ABOUT THE AUTHORS

Dr. Makhijani is an ophthalmology resident at Bronx-Lebanon Hospital, New York. He was formerly a software engineer, having worked at Microsoft.
Dr. Lujan is an associate at West Coast Retina, San Francisco, and a research scientist at the University of California, Berkeley.

DISCLOSURES: Drs. Makhijani and Lujan are co-inventors of Directional OCT. Dr. Lujan is also founder of OCTMD.org.

Optical coherence tomography systems have been the subject of numerous advances in hardware and software in recent years, but improvements in techniques for capturing OCT images have followed a parallel track as well.  Enhanced depth imaging (EDI) is a powerful way of overcoming the inherent weaknesses of signal fall off in spectrometer-based systems,1 and high-dynamic range (HDR) display provides tissue-specific visualization windows.2  While these innovations in the acquisition of OCT overcome limitations of current hardware, neither of them leverages the optical properties of the retina itself to reveal new anatomical information.

Another technique is called directional OCT (D-OCT), which involves taking OCT scans of the same patch of retina from multiple pupil entry positions and a set of analysis algorithms for registering these “tilted” images and visualizing information encoded within them.3  

D-OCT works by taking advantage of a property found in photoreceptors called directional reflectivity—the optical property that dictates that light reflects differently when coming from different angles. This property is independent of the OCT hardware system being used, as they are intrinsic to the retina itself. Changes in the reflectivity of different components of the photoreceptors are not only important to recognize for interpretation of OCT images, but directional reflectivity can be exploited to gain additional information about the health of the macula.

This article explores the development of D-OCT, including its applications and limitations.

Variability of Photoreceptors

D-OCT is sensitive to the orientation of the photoreceptor outer and inner segments because of the directional reflectance their wave-guiding properties cause. In healthy eyes, photoreceptors are optimized to point toward the undilated pupil and are actively “steered” to orient themselves to intercept rays of light along their axis.4

Light reflected from the photoreceptors back to the OCT sensor from structures currently called the ellipsoid zone (EZ) and the interdigitation zone (IZ) have been found to be maximal through the optical center of the pupil. D-OCT demonstrates that the reflectivity of the IZ band decreases dramatically and that of the EZ band to a lesser degree when OCTs are acquired through the sides of the pupil. This has been called the optical Stiles-Crawford effect.5

The axons of photoreceptors also demonstrate directional reflectivity, which, along with Müller cell processes, comprise the Henle fiber layer (HFL). Because HFL is made up of a dense bundle of cylindrical filaments, its reflectivity increases as incident light becomes more perpendicular to its orientation.6-8

How OCT Got to Where It Is Today

 
Optical coherence tomography (OCT) has progressed more rapidly from its inception to myriad research and clinical applications than any other ophthalmic technology.

OCT began as an ophthalmic application of the Michelson interferometer. David Huang, MD, PhD, and colleagues at Massachusetts Institute of Technology first described OCT in 199117 as a revolutionary non-invasive way to achieve microscopic-resolution in vivo cross-sectional imaging. The orderly organization of cellular components into layers in the retina and the direct relationship between anatomy and pathological processes led to the rapid adoption of OCT technology in ophthalmology and optometry.

Significant technological advances in OCT have translated rapidly into commercial systems over the past decade as the clinical demands and the utility of the devices have increased. Hardware-based improvements in speed, resolution and depth became possible with improved light sources and components. These have been translated from the lab to the marketplace as systems that incorporate spectrometers, referred to as spectral-domain or Fourier domain OCT,18 and even faster systems that cycle through different wavelengths of light, known as swept-source OCT.19

Advances in software have paralleled improvements in hardware. Commercial automated segmentation algorithms have been developed that identify the inner surface (internal limiting membrane) and outer surfaces (retinal pigment epithelium or Bruch’s membrane) of the retina to quantify the thickness and volume of the total retina. More advanced software has been developed to segment some retinal sub-layers and quantify pathological features such as drusen.20 Registration of retinal cross-sections and volumes between scan dates have further allowed for the qualitative comparison and quantitative analysis of patients over time to make treatment decisions and power clinical trials.

