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

PURPOSE To analyze the repeatability of measuring nerve fiber length (NFL) from images of the human corneal subbasal nerve plexus using semiautomated software. METHODS Images were captured from the corneas of 50 subjects with type 2 diabetes mellitus who showed varying severity of neuropathy, using the Heidelberg Retina Tomograph 3 with Rostock Corneal Module. Semiautomated nerve analysis software was independently used by two observers to determine NFL from images of the subbasal nerve plexus. This procedure was undertaken on two occasions, 3 days apart. RESULTS The intraclass correlation coefficient values were 0.95 (95% confidence intervals: 0.92-0.97) for individual subjects and 0.95 (95% confidence intervals: 0.74-1.00) for observer. Bland-Altman plots of the NFL values indicated a reduced spread of data with lower NFL values. The overall spread of data was less for (a) the observer who was more experienced at analyzing nerve fiber images and (b) the second measurement occasion. CONCLUSIONS Semiautomated measurement of NFL in the subbasal nerve fiber layer is highly repeatable. Repeatability can be enhanced by using more experienced observers. It may be possible to markedly improve repeatability when measuring this anatomic structure using fully automated image analysis software.