Major ReviewOptical coherence tomography: Imaging of the choroid and beyond
Introduction
The word choroid comes from the ancient Greek: korio-aydez, for korio (χoριo): a membrane around the fetus, and aydez (ειδησ): that looks like. In Latin this word meant network. Approximately 95% of the blood flow in the eye goes to the uvea, with the choroid accounting for more than 70%.162 The choroid has the highest blood flow per unit weight of any tissue in the body, about 20 to 30 times greater than that of the retina.2 One main function is to supply oxygen and metabolites to the outer retina, retinal pigment epithelium (RPE), and possibly the prelaminar portion of the optic nerve,79 and it is the only source of metabolic exchange for the avascular fovea. The photoreceptors inner segments, replete with mitochondria, have the highest rate of oxygen use per unit weight of any tissue.227 The choroid also acts as a heat sink, absorbs stray light, and, in birds, aids in accommodation.152
Many disease processes originate from the choroid or materially impact it because of its proximity to the affected retina or sclera (i.e., age-related macular degeneration, glaucoma, and diabetic retinopathy). Choroidal neovascularization (CNV) related to myopia, angioid streaks, multifocal choroiditis, and polypoidal choroidal vasculopathy (a variant of CNV) originates from the choroid. The high blood flow in the choroid predisposes the site to metastatic or embolic spread of tumors and infectious diseases. Some inflammatory disorders involving the posterior segment of the eye seem to target the choroid (e.g., Vogt-Koyanaghi-Harada disease, birdshot chorioretinopathy, and choroidal granulomas in sarcoidosis or tuberculosis). Finally, in animal models emmetropization and the first steps in control of refractive error are dependent on the choroid.152 Advances in imaging have greatly increased our ability to visualize the choroid, allowing better understanding of the choroid in health and disease.
The optic vesicles form as outpouchings of the forebrain and then invaginate and form a double-walled optic cup. The cup has two distinct layers: one eventually destined to form the retina and the other the RPE. The inferior portion has two edges where the cup joins, forming the choroidal fissure. This potential gap allows the hyaloid artery to enter the eye. The uvea develops from the mesoderm and migrating neuroectoderm that surround the optic cup. The choroid is the posterior part of the uvea, the middle tunic of the eye. The mesodermal cells surrounding the cup start to differentiate into vessels concomitantly with the RPE. The choriocapillaris starts to form at about the fifth to sixth week of embryogenesis. The basal lamina of the RPE and of the choriocapillaris define the boundaries of the nascent Bruch's membrane by week 6.194 The choriocapillaris becomes organized with luminal networks well before formation of the rest of the choroid. The posterior ciliary arteries enter the choroid during the eighth week of gestation, but it takes until week 22 before arteries (which have a continuous layer of smooth muscle cells) and veins become mature. Melanocytes precursors migrate into the uveal primordia from the neural crest at the end of the first month and start differentiating in the seventh month. The pigmentation of the choroid begins at the optic nerve and extends anteriorly to the ora serrata and is completed by about 9 months.145 Thus the choroid derives from different cell lines than the retina and the RPE, both derived exclusively from neural ectoderm. The sclera is derived from mesenchymal condensation starting anteriorly and completed posteriorly by week 12.
The sclera is thickest (1.0 mm) at the posterior pole around the optic nerve and thinnest (0.3 mm) beneath the insertions of the rectus muscles, measuring 0.5 mm at the equator. The sclera is organized in three layers: outer episclera, scleral stroma, and inner lamina fusca. The superficial episclera is a connective tissue that contains melanocytes, a few fibroblasts, lymphocytes, and relatively numerous blood vessels. The scleral stroma is the largest portion of the sclera and is composed of branching bundles of collagen fibers, together with numerous elastic fibers, extracellular matrix, and a few fibroblasts. The lamina fusca, the innermost layer, can be distinguished by its brownish color, due to the presence of numerous embedded melanocytes, and blends with the suprachoroidal and supraciliary lamellae of the uveal tract. Collagen in the lamina fusca is arranged in thin, small bundles. This layer also contains elastic fibers. The sclera, like the cornea, is essentially an avascular fibrous structure, except for the vessels of the superficial episcleral plexus169 and the intrascleral vascular plexus located just posterior to the limbus. A number of channels, or emissaria, penetrate and break the continuity of the sclera for the perforating ciliary nerves, short and long posterior ciliary arteries, anterior ciliary arteries, and vortex veins. Interesting variations of the sclera are associated with myopia, particularly high myopia. Eyes with high myopia have thinning of the sclera, with a decrease in the amount of collagen in the sclera. The distribution of collagen fibrils is different from that of emmetropic subjects, with an increased proportion of smaller diameter collagen fibrils,32, 134, 175 and the sclera is more elastic.135
The ophthalmic artery, the first branch of the internal carotid artery, branches to form the central retinal artery and the posterior ciliary arteries (PCAs). There are many variations, but the posterior choroid is supplied by two major PCAs, the medial and lateral PCAs, in approximately 90% of eyes.78 From their origin, the PCAs divide into a large number of branches as they course toward the eye. Two major branches, called the long PCAs, eventually supply the anterior uvea. Smaller short PCAs, usually about 20 in number, enter the eye, particularly around the optic nerve and macular regions.65
The choroid is primarily composed of blood vessels, but also has connective tissue, pigment, and intrinsic choroidal neurons. Birds have fluid-filled lacunae identified as a true functional lymphatic system in their choroid.140 Schroedl et al found that humans don't have typical choroidal lymphatic vessels, but have macrophage-like cells that stain positively for a lymphatic endothelium-specific marker (lymphatic vessel endothelial hyaluronic acid receptor).193 Humans also have cells with non-vascular smooth muscle-like elements in the choroid.133, 170 Their predominant subfoveal localization suggested that they may play a role in stabilizing the fovea against movement caused by the contracting ciliary muscle during accommodation.42 The human choroid has intrinsic choroidal neurons, which have been theorized to participate in autoregulation of blood flow.36 It is possible the presence of lymphatic and intrinsic neuronal cells may be atavistic. The choroid is attached to the sclera by strands of connective tissue that are easily separated anteriorly creating a potential space between them, the suprachoroidal space, and is tightly adherent to the optic nerve.