As OCT speed has increased, the opportunity for more interesting volumetric analysis has become possible, specifically en face OCT imaging21 and OCT angiography.22 Modern en face imaging uses a subset of the volumetric OCT data in conjunction with segmented anatomical contours to reveal pathological processes and anatomic relationships that are not obvious from cross-sectional OCT imaging alone and that allow for comparison to other imaging modalities.23

OCT angiography also utilizes en face OCT imaging, but leverages increased system speed and innovative software to correct bulk eye motion and to be sensitive to flow within the retinal vasculature.24,25  While OCT angiography is still in its early adoption phase, its ability to detect early choroidal neovascularization and areas of non-perfusion clearly make it a game changer in the retinal imaging landscape.26

In a normal eye imaged with conventional OCT, the obliquely oriented HFL has a reflectivity very similar to that of the underlying cellular machinery found in the outer nuclear layer (ONL). As the OCT beam moves to the side of the pupil and the light orientation becomes more perpendicular, the contralateral HFL becomes more hyper-reflective than the underlying ONL. At the same time, the ipsilateral HFL becomes relatively hypo-reflective, revealing a hyper-reflective synaptic outer plexiform layer above it and the ONL beneath.

The ONL-HFL Boundary

Owing to these changes in reflectivity, utilizing D-OCT allows the precise and reproducible identification of the cross-sectional thickness3 and the ONL three-dimensional volume.9 While the ONL-HFL boundary may sometimes be visible using frame-averaged images when one is seeking it out, it cannot be visualized on volumetric data without incorporating D-OCT and acquiring images through multiple pupil positions.  

Furthermore, the thicknesses of the ONL and the HFL are not correlated, prohibiting retrospective computation and discouraging lumping the two structures together as one. Finally, ONL measurements using D-OCT have been found to be repeatable and reproducible among different operators and graders, so that a change in the ONL thickness of more than 5 µm can be ascribed to ONL loss and not variability of the technique itself.10

Visualization of the directional reflectivity changes within a three-image D-OCT set can be achieved by generating a chromatic hybrid D-OCT image.11 This image is created by registering the “tilted” off-axis OCT images to the central pupil position “flat” image, normalizing the overall reflectivity, and assigning each individual image a separate color channel: red; green or blue.

If a structure is equally reflective from each pupil entry position, the colors cancel out and the pixels remain grayscale. However, if significantly more reflectivity occurs through the center position, such as the photoreceptor IZ band, then the resulting image will show that band as green. Similarly, blue and red show maximally reflective structures from one of the peripheral pupil positions, like HFL (Figure 1). Note that the subject’s fixation and scan area do not change; only the angle of incidence of the OCT beam on the retina.  

Figure 1. Directional optical coherence tomography (D-OCT) in a normal subject. A) Chromatic hybrid D-OCT showing directionally reflective structures in colors related to the pupil entry position that gives maximal reflectivity. Ellipsoid zone (EZ) and interdigitation zone (IZ) appear as green because the green channel is from light entering through the center of the pupil; the Henle fiber layer (HFL) is red and blue because its maximal reflectivity varies based on which side of the pupil light enters. Areas with equal reflectivity at each pupil entry position appear as grayscale as the colors from each individual scan cancels out. B) Standard central light entry position shows maximal IZ and EZ reflectivity (white arrowhead) and moderate HFL reflectivity (black arrowhead). C) Light entering from left pupil shows markedly decreased IZ reflectivity and moderately reduced EZ compared to B (white arrowhead) and maximal HFL reflectivity (black arrowhead). D) Light entering from the right pupil also shows decreased outer retinal band reflectivity (white arrowhead) and minimal HFL reflectivity (black arrowhead).   Figure 2. Directional optical coherence tomography (D-OCT) in a patient with central serous chorioretinopathy and subretinal fluid. A) Chromatic hybrid D-OCT shows directionally reflective structures in colors related to the pupil entry position that gives maximal reflectivity. Ellipsoid zone (EZ) and interdigitation zone (IZ) reflectivity follows the pattern of normal green appearance away from the area of subretinal fluid (white arrowheads). In the area of active subretinal fluid, the EZ and IZ are shown in a rainbow of colors indicating a change in alignment caused by the fluid. The colors that predominate are those that relate to the pupil entry positions toward which the photoreceptors are oriented. The variable reflectivity from the IZ and EZ are shown in B-D (white arrowheads). The Henle fiber layer (HFL) demonstrates the normal appearance in the temporal and nasal retina, but overlying the subretinal fluid, the geometry is altered in a way that the HFL appears flat and brightest in the green central channel.


