Automated Image Registration Method for In Vivo Confocal Microscopy of the Corneal Sub-basal Nerve Plexus

Abstract

In vivo confocal microscopy (IVCM) assesses corneal innervation in the sub-basal nerve plexus but is typically quantified manually from a single z-scan, limiting biomarker extrapolation for diagnosing limbal stem cell deficiency (LSCD). We developed an automated 3D reconstruction method of IVCM image volumes to improve sub-basal nerve density quantification. Our dataset comprised 99 IVCM stacks from 63 LSCD eyes (51 patients) and 23 stacks from 15 normal eyes. We designed an image registration algorithm combining phase correlation and homography transformation, which achieved a pairwise image correlation of 0.69 and mutual information of 0.60, significantly outperforming manual registration (0.60 and 0.43, respectively; p<0.001). Validation on an independent dataset of 325 volume scans from 24 eyes of 12 unilateral, severe LSCD patients yielded a correlation of 0.75 and MI of 0.76. This method enhances sequential IVCM scan alignment and supports more accurate, reproducible 3D evaluation of LSC biomarkers.

Figure 1.

Diagram outlining the registration process on a subset of z-scans. Beginning with an unregistered volume scan, the algorithm determines whether to use an untransformed image, the phase correlation approach, or the homography transform approach to deform z-scans to generate a registered volume. The maximum pixel value at each point (x, y) in the image stack of registered z-scans is selected for the image projection.

Figure 2.

Comparison boxplots stratified by disease severity of a) fiber count and b) standardized fiber length derived from proposed registration method and manual registration

Figure 3.

Comparison of 2D registrations generated with manual registration, phase correlation only (affine transformation), homography transform only (perspective transformation), and our combined registration approach for a) control, b) mild, c) moderate (cropped for detail), and d) severe example cases. Aliasing is present in each of the manually registered projections, particularly noticeable in the mild case (b) in the bottom left corner of all methods except the proposed approach, and the manual registration for the severe case (d).

Figure 4.

Comparison of a) three sequential scans in a z-series and maximum projections for each registration method b) no registration, c) manual registration, and d) our proposed registration method