The blood from the short PCAs enters the eye and travels through successively smaller arterioles within the choroid to arrive at the choriocapillaris. The choroid is traditionally described as arranged in layers of vessels from the outer to inner part of the choroid the Haller layer, the Sattler layer, and the choriocapillaris. The Haller layer contains larger choroidal vessels, and the Sattler layer has medium-sized vessels. There is no distinct border between these layers or even an established definition of what is meant by large or medium. Blood pressure is reduced from about 75% of systemic at the short PCAs to that in the choriocapillaris, measured in rabbits as approximately 5 to 9.5 mm Hg greater than the intraocular pressure.121 The choriocapillaris is a planar layer of small vessels with a lumen slightly larger than a typical capillary. The network of vessels in the choriocapillaris is so tightly arranged in the posterior portion of the eye that specific capillary tubes are more difficult to discern. Given the packing density of the choriocapillaris, along with the larger lumen size of individual tubules, the summation of the cross-sectional area of the choriocapillaris, lumens appears to be quite large. In the peripheral portions of the eye anatomic arrangements representing lobules are suggested, but there is no specific anatomic lobular structure in the posterior portion of the eye. There is a controversy regarding the choroidal angioarchitecture. Anatomically, the lobular appearance of the choriocapillaris exists only in part of the posterior pole, as has been demonstrated by Fryczkowski using vascular casts and scanning electron microscopy.51 Fryczkowski also showed that the choroidal anatomical lobules are not identical with those observed by fluorescein angiography. He concluded that two models of choroidal lobules should be distinguished: anatomical and functional. The anatomical lobule contains a central collecting venule and peripherally located feeding arterioles. The functional lobule is centered by a main feeding arteriole and has peripherally located draining venules. Using fluorescein angiography, the functional lobuli are seen as centrally filled areas even in areas that are anatomically non-homogeneous structures or non-lobular structures in the posterior pole.51
The vitality of the choriocapillaris is maintained in part by constitutive secretion of vascular endothelial growth factor (VEGF) by the RPE.189 The choriocapillaris is highly polarized,10 with the internal surface having localized attenuations of the capillary wall known as fenestrations. These fenestrations appear to increase the amount of material leaving the capillaries and direct the flow toward the RPE. Similarly sized vessels in the retina do not have fenestrations. Soluble VEGF isoforms are required for fenestrations to occur,189 and these fenestrations disappear with VEGF withdrawal.163
Venous drainage of the choriocapillaris is mainly through the vortex vein system with minor drainage through the ciliary body via the anterior ciliary veins. Postcapillary venules form afferent veins that converge to form the ampullae, which in turn empty through the vortex veins. There are often four vortex veins per eye, but the number may range from 3 to 8.187 Occasionally highly myopic eyes may have a vortex vein, called a ciliovaginal vein, located in the posterior pole that drains near or through the optic nerve. The typical vortex vein passes obliquely through the sclera for a distance of about 4.5 mm as it exits the eye and drains into the superior and inferior ophthalmic veins.73
Hayreh observed special features of choroidal vasculature in humans and monkeys.76 The choroidal arteries do not anatomose with one another, and each behaves like an end-artery. Hayreh quotes Duke-Elder (page 273) “The tendency for inflammatory and degenerative diseases of the choroid to show a considerable degree of selective localization, despite the fact that anatomically the vessels would appear to form a continuous meshwork, has given rise to speculations regarding the anatomical isolation of specific choroidal areas.”76 The flow within the choriocapillaris is a function of differential pressure gradients, which has the effect of allowing flow into regions that may have suffered a decrease in supply from a problem in a feeding choroidal arteriole.
In the early phase of fluorescein angiography, it is common to see areas of the choroid that do not appear to fill with dye as quickly as adjacent regions, called watershed zones by Hayreh. A common watershed zone appears as a stripe, one to several millimeters wide, running vertically at the temporal border of the optic disc. This watershed zone is thought to be the boundary between areas of the choroid supplied by the medial and lateral PCAs and occurs in 60% of eyes.76 The arteries and veins supplying each region of the choroid do not have a parallel course and the segment supplied by an artery does not correspond exactly with that drained by a vein. There is always a certain amount of overlap between the adjacent arterial segments via the veins.77 The peripapillary choroid has an important role in the blood supply of the anterior optic nerve, including the optic disc79 and has a segmental blood supply.
