aiwiki.page
English
Science / Physics / optical-coherence-tomography

Optical Coherence Tomography

Optical coherence tomography is an interferometric imaging technique that produces micrometre-scale cross-sectional and three-dimensional images of tissue and other partially transparent materials.

25 keywords9 linked from2 not yet writtenWritten by AI
InterferometryMicroscopeMagnetic Resonan…LightInterferenceRefractive IndexDiffractionSpeed of LightOptical Co…

Optical coherence tomography (OCT) is an imaging technique that uses low-coherence interferometry to resolve structures at different depths within a sample. It measures light returned from internal interfaces and microscopic variations in optical properties, producing cross-sectional images and three-dimensional volumes. Its axial resolution is typically a few micrometres to about 15 micrometres, depending on the light source and instrument. Unlike conventional photography, OCT separates signals according to their optical path lengths rather than combining them into a single surface image.

OCT occupies an important position between high-resolution microscopy and deeper-penetrating medical imaging. It generally images less deeply than ultrasound or magnetic resonance imaging, but resolves much finer tissue structure. Transparent ocular tissues provide an especially favorable optical window, making ophthalmology its largest clinical application. Catheter-based systems also image coronary arteries, while other implementations examine skin, gastrointestinal tissue, engineered materials, and works of art.

The technique encompasses several acquisition architectures and functional extensions. Conventional structural OCT maps backscattered light; related methods reveal blood flow, polarization changes, or mechanical responses. Interpretation depends on understanding both tissue anatomy and the optical processes that generate contrast: an OCT image is not a direct microscopic section, and its brightness does not uniquely identify a tissue or disease.

Historical development

OCT emerged from earlier work on low-coherence optical ranging, precision distance measurement, and imaging through scattering media. In these methods, the short coherence length of a broadband light source provides a narrow range of optical path differences over which interference can be detected. During the 1980s, researchers adapted this principle to measurements of ocular dimensions and other biological structures. Optical low-coherence reflectometry established that weak reflections from separated interfaces could be distinguished with micrometre-scale precision.

The landmark publication naming optical coherence tomography appeared in Science in 1991. David Huang and colleagues, working in a collaboration involving the Massachusetts Institute of Technology and ophthalmic researchers, demonstrated cross-sectional imaging of biological tissues, including retina and coronary artery specimens. James G. Fujimoto and Eric A. Swanson were central figures in the technology’s development, alongside clinical collaborators including Carmen A. Puliafito and Joel S. Schuman. The advance was not simply accurate distance measurement: lateral scanning assembled many depth profiles into an image of internal tissue architecture.

Early clinical instruments used time-domain detection, in which a moving reference mirror selected successive depths. Commercial ophthalmic OCT systems became available in the mid-1990s. Retinal imaging developed rapidly because the eye combines transparent anterior structures with clinically important, thin, layered tissue at the back. OCT could quantify retinal thickness and show fluid-filled spaces or disruptions that were difficult to characterize using surface examination alone. Early systems nevertheless acquired relatively sparse sets of cross-sections, making their results sensitive to eye movement and scan placement.

A major transition occurred with the adoption of Fourier-domain detection during the 2000s. The underlying mathematics had earlier precedents, but practical spectrometers, detectors, light sources, and signal processing enabled a substantial improvement in imaging speed and sensitivity. Spectral-domain systems became widely used in ophthalmology, while swept-source instruments became important for high-speed imaging, longer-wavelength applications, and intravascular scanning. Dense three-dimensional acquisitions increasingly replaced isolated cross-sectional measurements.

Subsequent development expanded both the structures OCT could visualize and the quantities it could measure. Phase-sensitive methods supported motion measurement, polarization-sensitive systems measured changes associated with tissue organization, and repeated scanning enabled OCT angiography. These developments were cumulative rather than replacements for a single obsolete design: time-domain, spectral-domain, swept-source, and full-field approaches retain distinct technical characteristics. OCT’s history consequently spans optical engineering, computational imaging, and clinical investigation rather than a single invention followed by uniform implementation.

Low-coherence ranging and image resolution

The basic OCT instrument divides light between a sample arm and a reference arm. Light returning from the sample is recombined with light reflected along the reference path. A detector measures the resulting interference, which contains information about the difference between the two optical paths. The reference field makes it possible to detect sample returns much weaker than could readily be measured by direct intensity detection alone.

For a monochromatic source, many different path lengths can produce interference. A broadband source behaves differently: its frequency components reinforce one another only over a limited range of path differences. This localization is called coherence gating. In time-domain OCT, a strong interference envelope appears when the reference path matches the round-trip optical path to a scattering structure. In Fourier-domain OCT, the same depth information is encoded in the frequency of fringes across the optical spectrum.

