LAUSR

Abstract Our aim was to compare regeneration of the sub‐basal nerve plexus (SBNP) and superficial nerve terminals (SNT) following corneal epithelial injury. We also sought to compare agreement when quantifying nerve parameters using different image analysis techniques. Anesthetized, female C57BL/6 mice received central 1‐mm corneal epithelial abrasions. Four‐weeks post‐injury, eyes were enucleated and processed for PGP9.5 to visualize the corneal nerves using wholemount immunofluorescence staining and confocal microscopy. The percentage area of the SBNP and SNT were quantified using: ImageJ automated thresholds, ImageJ manual thresholds and manual tracings in NeuronJ. Nerve sum length was quantified using NeuronJ and Imaris. Agreement between methods was considered with Bland‐Altman analyses. Four‐weeks post‐injury, the sum length of nerve fibers in the SBNP, but not the SNT, was reduced compared with naïve eyes. In the periphery, but not central cornea, of both naïve and injured eyes, nerve fiber lengths in the SBNP and SNT were strongly correlated. For quantifying SBNP nerve axon area, all image analysis methods were highly correlated. In the SNT, there was poor correlation between manual methods and auto‐thresholding, with a trend towards underestimating nerve fiber area using auto‐thresholding when higher proportions of nerve fibers were present. In conclusion, four weeks after superficial corneal injury, there is differential recovery of epithelial nerve axons; SBNP sum length is reduced, however the sum length of SNTs is similar to naïve eyes. Care should be taken when selecting image analysis methods to compare nerve parameters in different depths of the corneal epithelium due to differences in background autofluorescence. Highlights4 weeks after corneal epithelial injury overt anterior segment inflammation is absent.Corneal sub‐basal nerve plexus density is reduced 4 weeks after sterile injury.Recovery of the corneal superficial nerve terminals is evident 4 weeks after injury.The image analysis technique will affect absolute nerve density measurements.Automated‐thresholding can overestimate sub‐basal nerve plexus area.