Almost every tissue in the body has some form of autoregulation, but the extent of the autoregulation of the choroid is controversial. Some investigators have shown the choroid has no autoregulation when the perfusion pressure gradient is decreased by raising the intraocular pressure (IOP).2, 50 Others have shown the choroidal blood flow varies with IOP, perfusion pressure,102 endogenous nitric oxide production,167 and vasoactive secretory production of choroidal ganglion cells.119 Various studies have suggested the choroid has some autoregulatory capacity during changes in ocular perfusion pressure.168, 182, 183 Moreover, Polska et al found that the mechanisms regulating choroidal blood flow in the human fovea compensate better for an increase in arterial blood pressure than for an increase in intraocular pressure.168 Autoregulation in regular tissues usually keeps the oxygen partial pressure at a relatively low, physiologic level. The inner choroid's oxygen partial pressure is so high that it is likely that the choroidal circulation is regulated by additional factors. For example, CD-36 is a scavenger receptor expressed in the basal RPE, and Houssier et al showed that CD-36–deficient mice fail to induce COX-2 and subsequent VEGF synthesis at the level of the RPE and develop progressive degeneration of the choriocapillaris.89 Therefore, CD-36 binding in the RPE seems to be one of the factors maintaining the inner choroid.
One explanation proposed for what may appear to be incomplete autoregulation is that the choroidal blood flow is much higher than in other tissues, and there is a low oxygen extraction ratio. Autoregulation is an adaptive compensatory mechanism to adjust blood flow according to the local needs of the tissue supplied. The high oxygen partial pressure and low oxygen extraction imply that the flow in the choroid is maintained at a level much greater than the local needs of the choroid or RPE seem to dictate. On the other hand the O2 delivered to the outer retina is all consumed by the mitochondria in the inner segments of the photoreceptors, and it seems unlikely that a direct feedback mechanism exists between the oxygen utilization by the inner segments of the photoreceptors and the choriocapillaris because diseases that cause acute destruction of the outer retina such as acute zonal occult outer retinopathy are not associated with decreased thickness of the choroid.56 Indirect trophic mechanisms have not yet been elucidated.
There have been other reasons proposed for the large blood flow in the choroid. The high metabolism of the outer retina generates heat, and the choroidal blood flow may act as a heat sink. The amount of light energy delivered to the retina by incoming light is insufficient to cause a significant elevation in temperature and thus is an unlikely explanation.68 The high metabolism in the outer retina may produce enough heat to require mechanisms to reduce the local temperature. Even this hypothesis is suspect because vasodilation occurs in response to increased temperature, which has not been demonstrated in normal eyes. The choroid contains melanocytes that improve optical function by absorbing scattered light, and may also indirectly protect against oxidative stress. These melanocytes exist in an environment of high O2 partial pressure that, along with the light exposure, may be a risk factor for malignant transformation to melanoma.
Because of its localization between the overlying pigmented RPE and the underlying opaque and rigid fibrous sclera, the choroid is difficult to visualize. Methods using light reflection or fluorescence generation are impeded by the pigment in the RPE and choroid. Conventional optical coherence tomography (OCT) is affected by the effects of melanin and also the scattering properties of the blood and blood vessels.
Fluorescein is stimulated by blue light with a wavelength between 465 and 490 nm and emits a green light with a peak emission between 520 and 530 nm and a curve extending to approximately 600 nm. Both the excitation and emission spectra from fluorescein are blocked in part by melanin pigment, which acts to decrease visualization of the choroid. Fluorescein extravasates rapidly from the choriocapillaris and fluoresces in the extravascular space, and this also prevents delineation of the choroidal anatomy. The analysis of the choroid using fluorescein angiography is also limited by light absorption and scattering by the pigment in the RPE and choroid and by the blood in the choroid. Indocyanine green (ICG) absorption peak is between 790 and 805 nm and fluoresces over a somewhat longer wavelength range, depending on the protein content and pH of the local environment. Longer wavelengths have the attribute of penetrating the pigmentation of the eye better than those used with fluorescein angiography. ICG is 98% protein bound (with 80% binding to larger proteins such as globulins and alpha-1-lipoprotein6, 24, 41) and thus is less likely to leak from the normal choroidal vessels. During the earlier phases of ICG angiography, the choroidal vessels are fairly easy to discern, but vertical summation makes it difficult to delineate individual layers. ICG angiography allows a precise anatomic and dynamic evaluation of the choroidal circulation: the arteries, arterioles, venules, and veins, not only in the posterior pole, but out to the periphery and the vortex veins. The normal or abnormal filling of these choroidal vessels can be appreciated. Over the course of the angiogram some staining of the extravascular tissue, particularly Bruch's membrane, occurs, obscuring visualization of deeper structures later in the angiographic sequence.
Contact B-scan ultrasonography typically uses a 10-megahertz probe placed on the eyelid. The probe used contains a piezoelectric crystal that is stimulated to vibrate by a short burst of an electrical current. When the vibration stops, the piezoelectric crystal is used as a detector. Reflected sound waves cause the crystal to vibrate, which in turn causes the production of an electrical current. The reflected sound wave varies in strength with the position of the reflecting structure in the eye. Deeper structures produce weaker reflections. The decrease in signal strength with time of flight is corrected by increasing the gain of the amplifier during this interval. The direction of the crystal is moved slightly by a motor within the probe. Many axial scans, or A-scans, are summed to produce a two-dimensional image known as a B-scan. The methods using sound reflection have poor resolution. The axial resolution of A-scan or B-scan ultrasonography is theoretically about 150 μm. In reality the sound beam produced by a piezoelectric crystal in a conventional B-scan probe has a main lobe and several side lobes.80 Even the main lobe can be 1 mm in diameter at the surface of the retina. That means the image produced by an individual A-scan is a summation of a wide area of the back of the eye. For example, a typical B-scan of the optic nerve will not show the cup unless there is a large cup-to-disk ratio. Another problem with ultrasonography is that the exact location of the image obtained is not known. The general region can be estimated by evaluating relationships with neighboring structures.