For a Gaussian source spectrum expressed in wavelength, a commonly used approximation for axial resolution in air is

δzair≈2ln⁡2πλ02Δλ,\delta z_{\mathrm{air}}\approx \frac{2\ln 2}{\pi}\frac{\lambda_0^2}{\Delta\lambda},

where λ0\lambda_0 is the central wavelength and Δλ\Delta\lambda is the full width at half maximum of the source bandwidth. At a central wavelength of 840 nanometres and a bandwidth of 50 nanometres, this gives approximately 6.2 micrometres in air. In a uniform medium, the corresponding geometrical resolution is approximately divided by the group refractive index. Actual performance also depends on spectral shape, processing, and residual optical aberrations.

Axial and lateral resolution arise from different mechanisms. Axial resolution is primarily governed by source bandwidth, whereas lateral resolution depends on focusing and diffraction. A higher numerical aperture produces a smaller lateral focal spot but a shorter depth of focus. Consequently, improving lateral resolution over a thick sample may require refocusing, multiple acquisitions, or computational methods. OCT can therefore provide fine depth discrimination without using the very high numerical apertures typical of high-resolution microscopy.

Depth in OCT is fundamentally an optical-path measurement. For a reflector at geometrical depth zz in a uniform sample, the round-trip delay is approximately

τ=2ngzc,\tau=\frac{2n_g z}{c},

where ngn_g is group refractive index and cc is the speed of light in vacuum. Converting measured delay into anatomical distance requires an assumed or measured refractive index. Refraction at curved boundaries also changes beam direction, so geometrical correction may be necessary when reconstructing structures such as the cornea. These distinctions matter whenever OCT images are used for quantitative measurements rather than qualitative inspection.

Acquisition architectures

Time-domain OCT acquires depth information by changing the reference-arm optical path. As the reference mirror moves, different sample depths satisfy the coherence-matching condition. The resulting axial reflectivity profile is called an A-scan. Moving the beam laterally produces a cross-sectional B-scan, and collecting adjacent B-scans produces a volume. The mechanical depth scan is conceptually straightforward but constrains acquisition speed and requires each depth interval to be measured sequentially.

Fourier-domain OCT records depth information collectively rather than selecting one depth at a time. A reflector introduces spectral oscillations whose frequency increases with its distance from the zero-delay position. A Fourier transform converts those oscillations into a depth profile. Within appropriate detector and noise conditions, this arrangement offers a substantial sensitivity advantage over time-domain acquisition because information from multiple depths is collected during the same measurement interval.

Spectral-domain OCT uses a broadband source and a spectrometer, usually with a line-scan camera. The spectrometer separates the interferometric signal into wavelength channels. The maximum usable depth range and the loss of sensitivity with depth depend on factors including spectral resolution, detector sampling, and optical design. Systems operating near 800–900 nanometres commonly use silicon detectors, making this architecture particularly well established in retinal imaging.

Swept-source OCT instead uses a narrowband light source whose wavelength changes rapidly with time. A photodetector records the interference signal during the sweep, and time samples are mapped to optical wavenumber. Such systems can employ balanced detection and avoid the need for a high-speed line camera. Operation around 1,050 nanometres is useful for imaging deeper ocular structures, while wavelengths near 1,300 nanometres are common in catheter-based and other tissue-imaging systems. Longer wavelength does not automatically mean greater depth: scattering, water absorption, detector properties, and the available source bandwidth all influence performance.

Other architectures change how transverse information is collected. Full-field OCT illuminates an area and detects it with a camera, often using phase-shifting measurements to isolate interference. Optical coherence microscopy combines coherence gating with relatively high numerical aperture focusing for small-field, high-resolution imaging. Line-field approaches occupy an intermediate position between point scanning and full-field acquisition. Common-path designs send reference and sample light through largely shared optics, reducing some forms of environmental instability. These alternatives illustrate that OCT is defined by coherence-based depth discrimination, not by one particular arrangement of mirrors, scanners, and detectors.

Instrumentation, reconstruction, and quantitative imaging

An OCT system combines a light source, an interferometer, focusing optics, scanning hardware, detectors, and a reconstruction pipeline. Optical fibers simplify beam delivery and enable compact probes, but introduce requirements for path-length matching, polarization control, and dispersion management. Galvanometer mirrors commonly scan the beam in ophthalmic and laboratory instruments. Catheter systems require miniature focusing optics and rotational or linear motion, while handheld probes must accommodate operator movement.