D-OCT of Photoreceptors

Normal eyes demonstrate an orderly axial orientation of photoreceptors, which D-OCT visualizes by the maximal intensity of the hyper-reflective outer retinal bands when the OCT beam is through the center of the pupil and the cross-sectional image appears flat. A decrement in the amount of reflectivity of these photoreceptor bands may be due to the loss of photoreceptor inner and outer segments from macular disease or the alteration of the orientation of the cell processes, or both. The term photoreceptor “disruption” has been used indiscriminately throughout the literature to describe all of these situations.  

D-OCT may make it possible to distinguish photoreceptors that are truly lost from those that are just misaligned. Loss of photoreceptor inner and outer segments may result in no signal being generated from these bands at any imaging angle, whereas if only the orientation is altered, an OCT pupil-entry position may be found that preserves reflectivity. Change in orientation of the photoreceptor may be due to the effect of fluid in the retina (Figure 2), or a loss of contact between the outer segments and the underlying retinal pigment epithelial cells. Either of these situations may provide early diagnostic and prognostic information about disease progression over time.

Understanding the orientation of the HFL fibers can allow you to interpret its increased visualization overlying pathological features such as drusen, subretinal fluid or focal choroidal excavation. Similarly, loss of ONL may result in increased HFL visualization in cases of hydroxychloroquine toxicity.12 All of these are due to changes in the macular geometry to make HFL more visible. D-OCT can confirm these findings, and perhaps allow a more sensitive marker even when early loss of ONL has occurred.  

Because the photoreceptor nuclei found in the ONL have the potential to regenerate, ONL thinning is an unambiguous sign of disease progression. Indeed, ONL loss has been found histologically well away from the discreet areas of geographic atrophy on fundus autofluorescence currently used to follow AMD progression.13 Studies are under way to determine if the longitudinal assessments of ONL thickness and volume that D-OCT provides may be an early biomarker for AMD progression.14

Flat: Where It’s At

Thickness values derived from OCT segmentation analysis have become standard to evaluate the effectiveness for various therapies. Total retinal thickness, central foveal thickness and macular volume when used as clinical trial endpoints depend on the accuracy of automated algorithms in defining borders of the inner and outer neurosensory retina. While measurements of “tilted” scans have been found to induce error due to a geometric effect,15 a more subtle and significant measurement error exists that the principles of D-OCT can explain.

In studies that measure total retinal thickness using the RPE, an error arises from the directional sensitivity of the photoreceptor outer segments at the IZ. In a clinical dataset, more than 40 percent of scans were not “flat” and the segmentation algorithm was “pulled” lower in the scan toward the RPE.16 This effect is relatively small, although in a large normative database it adds to unnecessary noise that’s avoidable once it’s recognized.  Indeed, how the image is acquired will affect what we see and measure.

Limitations of D-OCT

Take-home Point

Optical coherence tomography (OCT)  has created a revolution in clinical practice and research by enabling a novel means of diagnosing, managing and prognosticating retinal disease. Directional OCT provides a means to further harness the optical properties of the retina to enhance and more precisely quantify and characterize macular pathology.
While D-OCT is a technique of acquisition and analysis that has been implemented on existing commercial systems, its full potential may only be realized with specialized hardware. Current limitations result from the user having to acquire multiple conventional scans.

When the user moves the machine to a new scanning position, even small changes in a patient’s fixation can cause imprecise registration between scans. While there is a wealth of information that can be ascertained without perfect registration, this motion ultimately limits the precision in quantifying the status of photoreceptors and detecting the smallest possible changes.  

Secondly, and perhaps most importantly for clinical adoption, acquiring multiple scans requires a longer overall scanning time. Hardware systems that accomplish eye tracking and rapid sequential steering of the OCT beam to different positions within the pupil are in development and would benefit from commercial adoption.  RS

