibre recognition software optimized for the statistically reliable morphometric characterization of corneal nerve patterns.
Methods: :
Confocal microscopy (Heidelberg Retina Tomograph II / Rostock Cornea Module; image size 400400 µm, 384384 pixels, 8 bit) was performed in healthy volunteers as well as patients with diabetes mellitus. An in-house developed image processing algorithm based on phase correlation was used to analyze and reduce motion artifacts in volume scan image sequences and implemented in C++. The interactive 3D surface plot plug-in for ImageJ (ImageJ 1.40g, http://rsb.info.nih.gov/ij/) was used to visualize the subbasal nerve plexus (SNP) layer in form of the depth map in 3D, with nerve fibers clearly visible even in anterior corneal mosaic (ACM) areas. Consequently, an in-house developed automated morphological and topological quantification of nerve fibres of SNP was performed.
Results: :
Automated morphological and topological quantification algorithm allows the detailed analysis of SNP structures both in healthy cornea and in cases of diabetic neuropathy. The SNP parameters are calculated in a series of image-processing steps and can be used to compare single nerve fibres as well as whole networks of nerve fibres to each other. The elimination of motion artifacts in CLSM volume scans with following reconstruction of SNP structures even in the presence of severe ACM is possible
Conclusions: :
The novel method presented permitted reconstruction and morphological analysis of the SNP layer. This approach provides a sound basis for extended studies of corneal nerve regeneration or degeneration as well as for use in clinical practice.
Keywords: cornea: clinical science • imaging/image analysis: clinical • diabetes