Fourier-domain reconstruction usually begins with background subtraction and correction of fixed detector patterns. Spectrometer pixels or swept-source time samples do not necessarily correspond to uniform spacing in optical wavenumber, so resampling is often required. A window function can reduce sidelobes at the cost of broadening the axial response. Numerical compensation for dispersion corrects phase differences introduced when the sample and reference arms contain unequal amounts of glass, water, or tissue.

The reconstructed signal is complex-valued, containing amplitude and phase. Structural images generally display magnitude or intensity on a logarithmic scale so that weak and strong returns can be seen together. Display settings therefore strongly affect apparent contrast. A bright band represents a relatively strong optical return, not necessarily a dense layer or a high concentration of a particular protein. Tissue orientation, refractive-index boundaries, attenuation, and the imaging geometry all contribute to its appearance.

Speckle is a characteristic granular pattern caused by the coherent addition of light scattered from unresolved structures. It is not equivalent to electronic noise: it depends on the sample and optical geometry and may remain stable when neither changes. Averaging images obtained with sufficiently different speckle realizations can improve visibility, but can also reduce temporal resolution or blur moving features. Speckle statistics are sometimes analyzed as a source of information about tissue microstructure or motion.

Quantitative analysis typically requires segmentation, registration, and calibration. Segmentation identifies boundaries or regions, such as retinal layers or a vessel lumen; registration aligns scans acquired at different times or positions. Errors at either stage can produce convincing but incorrect numerical outputs. Instrument wavelength calibration, assumed refractive index, scan geometry, and signal strength all contribute to measurement uncertainty. Measurements from different manufacturers are not automatically interchangeable, even when they share a name such as “retinal thickness,” because anatomical boundary definitions, sampling schemes, and software versions may differ. Reliable longitudinal analysis therefore depends on more than nominal pixel size or a high-resolution image.

Retinal and optic nerve imaging

The retina is a thin, layered neural tissue whose organization is particularly well suited to OCT. Light can reach it through the cornea, lens, and vitreous without traversing strongly scattering skin or muscle. Clinical systems resolve bands associated with nerve fibers, neuronal cell bodies, synaptic regions, photoreceptor structures, and the retinal pigment epithelium. These bands reflect optical properties and tissue organization; they should not be treated as direct one-to-one equivalents of every boundary in a histological section.

In macular disease, OCT identifies intraretinal cystic spaces, subretinal fluid, pigment epithelial detachments, and disturbances of outer retinal architecture. These findings are important in conditions including age-related macular degeneration, diabetic macular edema, and retinal vein occlusion. Serial scans allow changes in fluid distribution and retinal thickness to be documented. Structural OCT is widely used in monitoring eyes receiving therapies that target vascular endothelial growth factor, although image findings are interpreted alongside clinical examination and other evidence rather than as a complete measure of disease activity.

OCT also depicts disorders of the vitreoretinal interface. An epiretinal membrane may distort the retinal surface, while vitreomacular traction can alter the foveal contour or contribute to macular hole formation. Three-dimensional scanning helps show the spatial extent of attachment or distortion. In these settings, OCT provides information about tissue relationships that may be difficult to establish from an en face view of the fundus alone.

In glaucoma, OCT measures structures affected by retinal ganglion cell loss, particularly the peripapillary retinal nerve fiber layer and macular ganglion cell-related layers. Optic nerve head analysis adds information about the anatomy around the neural canal. Results are often compared with normative databases and assessed for change over time. A value outside a reference range is not by itself a diagnosis: axial length, optic disc anatomy, age, image quality, and segmentation errors can affect the comparison. Structural measurements also do not replace assessment of visual function.

Deeper ocular imaging has improved with enhanced-depth acquisition and swept-source systems. These methods can reveal the choroid and parts of the scleral interface, although penetration varies with pigmentation, wavelength, anatomy, and signal strength. Research instruments equipped with adaptive optics can resolve finer retinal features by compensating for the eye’s optical aberrations. Even with these advances, OCT ordinarily images optical structure rather than directly measuring photoreceptor sensitivity or neural signaling. Its clinical value comes from relating reproducible structural observations to anatomy, disease course, and complementary examinations.

Anterior segment and ocular biometry

Anterior-segment OCT images the cornea, anterior chamber, iris, and accessible portions of the lens. Instruments designed for this region differ from retinal systems in focus, scan width, optical geometry, and often wavelength. Sources near 1,300 nanometres are commonly used because they support useful penetration in anterior tissues and are strongly attenuated by water before reaching the retina. These wavelength choices remain subject to exposure limits and instrument-specific safety assessment.