In non-pathologic conditions the reflectivity of the choroid is difficult to distinguish from the overlying retina and the underlying sclera. This ambiguity raises questions as to what, exactly, is being measured when tumor thicknesses are evaluated by contact B-scan ultrasonography. In diseases that cause a thickening of the choroid the reflectivity may decrease, affording increased contrast between the choroid and the retina and sclera. When this thickening decreases, contrast also decreases. Given the low resolution of contact B-scan ultrasonography, changes in choroidal thickness cannot be measured accurately. Contact B-scan ultrasonography is good for visualizing larger tumors and gross changes in ocular curvature such as that seen in pathologic myopia.
Laser Doppler flowmetry, developed by Riva and associates in 1994, is a noninvasive technique for the investigation of local choroidal blood flow and its regulation in the fovea.181 For choroidal blood flow evaluation by laser Doppler flowmetry, the foveal region is chosen as the measuring site because it is free of retinal vessels, and the patient fixates directly on the laser light. The resultant Doppler signal arises predominantly from the choriocapillaris,181 by contributions from the vertical summation of signals generated by other choroidal vessels. This technique evaluates only a limited area of the posterior pole that may or may not be representative of the flow in other regions of the choroid. Future development of Doppler and phase variance OCT techniques offer opportunities to evaluate the flow within various levels in different regions of the choroid and are likely to supplant laser Doppler flowmetry.
In dry air the speed of sound is 343.2 meters per second. The sound velocity in an average phakic eye is 1,555 meters per second.85 The time it takes sound to travel the length of a phakic eye with an axial length of 24 mm is approximately 15.4 microseconds, an interval easy to measure. Because light travels at 3 × 108 m/sec, it is not possible to determine the time-of-flight delay on a micron-scale level of resolution using an external system of time measurement. The time it takes light to travel several microns is the same time it would take an electron to travel in a circuit. Measurement of multiple reflections with a detector and the subsequent required electronic circuitry requires conduction paths for electrons that are much longer than the variations in path length of the photons. An ingenious way to determine how long light takes to travel a given distance is to use the wave-like character of the light itself as its own internal clock. That is what a Michaelson interferometer does; the wavelength of the light is used as its own timing standard. The micron-scale resolution is achieved by comparing the time of flight of the sample reflection with the known delay of a reference reflection by looking for phase differences in the light interference.
Coherency of light is a measure of how correlated one wave of light is with another. Temporal coherency is a measure of the correlation one wave has with another generated at a different time. Coherence length is the distance light would travel during the coherence time. Light produced by a conventional laser has a long coherence time because the waves of light are similar. Thus the coherence length of a typical laser can be long—meters or more. An interferometer could be constructed with a conventional laser in which light reflected from a distant sample is compared with light from a much shorter reference arm. The differences between the path length of the sample and reference arms would cause gross differences in the number of wave lengths of light, but the interference is related more to the fractional differences in phase of the waves and not the gross differences in whole number of wavelengths. This approach allows us to measure small changes in the sample arm path length with great precision, but the same instrument can't differentiate that path length from one multiple wavelengths longer or shorter with the same fractional phase difference. One example of how this type of interferometer could be used is a “James Bond” device that aims a laser light beam at some structure in a room, such as a picture or even a window, from a great distance away. Anyone speaking in the room will cause these objects to vibrate. The small path length changes induced by the vibrations can be detected by comparing the reflected light (i.e., the sample arm) to a reference arm. The resultant interference signal can be used to obtain the original sound information, no matter if the reflecting structure is tens or hundreds of meters away.
It is possible to produce light with a short coherence length. In this situation the waveform of light produced is the same for all of the light rays produced at any one instance, but this waveform is different from other waveforms produced at other times. This approach essentially puts a time stamp on the waveform. Low coherence light split into a reference arm can only interfere with light from the sample arm if the path lengths are the same or are nearly the same. If the reference arm is varied in length the reflectivity of various points in the sample arm can be determined. Only light from the small area, as defined by the coherence length, produces signal. The smaller the coherence length the smaller the area sampled. The smaller the area that is sampled, however, the less chance light rays will be reflected back to the interferometer. The more light that is sampled by the interferometer the more certain we can be about the true reflectivity of the sample. To get enough light to form a good image, the tissue has to be sampled for a finite time. This concept forms the basis, and illustrates the limitations, of time-domain OCT. The advantages of time-domain instruments are that they are conceptually simple and relatively easy to manufacture. Their chief disadvantage is that the illuminating beam goes through the full depth of the tissue even though only one point is sampled at a time. The total light exposure to the eye is restricted by safety standards, so there is a limit the signal to noise ratio of the information obtained by time domain OCT.
In high school textbooks interference is shown as an all or nothing event, either light rays constructively or destructively interfere. In reality the result of interference of coherent light is a varying brightness depending on the phase relationship of the two beams. If the two light beams have a slightly different frequency a new interference or beat frequency is detectable. Using short coherence length light interference from various depths of tissue produces a range of frequencies depending on the depth of reflection. If the interference pattern is projected through a grating, the various frequency components will be dispersed. The closer the match between the two path lengths is, the lower the frequency. Deeper structures will have a greater path length mismatch and consequently will produce an interference signal composed of higher frequencies. These various signals and frequencies can be detected simultaneously. Using a Fourier transform, it is possible to determine where, and how strongly, different reflections in the sample arm are. In effect reflections from various depths are frequency encoded. The decoding process produces useful information from all levels simultaneously. This forms the basis for spectral-domain (SD)-OCT. Because the illuminating beam is used efficiently, as all layers produce a signal during the scan time, it is possible to capture information faster for any given A-scan than with time-domain OCT. SD-OCT devices typically scan with speeds up to 100 times faster than time-domain OCTs.
