Where is retinal optical coherence tomography heading?

Optical coherence tomography (OCT) is a rapidly emerging, noninvasive, optical diagnostic-imaging technique that enables in-vivo cross-sectional visualization of internal microstructure in biological systems at a resolution of a few micrometers.¹ Novel high-speed detection techniques and the recent development of tunable light sources with ultrabroad bandwidth have revolutionized imaging performance and the clinical feasibility of OCT. As a consequence, it has become an optical analog to computed tomography and magnetic-resonance imaging. Although OCT does not enable full-body imaging, it facilitates noninvasive optical biopsy, i.e., micron/cellular-resolution 3D visualization of tissue morphology.²

The eye provides easy optical access to both the anterior segment and the retina because of its essentially transparent nature. For this reason, ophthalmic (and especially retinal) imaging has so far been the first and most successful clinical application for OCT. After all, nearly 50% of all OCT publications to date have been published in ophthalmic journals, thus supporting this observation. In addition, more than half a dozen companies offer the technology in its fourth generation. Consequently, 3D retinal OCT represents the fastest adopted imaging technology in the history of ophthalmology.

Figure 1. 3D optical-coherence tomography (OCT) at 800 and 1060nm of (a)–(d) a normal retina and (e)–(h) a patient with retinitis pigmentosa. (b, f) High-definition (4096 depth scans) 800nm 3D OCT scan over 35°. (d, h) High-definition (2048-pixel) 1060nm 3D OCT scan over 35°. (c, g) En-face wide-field (35°×35°). (a, e) En-face zoomed-in fundus image of the choroid using 1060nm 3D OCT. Arrows indicate enhanced choroidal visualization. (Adapted from Považay and collaborators.³)

While third-generation commercial retinal scanning systems (Stratus OCT) were based on time-domain OCT, enabling up to 400 ‘A scans’ (1D depth measurements) per second, fourth-generation (spectral- or Fourier-domain-based) systems can presently perform up to 100 times more measurements, delivering either highly sampled (high-definition) 2D tomograms or 3D (volumetric) imaging of the retina. The clinical benefit is demonstrated by highly sampled 3D visualization of the retina in a reasonably short data-acquisition period, resulting in more reliable and reproducible 2D thickness maps of (intra-)retinal layers. In turn, this might lead to improved diagnosis of retinal pathologies or better understanding of retinal pathogenesis, and enhance (objective) monitoring of novel therapy approaches.

Thus far, commercial retinal OCT has been performed mainly in the 800nm regime. Although these systems can resolve all major intraretinal layers, they do not penetrate far beyond the retina, resulting in only limited visualization of the choriocapillaris and choroid. In clinical OCT, turbid ocular media (e.g., cataracts) represent a significant challenge in retinal imaging, however. Since scattering in biological tissues decreases monotonically with increasing wavelength, OCT imaging at 1060nm can achieve deeper tissue penetration, thus enabling delineation of choroidal structure.^4,5

In parallel to several other groups, we have initiated 3D OCT at 1060nm and demonstrated enhanced visualization of the choroid up to the sclera. We use a cost-effective, easy-to-implement system based on a high-speed indium-gallium arsenide linear 1024-pixel array (SUI-Goodrich), enabling 47,000 A scans per second at 5–8μm axial resolution with a 2.6mm tissue scanning depth. Figure 1shows a comparison of 3D OCT at 800 versus 1060nm in a normal and a pathologic eye.^{3, 5}

Figure 2. Possible strategy for clinical cellular-resolution OCT in a patient with Type-2 Macular Telangiectasia. (a) Prescreening over 20°×20° (512×128 depth scans) using a commercial 3D OCT scanner at 800nm. (b) Detection of impaired intraretinal morphology using a representative cross section from (a). Zoom in at (c)–(e) normal (yellow dashed rectangle in b) and (f)–(h) pathologic (white dashed rectangle in b) eyes using cellular-resolution OCT. Volumetric rendering at (c) 6°surrounding the retinal fovea and (f) 0°. En-face images at the level of (d) the capillaries in the inner nuclear layer at 6°and (e) the tips of the outer photoreceptors at 6°, extracted from (c). (g) Cross sections and (h) en-face images at the level of the retinal-pigment epithelium, extracted from (f). (Adapted from Považay and collaborators.³)

For ophthalmic retinal OCT imaging, the cornea and lens act as the ‘imaging objective,’ thus determining the numerical aperture and hence the beam diameter in the retina. The latter specifies the transverse OCT resolution, which is on the order of ∼20μm for a beam of ∼1mm diameter (at 800nm), approximately one order of magnitude worse than common axial OCT resolutions. This can be improved by dilating the pupil and increasing the measurement beam diameter. However, for large pupil diameters, ocular aberrations limit the minimum focused spot size on the retina, even for monochromatic illumination. An alternative and promising approach uses adaptive optics (AO)—originally developed to improve the resolution of astronomical imaging—to minimize ocular aberrations and reduce the spot size.

We have interfaced AO with 3D ultrahigh-resolution OCT for in vivo cellular-resolution retinal imaging. We used a deformable mirror (Mirao52, Imagine Eyes, France) with a unique performance in terms of amplitude (±50μm stroke) and linearity, allowing for correcting highly aberrated normal or pathologic eyes.⁶ We also used a 140–160nm titanium-sapphire laser (Femtolasers Integral, Femtolaser, Vienna, Austria) and a CMOS Basler sprint spL4096-140k camera (Basler AG, Germany), generating 160,000 A scans per second (1536 pixels each) and resulting in ultrahigh-speed cellular-resolution retinal imaging with an isotopic resolution of 2–3μm. Figure 2 demonstrates clinical cellular-resolution AO OCT in a patient with Type-2 Macular Telangiectasia. A commercial 3D OCT system is used to prescreen a larger volume to identify suspicious locations. These areas are then investigated (for the time being still using a separate system) with AO OCT at cellular-resolution level, revealing quite normal vasculature and photoreceptor appearance at 6° surrounding the fovea (the small area responsible for our central, sharpest vision), compared to severe impairment at 0°. These state-of-the-art retinal OCT technologies will serve as prerequisites to significantly improve functional extensions like optophysiology, Doppler, spectrosopic-, or polarization-sensitive OCT in the near future.

Commercial fourth-generation 3D ophthalmic OCT systems seem to have a significant diagnostic impact in both daily clinical routine and novel therapeutic trials. After introducing 3D visualization of the healthy and pathologic human retina, the emphasis is now shifting toward filtering out proper biomarkers for improved diagnosis. It is not obvious where retinal OCT is heading. Despite several proof-of-principle demonstrations of significantly improved state-of-the-art retinal OCT imaging, clinical studies of larger, properly selected patient cohorts are required to demonstrate improved clinical impact of and therefore justify the increased technological investment required for novel retinal OCT techniques.

The author acknowledges all members of the biomedical-imaging group in the School of Optometry and Vision Sciences of Cardiff University, and Alan C. Bird and Catherine A. Egan from the Medical Retina Service at Moorfields Eye Hospital, London (UK). Financial and equipment support from the following institutions is also acknowledged: Cardiff University, the European Commission's Sixth and Seventh Framework Programme projects FP6-IST-NMP-2 STREPT (017128) and FUN OCT (FP7 Health, contract 201880), Action Medical Research (contract AP1110), the UK Department of Trade and Industry (project 1544C), FEMTOLASERS GmbH, Carl Zeiss Meditec Inc., Maxon Computer GmbH, Multiwave Photonics, and the Lowy Foundation, Sydney (Australia).