2.2.1 Light in Flight

In 1971, Michael Duguay, then a researcher at AT&T Bell Laboratories, first proposed that the capture of echoes of light or “light in flight” could yield useful data about the composition of biological tissues. ^{3 , 4} With the use of a laser-activated Kerr shutter, he proposed a method of freezing the motion of light in order to noninvasively image tissues. ^{3 , 4} In the same year, Eric Ippen left Bell Labs for Massachusetts Institute of Technology (MIT; Cambridge, MA, USA) where doctoral student James Fujimoto soon joined him. ⁵ The pair and their collaborators began the hard work of developing a practical application for Duguay’s theoretical proposal, and the work that would ultimately culminate in the invention of OCT began. ⁵

By the 1980s, this group had their sights on skin and skin diseases as the first target for applying femtosecond laser technology to clinical medicine. ⁵ The researchers soon found that skin yielded a high degree of optical scattering in response to laser light and thus shifted their focus to the eye. ⁵ With transparent structures throughout its length—from cornea to lens to vitreous to retina—the eye proved to be the optimal test case for developing a practical medical application for capturing “light in flight.” ⁵

Ex vivo bovine and rabbit eyes were the subject of early work in this arena, and experiments on these animal models provided useful insights. ⁵ The earliest experiments utilized laser light with a wavelength of 625 nanometers (nm), resulting in scans with a sensitivity of −70 decibels (dB). ⁵ Later work would find that longer wavelengths were more effective in reducing attenuation from optical scattering, thus improving sensitivity. ⁵ For context, through this change among others, modern OCT machines have achieved three log units better sensitivity than those early scans. ⁵

The original experiments sought to “see inside” of ophthalmic tissues through nonlinear cross-correlation, in which the instrument produced two beams of light: one directed at the tissue and one reference beam with a variable time delay. ^{5 , 6} By assessing the varying echo profiles and time delays of these beams (e.g., by analyzing and comparing the backscattered and backreflected beams of light), the prototype devices produced inference patterns related to the structure of the tissue, culminating in the generation of an axial scan (A-scan). ^{5 , 6} As early as this initial work, there was a focus on applying the technology to assess pathological states. ⁵

2.2.2 Interferometry

In the late 1980s, superluminescent diode interferometers replaced femtosecond lasers as the primary light source in OCT research. ⁵ Interferometry hastened the development of a feasible clinical product because it enabled the creation of instruments that were less expensive yet offered improved sensitivity. ⁵ Interferometry had its roots in the work of Sir Isaac Newton. ^{5 , 7} Its first real-world application was in the telecommunications industry, in which this technology improved the transmission of optical data. ^{5 , 7} Early OCT work used Newton’s classic technique of low-coherence or white-light interferometry. ⁷

Interferometers worked by comparing an optical beam with a reference beam. ⁷ First, a laser source emitted a light, which a partially reflecting mirror called a beamsplitter divided into two perpendicular beams: one beam that would travel into the tissue of interest (e.g., the eye; this was the optical beam), and a reference beam. ⁷ A mirror at a known distance reflected the reference beam such that it traveled back to the beamsplitter at a known time delay (e.g., it served as a time reference). ⁷ At the beamsplitter, this reference beam interfered with the optical beam after it was backreflected and backscattered by the tissue. ⁷ Upon returning from the eye, the optical beam had multiple echoes resulting from the structural variations among the tissues within the eye. ⁷ In other words, intraocular structures had variable microscopic composition and were located at different distances from the light source such that each tissue type reflected and scattered the optical beam differently. ⁷ Thus, the variable echoes of light within the optical beam corresponded to microstructural nuances within the eye. ⁷ Ultimately, a detector compared the reflection and scattering of both beams, measuring the interference or correlation between them. ⁷ This method of time domain detection enabled ultrahigh-resolution time and distance measurements and eventually gave rise to the earliest commercial OCT instruments. ^{5 , 7}

Fercher and his collaborators at Medical University of Vienna (Vienna, Austria) published the first application of interferometry in medicine in 1988, measuring the axial length of in vivo human eyes; their results correlated with the acoustically measured axial lengths within 0.03 mm. ⁸ Despite its advantages in precision and speed, interferometry was not a practical medical tool. ⁵ Researchers and subjects struggled with its sensitivity to movement and vibration and the fact that bulk optics required very precise alignment in order to avoid signal loss. ⁵ In the US, John Apostolopoulous, then an undergraduate student at MIT, pioneered much of the early work in interferometry. ⁵ His experiments proposed a technique similar to nonlinear cross-correlation, but he substituted an inexpensive low-coherence diode laser for the femtosecond lasers used in the latter technique. ⁵ Apostolopoulous’s experiments did not have sufficient sensitivity for generating scans of the eye, but he described a theoretical means for doing so in his unpublished 1989 bachelor’s thesis. ⁵