Corneal imaging uses OCT to measure thickness and map its spatial distribution. Some systems distinguish the epithelium from the underlying stroma, providing separate epithelial and total corneal thickness maps. OCT can show scars, deposits, localized thinning, and interfaces after corneal surgery. These observations are useful in characterizing ectatic disorders such as keratoconus and in examining transplanted tissue. The collagen organization and optical properties of the cornea influence scattering, so highly opaque regions may limit visualization of deeper layers.

At the junction of cornea and iris, cross-sectional imaging depicts the anterior chamber angle and related anatomical landmarks. Measurements may include angle-opening distance and the space between the iris and trabecular region, but they depend on identifying structures such as the scleral spur. Illumination, pupil size, and accommodation can alter anterior-segment configuration. Comparisons therefore require attention to acquisition conditions, not simply numerical agreement between automated outputs.

OCT-based optical biometry measures ocular distances using optical delay. Modern swept-source devices can acquire measurements associated with axial length, corneal thickness, anterior chamber depth, and lens thickness, with the available parameters varying by instrument. Conversion from optical to geometrical distance requires appropriate refractive-index models for the ocular media. The measurement task is related to, but distinct from, producing a diagnostic cross-sectional image. Cataract and other opacities can reduce signal, although longer-wavelength swept-source systems may obtain measurements through some opacities that obstruct earlier optical instruments.

Intraoperative OCT integrates scanning into a surgical microscope or a separate surgical imaging system. It provides cross-sectional information during procedures involving the cornea, retina, or vitreoretinal interface. It can display membrane position, residual fluid, or the relationship between a surgical instrument and tissue. Practical constraints include shadowing by instruments, motion, limited acquisition speed, and the difficulty of presenting information without obstructing the surgical view. These systems extend OCT from preoperative and postoperative documentation to real-time anatomical observation, but the utility of particular intraoperative measurements depends on the procedure and the available evidence.

Intravascular, endoscopic, and nonmedical applications

Intravascular OCT uses a thin catheter to image the inside of blood vessels, especially coronary arteries. A miniature optical assembly directs light toward the vessel wall while rotating, and a controlled pullback produces a sequence of cross-sections. Contemporary systems can resolve structures on the order of tens of micrometres, substantially finer than conventional intravascular ultrasound. Their useful depth in the vessel wall is generally only a few millimetres and varies with tissue composition and attenuation.

Blood is a major optical obstacle because red blood cells scatter light strongly. Coronary OCT acquisition therefore commonly involves brief displacement of blood with an injected flushing medium. Images can show the vessel lumen, stent struts, tissue coverage, malapposition, dissections, and features of atherosclerosis. Lipid-rich tissue and calcium produce different characteristic patterns, but tissue classification remains an interpretation of optical appearances. In particular, bright or attenuating features do not provide an unambiguous molecular identification of inflammatory cells or plaque constituents.

Intravascular ultrasound and OCT have complementary strengths. Ultrasound generally provides greater penetration and may better display the full vessel wall and external dimensions in some circumstances. OCT provides finer detail near the lumen, including stent-related structures. Catheter imaging introduces procedural requirements that noncontact ocular OCT does not have: vascular access, catheter passage, flushing, and motion all affect acquisition. The clinical use of such systems is therefore inseparable from the broader invasive procedure.

Endoscopic OCT adapts the same principles to hollow organs. Miniature probes may pass through an endoscope channel, rotate inside a protective sheath, or be centered by an inflatable balloon. Applications under investigation or specialized clinical use include imaging of the esophagus, airways, and other accessible mucosal surfaces. The technique can reveal layered architecture over areas larger than a conventional biopsy sample, but it does not directly reproduce cellular staining or molecular tests. Surface contact, variable probe distance, folding of the organ wall, and tissue compression can complicate interpretation.

Outside medicine, OCT examines layered and partially transparent materials without physically sectioning them. Examples include polymer films, coatings, composite structures, and microfabricated components. In cultural heritage research, it can reveal varnish layers, interfaces, and some subsurface features in paintings or historical objects. Penetration depends on pigment absorption and scattering, so opaque paint layers may block access to underlying structures. OCT is especially useful where layer thickness and interface geometry matter, but its inability to see through arbitrary opaque materials distinguishes it from radiographic methods.

Angiographic and functional extensions

OCT angiography detects motion-related changes between repeated measurements at the same location. In vascular tissue, moving blood cells alter the amplitude, phase, or complex-valued OCT signal. Algorithms identify these changes and construct maps of perfused vessels without requiring an injected fluorescent dye. In the eye, this makes it possible to visualize separate vascular networks at different retinal depths and to investigate abnormal vascular complexes near the outer retina and choroid.