References
1. Spaide RF, Koizumi H, Pozzoni MC. Enhanced depth imaging spectral-domain optical coherence tomography. Am J Ophthalmol. 2008;146:496-500.
2. Liu JJ, Witkin AJ, Adhi M, et al. Enhanced vitreous imaging in healthy eyes using swept source optical coherence tomography. PLoS One. 2014;9:e102950.
3. Lujan BJ, Roorda A, Croskrey JA, et al. Directional optical coherence tomography provides accurate outer nuclear layer and Henle fiber layer measurements. Retina. 2015;35:1511-1520.
4. Westheimer G. Directional sensitivity of the retina: 75 years of Stiles-Crawford effect. Proc Biol Sci. 2008;275:2777-2786.
5. Gao W, Cense B, Zhang Y, Jonnal RS, Miller DT. Measuring retinal contributions to the optical Stiles-Crawford effect with optical coherence tomography. Opt Express. 2008;16:6486-6501.
6. Lujan BJ, Roorda A, Knighton RW, Carroll J. Revealing Henle’s fiber layer using spectral domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2011;52:1486-1492.
7. Otani T1, Yamaguchi Y, Kishi S. Improved visualization of Henle fiber layer by changing the measurement beam angle on optical coherence tomography. Retina. 2011;31:497-501.
8. Ouyang Y, Walsh AC, Keane PA, Heussen FM, Pappuru RK, Sadda SR. Different phenotypes of the appearance of the outer plexiform layer on optical coherence tomography. Graefes Arch Clin Exp Ophthalmol. 2013;251:2311-2317.
9. Lujan BJ, Russakoff D, Oakley J, Garvin M, Roorda A. True outer nuclear layer volumes using directional optical coherence tomography. Invest Ophthalmol Vis Sci. 2014; 55:4803.
10. Tong K, Lujan BJ, Zhou Y, Lin MC.  Directional OCT reveals reliable outer nuclear layer measurements. Invest Optom Vis Sci. In Revision.
11. Makhijani VS, Roorda A, Bayabo JK, Tong KK, Rivera-Carpio CA, Lujan BJ. Chromatic visualization of reflectivity variance within hybridized directional OCT images. Proc. SPIE 8571. Optical Coherence Tomography and Coherence Domain Optical Methods in Biomedicine XVII, 857105 (March 20, 2013).
12. Chen E, Brown DM, Benz MS, et al. Spectral domain optical coherence tomography as an effective screening test for hydroxychloroquine retinopathy (the “flying saucer” sign). Clin Ophthalmol. 2010;4:1151-1158.
13. Bird AC, Phillips RL, Hageman GS. Geographic atrophy: a histopathological assessment. JAMA Ophthalmol. 2014;132:338-345.
14. Lujan BJ, Bhavna A, McDonald HR, et al. Optically isolating photoreceptors as AMD biomarkers using directional OCT. Invest Ophthalmol Vis Sci. 2015;56:5947.
15. Hariri A, Lee SY, Ruiz-Garcia H, Nittala MG, Heussen FM, Sadda SR. Effect of angle of incidence on macular thickness and volume measurements obtained by spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2012;53:5287-5291.
16. Antony BJ, Stetson PF, Abramoff MD, et al. Characterizing the impact of off-axis scan acquisition on the reproducibility of total retinal thickness measurements in SD-OCT volumes. Transl Vis Sci Technol. 2015;4:3.
17. Huang D, Swanson EA, Lin CP, et al. Optical coherence tomography. Science. 1991;254:1178-1181.
18. Wojtkowski M, Leitgeb R, Kowalczyk A, Bajraszewski T, Fercher AF. In vivo human retinal imaging by Fourier domain optical coherence tomography. J Biomed Opt. 2002;7:457–463.
19. Choma M, Sarunic M, Yang C, Izatt J. Sensitivity advantage of swept source and Fourier domain optical coherence tomography. Opt Express. 2003;11:2183-2189.
20. Gregori G, Wang F, Rosenfeld PJ, et al. Spectral domain optical coherence tomography imaging of drusen in nonexudative age-related macular degeneration. Ophthalmology. 2011;118:1373-1379.
21. Jiao S, Knighton R, Huang X, Gregori G, Puliafito C. Simultaneous acquisition of sectional and fundus ophthalmic images with spectral-domain optical coherence tomography. Opt Express. 2005;13:444-452.
22. Proskurin SG, He Y, Wang RK. Determination of flow velocity vector based on Doppler shift and spectrum broadening with optical coherence tomography. Opt Lett. 2003;28:1227-1229.
23. Lujan BJ, Rosenfeld PJ, Gregori G, et al. Spectral domain optical coherence tomographic imaging of geographic atrophy. Ophthalmic Surg Lasers Imag. 2009;40:96-101.
24. Jia Y, Tan O, Tokayer J, et al. Split-spectrum amplitude-decorrelation angiography with optical coherence tomography. Opt Express. 2012;20:4710-4725.
25. Kraus MF, Potsaid B, Mayer MA, et al. Motion correction in optical coherence tomography volumes on a per A-scan basis using orthogonal scan patterns. Biomed Opt Express. 2012;3:1182-1199.
26. Jia Y, Bailey ST, Hwang TS, et al. Quantitative optical coherence tomography angiography of vascular abnormalities in the living human eye. Proc Natl Acad Sci U S A. 2015;112:E2395-2402.