Correcting Lateral Magnification in OCT Imaging of the Rat Eye

Abstract

Purpose: Lateral magnification in the rat eye is approximately five times that of the human eye. Thus, OCT scan locations will correspond with different retinal positions when imaging eyes with even small differences in axial length. In this study we specify model eye parameters to compensate for this lateral magnification and evaluate the consistency of retinal thickness estimation in the normal rat eye.

Methods: Ultrasonic immersion A-scans were performed in 10-28 month-old Brown Norway rats (n=50 eyes) to measure anterior chamber depth (ACD), lens thickness (LT), vitreous chamber depth (VCD), and axial length (AL). Average values for each parameter were then used to specify a three surface schematic eye, from which a lateral magnification scaling function (µm/degree) was derived. The resulting model was then used to scale the lateral dimensions of volumetric (n=31 B-scans) SD-OCT images in 8 rats (Spectralis, Heidelberg Engineering). Measurements of nerve fiber layer and ganglion cell layer (NFL/GCL), NFL/GCL to inner plexiform layer (NFL/GCL+IPL) and total retinal thicknesses were then taken from a circular scan pattern (d= 1.25 mm) centered on the optic nerve head (ONH). ONH diameter and area were measured from annotations of Bruch’s membrane opening. Linear regression analysis was performed to determine whether thickness measurements or ONH area were dependent upon axial length after image scaling.

Results: Model parameters (mean ± SD) were ACD: 1.07 ± 0.15 mm, LT: 4.49 ± 0.37 mm, VCD: 1.39 ± 0.20 mm, and AL: 6.95 ± 0.40 mm. The resulting angular scaling factor was 61.36±3 µm/degree (mean ± 95% CI). After correction for lateral magnification, retinal layer thickness measurements were (mean ± SD): NFL/GCL: 30 ± 2 µm, NFL/GCL+IPL: 75 ± 2 µm, Total retinal thickness: 193 ± 5 µm. The average ONH diameter was 290 ± 17 µm and ONH area was 66,446 ± 8,114 µm². Univariate linear regression after scaling showed that none of these parameters varied as a function of axial length (all R² < 0.15; P > 0.3).

Conclusions: The model eye parameters provide a basis for consistent scaling of SD-OCT images across animals in the normal rat eye. The small variation observed in retinal thickness measurements and lack of axial length dependence in these measures suggests that comparable retinal locations were identified between animals when using this method.

Keywords: 549 image processing • 552 imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • 551 imaging/image analysis: non-clinical