The two types of Fourier domain detection are SD-OCT and swept-source domain OCT (SS OCT). These use a different light source and detection method, but not necessarily a different wavelength.54 The SD-OCT approach uses an interferometer with a low coherence light source and measures the interference spectrum using a spectrometer and a high-speed line charge coupled device (CCD). The SS-OCT uses a frequency-swept light source and detectors that measure the interference output as a function of time.26, 27 There are some problems inherent in SD technology. The deeper tissues produce higher frequency signals, but the way the grating and detector sample this frequency is not linear. The higher frequencies are bunched together to a greater extent than lower frequencies. In addition, the sensitivity of the detection decreases with increasing frequency. This causes SD-OCT to have decreasing sensitivity and resolution with increasing depth.
The peak sensitivity of SD-OCT is where the reference and sample arms are the same length, called the zero-delay point. The further in the sample arm from the zero delay point a reflection originates, the weaker the resultant detected signal, even if the reflection does not vary in intensity because the detector decreases in sensitivity (Fig. 1). As a consequence of the roll-off in sensitivity, images from deeper in the eye are increasingly dark. Conventionally the zero delay point is placed at the level of the posterior vitreous. This is a logical place, because visualizing the vitreoretinal interface is important in evaluating many diseases. The vitreous is nearly transparent, and a high sensitivity is needed to produce useful images. The trade-off is that the choroid cannot be seen very well.
There are several techniques to better visualize the choroid: enhanced depth imaging OCT, image averaging, SS-OCT and using a longer wavelength.
A consequence of using a Fourier transform to decode the interferometric signal is that two conjugate images are developed. In practical use only one of these two images is shown, typically with the retina facing toward the top of the screen. If the peak sensitivity is placed posteriorly, typically at the inner sclera, deeper structures such as the choroid can be seen (Fig. 2). The upside down conjugate image of these structures is visualized, and the right-side-up image of structures in the orbit is blank because of the lack of any imaging information. This method of imaging the choroid is called enhanced depth imaging (EDI) OCT.207 It is now simply performed with SD-OCT instruments, for example, by selecting the “EDI” button when using the Spectralis (Heidelberg Engineering, Heidelberg, Germany). This feature is now available in the software of different OCT devices: the 3D OCT-2000 (Topcon Inc, Tokyo, Japan), Cirrus with high definition (HD) OCT machine (Carl Zeiss Meditec Inc, Dublin, CA), and the RTVue (Optovue Inc, Fremont, CA). These devices can perform the equivalent of EDI-OCT without requiring the user to invert the image. The new software version 6.0 for the Cirrus with HD OCT machine has the ability to capture images using EDI technique.
With the peak sensitivity deeper in the eye, the roll-off of sensitivity occurs in the vitreous, which as a consequence is not visualized well. With SD-OCT there is then a choice between two modes of imaging, a conventional scan in which the vitreous is visualized and the EDI mode used to see the choroid. For most eyes, both cannot be seen optimally at the same time. The increase in sensitivity using the EDI mode allows penetration of an additional 500–800 μm deeper into the eye depending on the amount of pigmentation and other factors. As will be discussed later, high myopes have thin choroids with little pigmentation, and the sclera is usually thin as well. As a consequence it is possible to not only see the full thickness of the sclera, but also visualize the orbital fat.
Spectral domain devices use a grating that disperses light onto the CCD. Video cameras commonly use CCDs that are a two-dimensional array of light sensitive elements. To detect the interference fringe from SD-OCT, only a single row of light-sensitive elements is required. These are called line CCDs. Each light-sensitive spot, sometimes called a well or a bucket, gathers an electron for each photon of light. The OCT scan line has to remain “parked” over the sampled tissue until the wells in the line CCD fill. After a period of detecting light the wells are dumped out and the electrical charge from each one is evaluated. The scan speed is restricted in part by the sensitivity of the CCD, the strength of the light source, the quantum efficiency of the detector, and the speed at which all of the individual elements of the CCD can be read. The line CCD has to be read at fixed intervals, which limits the speed of the OCT device.
To improve the signal-to-noise ratio and therefore image appearance, many B-scans can be averaged together, typically 50 to 100. With the Spectralis this can be accomplished using eye tracking, so the same location is sampled from one scan to the next, but averaging many B-scans is possible without eye tracking. For example the 3D OCT-2000 uses an alternate approach by analyzing the degree of matching between many images, eliminating the ones that don't match, and averaging only the ones that do.
A third method of generating OCT images uses a laser that sweeps across a range of wavelengths in an orderly fashion. The interference of the light from the sample and reference arms produces a signal that can be read out in nearly real-time by a photodiode. The SS-OCT uses a light source that is inherently more complicated than what is used by SD-OCT. The advantage is gained at the detection end of the instrument. The detector is generally simpler in design and capable of operating at higher speeds. There are other desirable features that can be exploited. The falloff in sensitivity for swept source OCT is much lower than for SD-OCT. In addition the wavelengths available for swept laser sources are generally somewhat longer, in the 1 micron region. This longer wavelength is capable of penetrating tissue to a greater extent than the relatively shorter wavelengths typically used for SD-OCT. Thus both the vitreous and choroid can be imaged simultaneously; there is no need to pick one or the other.
