Abstract
Purpose :
We have developed a novel technique to estimate the shape of the foveola using a commercially available OCT.
Methods :
After dilating the left pupil with 1% Tropicamide, 3 male subjects (ages 27, 30 and 50 years) fixated on the internal OCT target and the OCT (Spectralis) was adjusted until a foveal reflex was visible in the B-scan. A Badal target system, displaying concentric rings on a smart phone screen, was aligned with the internal OCT target. Five B-scans (30° scan, 100 ART) were collected while the subjects fixated on the central target and then rotated their eye to fixate on the intersection of the OCT scan line and the rings of the Badal target (-9.3° to 9.3° in 3.1° steps) in both the horizontal and vertical directions. The location of the reflex within each B-scan gave an estimate of the normal angle at the inner limiting membrane (ILM). B-scans were converted to Cartesian coordinates by correcting for a nominal amount of radial distortion. The foveal region of each off-axis B-scan was aligned with an on-axis B-scan using an ‘image registration’ algorithm in Matlab. The lateral distance of the reflex in each ‘registered’ image was estimated (Fig 1). Reflex angles and distances were used to create slope functions (y’ = b + cx) then integrated to estimate the shape of the foveola (y = a + bx + ½ cx2 where a = 0). The radius of the ILM was estimated from the linear fit coefficients (ie R ≈ 1/c) (Fig 2).
Results :
Slope functions for the 6 foveal meridians had an R2 between 0.784 and 0.953 indicating a strong relationship between the angle and lateral distance of the reflex. The foveal ILM radii of the three subjects were 1.04/0.49 mm, 0.30/0.30 mm and 0.74/0.39 mm for horizontal/vertical meridians. Some slope function estimates indicated that the fovea can be at an angle with respect to the OCT central axis.
Conclusions :
We have developed a technique to estimate foveal shape from central and off-axis OCT images. These estimates are consistent with previous numerical estimates of foveal shape.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.