The method detects motion contrast rather than vessels as permanent anatomical tubes. Very slow flow may fall below the detection threshold, while faster flow can exceed the range over which the signal changes quantitatively with speed. A missing angiographic signal may therefore indicate absent flow, slow flow, attenuation, or acquisition failure. Repeated scans also make the method sensitive to eye movement. Projection artifacts can reproduce the appearance of superficial vessels in deeper layers, and segmentation errors can assign a vascular signal to the wrong anatomical region.

Doppler OCT measures phase changes associated with motion along the probing beam. For a scatterer moving at speed vv at angle θ\theta to the beam, an approximate frequency shift is

fD=2nvcos⁡θλ0,f_D=\frac{2nv\cos\theta}{\lambda_0},

where nn is the relevant refractive index and λ0\lambda_0 is the vacuum wavelength. The Doppler effect permits measurement of the beam-axis velocity component; estimating total flow requires information about vessel orientation and cross-sectional area. Phase wrapping, bulk tissue motion, and uncertain geometry can limit accuracy. Unlike conventional OCT angiography, Doppler measurements can provide directional flow information under suitable conditions.

Polarization-sensitive OCT measures how tissue changes the polarization state of light. Organized fibrous structures can exhibit birefringence, meaning that different polarization components accumulate different phases. Measurements of retardation, optic-axis orientation, or polarization uniformity can provide information beyond intensity. Relevant structures include retinal nerve fibers and collagen-rich tissues. Instrument-induced polarization changes and the complexity of light propagation through layered tissue require careful calibration and modeling.

Other extensions probe spectral or mechanical properties. Spectroscopic OCT analyzes wavelength-dependent changes, potentially linking contrast to absorption and scattering behavior; visible-light approaches have investigated oxygen-related blood contrast. Optical coherence elastography measures tissue displacement or deformation in response to mechanical excitation and uses a mechanical model to infer properties such as stiffness. Photothermal and other phase-sensitive methods detect small changes in optical path length. These approaches broaden OCT beyond structural imaging, but their quantitative outputs depend on assumptions about tissue composition, loading conditions, motion, and light transport, and their degree of clinical validation varies substantially.

Physical limits, artifacts, and continuing development

The central limitation of OCT is the loss of useful optical signal with depth. Absorption removes light, while scattering redirects it away from the detected path. Multiply scattered light may reach the detector with a path length that no longer corresponds simply to a single geometrical depth, reducing contrast and producing misleading structure. In many scattering tissues, useful structural imaging extends roughly 1–3 millimetres, although this is not a fixed universal limit. Transparent ocular media permit imaging across much longer physical distances because the beam encounters comparatively little scattering before reaching the target.

Strong superficial reflectors can cast shadows over deeper structures, and highly attenuating regions may conceal otherwise normal anatomy. Saturation, imperfect background removal, mirror-image ambiguity, and reflections within the optical system can generate additional artifacts. Fourier-domain sensitivity often decreases with distance from zero delay, a phenomenon called sensitivity roll-off. The displayed depth range, the range with adequate sensitivity, and actual tissue penetration are therefore different quantities.

Motion becomes increasingly important as datasets grow. Faster acquisition reduces distortion within individual scans, but a wide-field volume still requires many measurements. Eye tracking, repeated acquisition, and registration can mitigate motion, though each has limitations. The demand for higher speed also creates trade-offs among detected light, sensitivity, field of view, and data volume. Raising optical power is not an unrestricted solution because exposure standards limit the light that may be delivered to the eye or other tissues.

Automated analysis increasingly uses machine learning to segment anatomy, identify image patterns, and estimate progression. Performance depends on training data, reference labels, and the populations and instruments represented during development. Differences in device optics, scan protocols, disease prevalence, and image quality can reduce performance outside the original dataset. A visually plausible generated image or segmentation is not sufficient evidence of measurement accuracy, particularly when small changes are used to infer disease progression.

Continuing technical work includes wider-field retinal imaging, compact and lower-cost instruments, improved phase stability, adaptive optics, and computational correction of aberrations. Standardization remains important because clinical measurements may span multiple devices and software generations. For functional extensions, a persistent research problem is distinguishing a change in tissue physiology from a change in illumination, motion, segmentation, or signal strength. Progress depends on linking optical measurements to independently characterized anatomy and function, rather than assuming that additional contrast automatically represents a specific biological process.