Abstract
Purpose :
The correlation between corneal nerve tortuosity and various pathologies has been shown several times. The clinical perception of corneal subbasal nerve tortuosity was recently shown to have two distinct forms, namely short-range tortuosity (SRT; low amplitude, high frequency) and long-range tortuosity (LRT; high amplitude, low frequency). We investigated if one or both forms of tortuosity index are capable of highlighting differences between healthy and diabetic with neuropathy individuals.
Methods :
Images from 10 healthy (Fig1, a) and 10 Type 1 or 2 diabetic subjects with neuropathy (Fig1, b) were acquired using the Heidelber Retina Tomograph (HRT-II) with the Rostock Cornea Module (Heidelberg Engineering GmbH, Heidelberg, Germany). Several images per subjects were acquired, all covering an area of 400x400 µm (384x384 pixels). For each subject, 5 non-overlapping images from the central part of the cornea were considered for this study. We automatically traced corneal nerves with a custom computerized procedure and a set of mathematical tortuosity measurements was derived. SRT and LRT were computed as a previously identified combination of these measurements.
Results :
For each subject, the average SRT or LRT value obtained in the 5 images was considered. Results revealed that each average index of tortuosity is able to classify subjects, using a simple linear discrimination criterion, with an accuracy of 80% and 70%, respectively (Fig1, c). The best accuracy, 85%, is however achieved by considering both indexes together. We also performed the classification using SRT and LRT derived from each single image, but results revealed a poor ability to differentiate between the two groups, with an accuracy of 63% (Fig1, d).
Conclusions :
We showed that the use of both tortuosity indexes together has a very good capability to differentiate healthy subjects from diabetic with neuropathy ones by analising 5 images from the central part of the cornea in each subject, but not when analising each image separately. The latter is probably due to the large variability of SRT and LRT in the different regions of the cornea and underlines the importance of standardizing the acquisition protocol, which should acquire images from a specific location of the cornea.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.