Abstract
Purpose :
We present a novel Sensorless Adaptive Optics OCT (SAO-OCT) capable of visualizing both retinal vasculature and Retinal Pigment Epithelium (RPE) in a clinical setting. By controlling the pupil size according to the image Field of View (FOV), diffraction-limited performance can be achieved at different imaging scales. In addition, by employing a polarization diversity detection scheme, Randomness of Polarization (RP) is measured and utilized for RPE segmentation of both its morphological features and tissue property.
Methods :
Transmissive adaptive elements and a zoomable collimator allowed multi-scale OCT imaging with variable pupil size. Three different image FOVs, from wide (9 mm x 9 mm), middle (6 mm x 6 mm), to small (3 mm x 3 mm), were acquired with corresponding beam diameters of 1 – 3 mm at the pupil for distinct lateral resolutions. For middle and large FOVs, only defocus was corrected, while small FOVs also corrected astigmatism. From a single imaging session, comprehensive retinal information was obtained via morphological structure, angiography, and RP. Aberration correction allowed high-resolution OCT angiography (SAO-OCTA), while RP measurement permitted exclusive identification and segmentation of the RPE using its intrinsic properties. The segmented RPE's topography was mapped to visualize retinal curvature and retinal detachments. By combining OCTA and RPE maps, we provide en face images with vessels and RPE topology observed simultaneously. Human imaging was ethics-approved at SFU.
Results :
A representative image set from a healthy volunteer is shown in Fig. 1, taken in a single session. Fig. 1(a) is used for overall morphological contrast, with regions of interest visualized in high-resolution in (b)-(c) by both vasculature and RP contrast; vessels in blue, RPE elevation map in yellow (high) to red (low). The RPE is uniform and continuous, shown with gentle gradation due to eye curvature. Contrarily, in pathological subjects, RP contrast clearly visualized RPE disruption as sudden colour discontinuities, and its relative elevation map with detailed vasculature provide a more comprehensive retinal image, which will be presented.
Conclusions :
Our proposed imaging system and retinal image formation method can provide diverse and comprehensive retinal structure analysis and investigation of retinal pathologies.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.