Three-Dimensional Reconstruction of Subbasal Nerve Density in Eyes With Limbal Stem Cell Deficiency: A Pilot Study

Abstract

Purpose: Corneal subbasal nerve parameters have been previously reported using 2-dimensional scans of in vivo laser scanning confocal microscopy (IVCM) in eyes with limbal stem cell deficiency (LSCD). This study aims to develop and validate a method to better quantify corneal subbasal nerve parameters and changes from reconstructed 3-dimensional (3D) images. Methods: IVCM volume scans from 73 eyes with various degrees of LSCD (mild/moderate/severe) confirmed by multimodal anterior segment imaging including IVCM and 20 control subjects were included. Using ImageJ, the scans were manually aligned and compiled to generate a 3D reconstruction. Using filament-tracing semiautomated software (Imaris), subbasal nerve density (SND), corneal nerve fiber length, long nerves (>200 μm), and branch points were quantified and correlated with other biomarkers of LSCD. Results: 3D SND decreased in eyes with LSCD when compared with control subjects. The decrease was significant for moderate and severe LSCD (P < 0.01). 3D SND was reduced by 3.7% in mild LSCD, 32.4% in moderate LSCD, and 96.5% in severe LSCD. The number of long nerves and points of branching correlated with the severity of LSCD (P < 0.0001) and with declining SND (R2 = 0.66 and 0.67, respectively). When compared with 2-dimensional scans, 3D reconstructions yielded significant increases of SND and branch points in all conditions except severe LSCD. 3D analysis showed a 46% increase in long nerves only in mild LSCD (P < 0.01). Conclusions: This proof-of-concept study validates the use of 3D reconstruction to better characterize the corneal subbasal nerve in eyes with LSCD. In the future, this concept could be used with machine learning to automate the measurements.