Comparative analysis of ocular biometrics using spectral domain optical coherence tomography with Purkinje image and optic nerve head alignments in mice

Abstract

Background: Mice are an emerging model for experimental myopia. Due to their small eye size, non-invasive optical coherence tomography is essential for evaluating ocular biometrics. There is currently no universally accepted protocol for those measurements. This study aims to compare ocular biometric measurements using two methods: Purkinje image-based alignment and optic nerve head alignment, utilizing spectral domain optical coherence tomography. Gaining an understanding of the implications of these methods in determining axial elongation in the normal growing eyes of wild-type C57BL/6J mice would offer valuable insight into their relevance for the experimental myopia model. Methods: Ocular dimensions and refractive development were measured on postnatal days P21 (n = 10), P28 (n = 15), and P35 (n = 8). The Purkinje image-based alignment (P1) was determined using a photorefractor and aligned perpendicular to the corneal apex using SD-OCT. In comparison, due to the absence of a fovea in the mouse retina, the optic nerve head (ONH) alignment was used. Variance analysis, regression analysis, and Bland–Altman analysis were performed to compare the differences between alignment methods as well as the replication by another operator. Results: Mice developed hyperopic ametropia under normal visual conditions. The photorefractor measured a technical variation of 3.9 D (95% CI, n = 170, triplicates). Bland–Altman analysis revealed a shorter (mean ± SD) axial length (− 26.4 ± 18.1 μm) and vitreous chamber depth (− 39.9 ± 25.4 μm) in the Purkinje image-based alignment. There was a significant difference in the relative growth trend in VCD (linear regression, p = 0.02), which was relatively stable and showed shortening when measured with ONH alignment from postnatal age 21 to 35 days. Conclusions: SD-OCT allowed precise in-vivo measurement and segmentation of ocular dimensions, regardless of the methods adopted. P1 alignment consistently resulted in significantly shorter VCD and AL compared to ONH alignment at most time points. When considering temporal changes from P21 to P35, both methods showed similar results, with significant elongation of ACD, LT, and AL as expected. However, our findings revealed a significant shortening of VCD over time with the adoption of ONH alignment, while the change in P1 alignment was relatively stable. Therefore, AL provides a better measure for evaluating ocular growth in mice using optical coherence tomography than VCD for myopia research.