There are trade-offs with swept source OCT. Although longer wavelengths of light may penetrate tissue better, particularly tissue containing melanin, the problem is that water absorbs longer wavelengths of light. This restricts the range, or bandwidth, of wavelengths that can be used in the eye, because the vitreous is mostly water. The greater the bandwidth of the light source, the better the resolution of the OCT image. If the bandwidth of swept source lasers centered at, for example, 1050 nm is large, the longer wavelengths will be attenuated by water absorption. This creates an asymmetrical profile of the spectrum used to illuminate the retina and deeper structures, effectively reducing the actual usable bandwidth of the light source. The light used to illuminate the retina cannot be increased to compensate, because the amount of light delivered to the eye is limited by standards that quantify the amount of light delivered to the cornea. The second disadvantage is the resolution of the OCT image is related to the square of the center wavelength of the light. Any increase in the wavelength has a detrimental effect on resolution for any given bandwidth. Because there is a limit to the available bandwidth of light that can be used because of water absorption, there is an upper limit to the resolution of SS-OCT instruments operating in the 1 micron region. Given the relatively good performance of SD-OCT in obtaining choroidal images, there may be less need to develop commercial devices in the 1 micron realm. Newer light sources operating at shorter wavelengths are being explored, and these may avoid the problem of water absorption. An SS-OCT using a large-bandwidth light source with a center wavelength of 850 nm could provide high-speed, high-resolution imaging with little falloff in sensitivity with depth.
OCT pictures are generally shown as 2D B-scan images acquired through the thickness of the retina. It is simple to obtain numerous adjacent B-scan images with any typical SD-OCT. After the successive images are aligned, a 3D volume can be reconstructed. Noise reduction through averaging can be done for each component of B-scan prior to creating the 3D volume. Noise reduction after the 3D volume is assembled can be completed with a variety of image processing techniques. Many SD-OCT instruments rely on their inherently fast scan rate to assemble a series of B-scans with the assumption that the patient did not make any saccades during the scan interval. Another approach is to align B-scans after acquisition, but this can result in image warping and local artifacts where the images in successive B-scans don't fit together well. A third approach is to use eye tracking not only to average individual B-scans, but also to maintain centration of the volume scan. An example of 3D volume rendering of the choroid is shown in Fig. 3. Although 3D rendering has been available in radiology for decades, there has been remarkably little penetration into ophthalmology. A B-scan can be informative, but in many cases can be compared to one page of a book. Volume rendering allows viewing and measurement of the choroid and any contained pathology in 3D from any arbitrary direction and would also be a useful tool to assess the three dimensions of any intrachoroidal process, such as a choroidal tumor or granuloma, and its relationships to adjacent structures.
Section snippets
Measurements and reproducibility of choroidal thickness
In using SD-OCT, whether it is Spectralis using EDI module, Cirrus with HD, RTVue, or SS-OCT, the outer border of the RPE and the inner border of the sclera usually are determined manually by the observer. Digital calipers are placed at the outer edge of the hyperreflective RPE line and the inner border of the hyperreflective surface located behind the large choroidal vessels, which is the scleral/choroidal interface. These two points are chosen so that the line traced between them is
Choroidal thickness variations in normal eyes
Brown and associates demonstrated diurnal fluctuations of CT measured using a non-contact optical biometer Lenstar LS 900 (Haag-Streit AG, Köniz, Switzerland).19 Determining CT was possible only in 28% of the patients with a single measurement. By repeating measurements and averaging the partial coherence interferometry signals, a peak of reflectivity corresponding to the scleral choroidal interface was estimated in another 70%.19 The authors demonstrated significant diurnal fluctuations of CT.
Description of a new entity
Spaide described a condition in non-myopes he named age-related choroidal atrophy (ARCA). Although CT appears to decrease with age, there are some patients who seem to have a pronounced loss of CT over time (Fig. 7).203 Despite relatively preserved Snellen visual acuity, these patients have visual complaints frequently involving reading and exhibit characteristic fundus changes quite similar to those seen in myopes. The larger choroidal vessels are easily discerned, many of the visible
Choroidal hyperpermeability in CSC
Since its first description by von Graefe in 1866, the understanding of central serous chorioretinopathy (CSC) pathophysiology has evolved. In the 1960s, Maumenee used fluorescein angioscopy to visualize the leaks from the level of the RPE. In 1967, Gass proposed that choriocapillaris hyperpermeability was the source of the fluid leak.58 Later in the 1980s, Marmor performed experiments on healthy rabbits and found that the RPE pumped fluid and that there was a normal flow of fluid from the
AMD
At the time when the genetic basis of AMD is being unraveled, some important questions concerning its pathogenesis remain unsolved. One of these questions is how the choroid modulates the expression of AMD. Geographic atrophy and age-related choroidal atrophy are both related to aging. Even though they can occur concomitantly, they are not necessarily associated.203 Histologic evidence regarding CT in AMD is mixed, with some studies suggesting a reduction in the choriocapillaris diameter and
Choroidal thickness in posterior uveitis
Many inflammatory diseases in the posterior segment of the eye are known to originate from, or at least involve, the choroid. The monitoring of CT may prove to be an accurate way to evaluate inflammatory processes and potentially to monitor the response to treatment in a non-invasive manner.
Choroid in inherited retinal dystrophies
Inherited retinal dystrophies comprise a large number of disorders, some better characterized than others. Phenotypically disparate conditions may share the same underlying genetic defect. In addition patients with the same genetic mutation may have large variations in disease severity. Whereas some disorders, such as dominantly inherited retinitis pigmentosa (RP), involve rhodopsin mutations, implying the potential for rod damage, many of these eyes will go on to also show cone, RPE, and, in
Chorioretinal atrophy in high myopia
In many developed countries, a major cause of vision loss is high myopia, which is often defined as 6 or more diopters spherical equivalent myopic correction. High myopia is associated with excessive and progressive elongation of the globe and results in a variety of fundus changes that lead to visual impairment, including lacquer cracks in Bruch's membrane, CNV, and chorioretinal atrophy.33, 82, 88, 103, 104, 113, 142, 157 Myopic chorioretinal atrophy is a common and serious problem, and yet
Diseases involving the sclera
With EDI-OCT it is possible to visualize the full thickness of the sclera in high myopes. The choroid is thin and often has sparse pigmentation, and the sclera is likewise thin. With SS-OCT the sclera can be visualized to a greater extent, and in some lightly pigmented eyes (albeit with low myopia) the outer boundary of the sclera can sometimes be appreciated. There are a number of potential interactions between the choroid and sclera that may influence ocular pathology and with improved
General information
OCT features of choroidal tumors have been extensively documented but are limited mostly to their effects on the overlying retina. Using conventional OCT, the choroidal findings in choroidal nevi are limited to the anterior surface.195 EDI-OCT of the choroid allows characterization of the thickness and reflective quality of small (less than 0.9 mm thick) non-pigmented choroidal lesions. We can roughly estimate the thickness of pigmented choroidal tumors in an indirect manner, by determining if
Choroid in glaucoma
Glaucoma is the world's leading cause of irreversible blindness.180 Theories of glaucomatous damage to ganglion cell axons usually focus on a mechanical etiology at the level of the lamina cribrosa, impairment of blood flow to the optic nerve and peripapillary area, or both. Some studies report an impairment of blood supply prior to demonstrable glaucomatous visual field defects, suggesting that the neuronal damage does not cause the impaired optic nerve blood flow.165 EDI-OCT in glaucoma
Optic nerve head drusen: enhanced visualization and analysis of adjacent optic nerve fibers
Optic nerve head drusen are acellular deposits, primarily composed of proteins, mucopolysaccharides, and calcium, that characteristically occur bilaterally in small, crowded optic nerve heads.114 They have been clinically observed in 0.3% of the population, although autopsy studies suggest an incidence of 2%.48 This discrepancy suggests clinical underestimation of the real incidence of optic nerve head drusen. Buried optic nerve head drusen, especially small ones, cannot be easily visualized by
Choroid in diabetes
Many histological studies have identified consistent vascular changes in the choroid of the diabetic patient52, 53, 81 that are quite similar to those seen in diabetic retinopathy: increased vascular tortuosity, vascular outpoutchings, microaneurysm formation, areas of nonperfusion, dilations and narrowing of vascular lumens, and choroidal neovascularization. The vessels involved are mainly the choriocapillaris, but also the larger choroidal vessels.52, 81 Early vascular changes involved the
Conclusion
The choroid is the main blood supply to the eye and should be considered in the pathogenesis of eye diseases. There are now methods to visualize the choroid: EDI-OCT and SS-OCT techniques allow visualization in nearly every patient. The CT in normal eyes can be measured and used as comparison to that found in various disease states. CT is negatively correlated with age, varies during the day, and possibly fluctuates with defocus over the course of a few minutes. The choroid is an extremely
Method of literature search
The authors conducted a search of Medline with PubMed. Even though the development of OCT methods to image the choroid is relatively recent since 2008, the search for choroidal anatomy, physiology, and histopathology goes back earlier, from April 1962. We stopped the search in September 2012. Search words were choroidal thickness, enhanced depth imaging optical coherence tomography, choroidal optical coherence tomography, swept-source optical coherence tomography, choroidal blood flow,
Disclosure
Dr Spaide is a consultant to Topcon.
The authors have no other financial or personal relationships with other people or organizations that could potentially and inappropriately influence their work.
References (240)
- et al.
Ocular and optic nerve blood flow at normal and increased intraocular pressures in monkeys (Macaca irus): a study with radioactively labelled microspheres including flow determinations in brain and some other tissues
Exp Eye Res
(1973) Posterior scleritis
Surv Ophthalmol
(1988)- et al.
An international classification and grading system for age-related maculopathy and age-related macular degeneration. The International ARM Epidemiological Study Group
Surv Ophthalmol
(1995) - et al.
Reproducibility of choroidal thickness measurements across three spectral domain optical coherence tomography systems
Ophthalmology
(2012) - et al.
The optic nerve head as a biomechanical structure: a new paradigm for understanding the role of IOP-related stress and strain in the pathophysiology of glaucomatous optic nerve head damage
Prog Retin Eye Res
(2005) - et al.
Choroidal thickness in polypoidal choroidal vasculopathy and exudative age-related macular degeneration
Ophthalmology
(2011) - et al.
Re-evaluation of “reactive lymphoid hyperplasia of the uvea”: an immunohistochemical and molecular analysis of 10 cases
Ophthalmology
(2000) - et al.
HLA associations and ancestry in Vogt-Koyanagi-Harada disease and sympathetic ophthalmia
Ophthalmology
(1990) - et al.
Indocyanine green angiography in birdshot chorioretinopathy
Ophthalmology
(1999) - et al.
Age-related macular changes in humans over 90 years old
Am J Ophthalmol
(1990)
Enhanced depth imaging optical coherence tomography of the choroid in highly myopic eyes
Am J Ophthalmol
Acute zonal occult outer retinopathy: a long-term follow-up study
Am J Ophthalmol
Uveal lymphoid infiltration. Report of four cases and clinicopathologic review
Ophthalmology
Deep retinal vascular anomalous complexes in advanced age-related macular degeneration
Ophthalmology
The blood supply of the optic nerve head and the evaluation of it—myth and reality
Prog Retin Eye Res
Diabetic choroidopathy. Light and electron microscopic observations of seven cases
Ophthalmology
Analysis of normal peripapillary choroidal thickness via spectral domain optical coherence tomography
Ophthalmology
Pathologic myopia and choroidal neovascularization
Am J Ophthalmol
Enhanced depth imaging optical coherence tomography of the sclera in dome-shaped macula
Am J Ophthalmol
Choroidal and retinal blood flow changes in degenerative myopia
Can J Ophthalmol
Radiation choroidal vasculopathy: an indocyanine green angiography study
Eye (Lond)
Subfoveal choroidal thickness in multiple evanescent white dot syndrome
Clin Exp Optom
Tractional retinal detachment following intravitreal bevacizumab (Avastin) in patients with severe proliferative diabetic retinopathy
Br J Ophthalmol
Binding of sulfobromophthalein (BSP) sodium and indocyanine green (ICG) by plasma alpha-1 lipoproteins
Proc Soc Exp Biol Med
Deformation of the lamina cribrosa and anterior scleral canal wall in early experimental glaucoma
Invest Ophthalmol Vis Sci
Reproducibility-repeatability of choroidal thickness calculation using optical coherence tomography
Optom Vis Sci
Fine structure of the choriocappillaris and retinal capillaries
Invest Ophthalmol
Retinal pigment epithelial detachments in the elderly
Trans Ophthalmol Soc UK
Doyne Lecture. Pathogenesis of retinal pigment epithelial detachment in the elderly; the relevance of Bruch's membrane change
Eye (Lond)
Toxoplasmosis
Semin Ophthalmol
Dominant inherited tilted disc syndrome and lacquer cracks
Eye (Lond)
Ocular toxoplasmosis in the fetus. Immunohistochemistry analysis and DNA amplification
Retina
Wolff's Anatomy of the Eye and Orbit
In vivo human choroidal thickness measurements: evidence for diurnal fluctuations
Invest Ophthalmol Vis Sci
Diurnal variations in axial length, choroidal thickness, intraocular pressure, and ocular biometrics
Invest Ophthalmol Vis Sci
Tears of the retinal pigment epithelium: an old problem in a new era
Retina
RPE tears after pegaptanib treatment in age-related macular degeneration
Retina
Indocyanine green: observations on its physical properties, plasma decay, and hepatic extraction
J Clin Invest
Severe myopia as a risk factor for progressive visual field loss in primary open-angle glaucoma
Ophthalmologica
Optical coherence tomography using a frequency-tunable optical source
Opt Lett
Sensitivity advantage of swept source and Fourier domain optical coherence tomography
Opt Express
Individual variations in human cone photoreceptor packing density: variations with refractive error
Invest Ophthalmol Vis Sci
Diagnostic testing and disease monitoring in birdshot chorioretinopathy
Semin Ophthalmol
Normal and staphylomatous sclera of high myopia. An electron microscopic study
Arch Ophthalmol
The posterior staphyloma of pathologic myopia
Trans Am Ophthalmol Soc
Effect of refractive error on the risk of ocular hypertension and open angle glaucoma
Trans Ophthalmol Soc UK
Substance P and calcitonin gene-related peptide intrinsic choroidal neurons in human choroidal whole-mounts
Histol Histopathol
Mapping choroidal and retinal thickness variation in type 2 diabetes using three-dimensional 1060-nm optical coherence tomography
Invest Ophthalmol Vis Sci
Three-dimensional 1060-nm OCT: choroidal thickness maps in normal subjects and improved posterior segment visualization in cataract patients
Invest Ophthalmol Vis Sci
A clinical technique and apparatus for simultaneous angiography of the separate retinal and choroidal circulations
Invest Ophthalmol
Cited by (363)
A Systematic Prospective Comparison of Fluid Volume Evaluation across OCT Devices Used in Clinical Practice
2024, Ophthalmology ScienceOptical coherence tomography in the management of diabetic macular oedema
2024, Progress in Retinal and Eye ResearchVisualization of choroidal vasculature in pigmented mouse eyes from experimental models of AMD
2024, Experimental Eye ResearchInvestigation of the color discrimination ability using the Farnsworth-Munsell 100-hue test and structural changes by SS-OCT in patients with transfusion-dependent beta-thalassemia
2023, Photodiagnosis and Photodynamic TherapyChoroidal vessel segmentation in SD-OCT with 3D Shape-aware Adversarial Networks
2023, Biomedical Signal Processing and ControlPractical guidance for imaging biomarkers in exudative age-related macular degeneration
2023, Survey of Ophthalmology
Financial Disclosure: Dr Spaide is a consultant to Topcon.
The authors have no other financial or personal relationships with other people or organizations that could potentially and inappropriately influence their work.