2.2.3 The Pivotal Role of Collaboration

The developers of OCT worked at an astonishingly fast pace to translate the “light in flight” principle to meaningful clinical impact. The diversity in background and level of training among those involved played a key role in facilitating this expeditious development. The team included a range of scientists from undergraduates like Apostolopoulous to senior principal investigators as well as the full spectrum of medical personnel from preresidency fellows to attending physicians. ⁵ In 1990, Eric Swanson, an engineer at MIT’s Lincoln Laboratories, joined the mix. ⁵ In contrast to many academic enterprises, the emphasis of the work at Lincoln Labs is deeply pragmatic; there is a substantial focus on Department of Defense technology and advanced engineering with an emphasis on feasibility and implementation. ⁵ Swanson worked on intersatellite optical communications and fiber optics networking. ⁵ Compared to the bulk optics previously used in interferometry, fiber optics mitigated alignment issues and enabled use in catheters or endoscopes. ⁵ The latter of these features enabled OCT development work in intravascular applications, including measurement of coronary artery plaques. ⁵ Swanson brought his content expertise and the Lincoln Labs implementation-oriented mindset to the OCT effort. ⁵ With the application of fiber optical elements and other contributions from Swanson, the feasibility of OCT improved dramatically: imaging speed became 100 times faster, the design became more compact (the prototype instrument that initially required a 1-square meter table could then sit on a platform just 19 inches wide), and the patient interface became more flexible. ⁵

2.3.1 OCT versus Ultrasound

Ultrasound (US) existed long before OCT, but both technologies hinged on strikingly similar principles. Whereas ultrasound generated images through the measurement of the time delay and intensity of backreflected and backscattered echoes of soundwaves, OCT sought to achieve the same with echoes of light waves. ⁷ Each technology had a set of advantages and drawbacks that made it most suitable for measurement of different biologic structures. ⁷ US required direct contact of the probe to the tissue it measured as well as the use of gel in order to couple the transmitter–receiver and the tissue, in order to transmit sound waves appropriately; OCT did not require physical coupling or the use of a coupling agent. ⁷ Although these features of US did not inhibit its application in imaging most structures of the body (e.g., intraabdominal organs), they made it less attractive for imaging the fine structures of the eye. ⁷ US waves had a sufficiently low frequency to propagate to the deep structures of the body, but this came with the trade-off of poorer image resolution. ⁷ In contrast, the high frequency light waves emitted in OCT precluded penetration through most opaque biological tissues due to the high degree of scattering and absorption but could achieve much finer resolution. ⁷ This principle limited OCT’s use to structures that were “optically accessible” (i.e., the optically clear components of the eye) and some structures in turbid media that could be accessed via catheters or endoscopes (e.g., coronary plaques). ⁷ Despite these drawbacks compared to US, OCT emerged as a superior technology for imaging of the eye’s intricate microstructure, not only because an instrument that did not require contact with the eye was more tolerable for patients, but also because the use of shorter wavelength light permitted the collection of data that highlighted details as fine as the intricate layers of the retina with a much higher axial resolution than US. ⁷

2.3.2 How OCT Works (Fig. 2‑2)

Starting from its earliest prototypes, OCT leveraged the intrinsic differences between tissues within the eye to generate detailed images. ⁷ Various tissue-dependent phenomena occurred when the light beam from OCT instruments passed through the structures of the eye. For one, tissues could transmit light; that is, light could continue propagating into deeper tissue layers and structures, much like sound waves in US propagate from superficial skin, subcutaneous tissue, and fascia to the deep viscera. ⁷ Second, tissues could absorb light. ⁷ Tissues that absorbed light, such as those containing melanin or hemoglobin, effectively removed certain wavelengths from the beam. ⁷ This principle explained why the retinal pigment epithelium, a melanin-containing layer, appeared distinctly different in OCT images than microstructures that contained no or different pigments. ⁷ Third, tissues could reflect light back toward the receiver. ⁷ Reflection occurred at the borders of tissues or substances with different indices of refraction, such as at the air-corneal interface or vitreo-retinal interface. ⁷ Finally, tissues could scatter light; scattering occurred due to compositional variations within cells, often due to the presence of major intracellular components including nuclei and other organelles. ⁷ Because of this principle, OCT distinguished the layers of the retina that were composed of organelle-laden cell bodies from those that contained mostly axons; the composition of the latter included primarily cytoplasm and cell membranes with fewer organelles, thus resulting in a lesser degree of scattering when the incident beam met that layer. ⁷ Of note, when tissues scattered light, it propagated in multiple directions; the portion of that scattered light that propagated in the reverse direction of the incident beam was called backscattered light. ⁷ Since OCT instruments only detected light along the same axis as the incident beam, they only detected the backscattered portion of all scattered light. ⁷ Due to these principles, tissues with higher degrees of absorption appeared darker in OCT images, and those with a higher degree of scattering (and thus a more substantial amount of backscattered light) appeared brighter. ⁷

These principles also explained why OCT was not optimal for imaging deep structures of the body. ⁷ Due to the strong absorption and scattering of light incident upon skin, the penetration of the beam was too shallow to reach deeper structures. ⁷ That is, when a tissue absorbed or scattered light too strongly, it effectively cast a shadow over deeper structures and tissues. ⁷ On the other hand, the primarily transparent structures of the eye transmitted most light and demonstrated a very low degree of reflection, absorption, and scattering. ⁷ OCT was developed with sufficient sensitivity to detect the very minor differences among layers that were all weakly absorptive, backreflective, and backscattering. ⁷

Even in its earliest iterations, OCT produced images with microstructural precision on par with histological biopsies. Though the results appeared similar to histologic samples (and the results of early OCT scans were validated through histology), the mechanism of OCT was entirely different. ⁷ Whereas histology relied on external markers of cellular or subcellular components to differentiate structures, OCT leveraged intrinsic features. ⁷

2.3.3 The First OCT Images

In the late 1980s, one of the authors of this chapter (Joel S. Schuman) became involved in the OCT development effort as a fellow at Massachusetts Eye and Ear Infirmary/Harvard Medical School. ⁹ He proposed measurements of the deeper structures of the eye (e.g., the retina and its intricate layers which, though all transparent, were optically distinct) via optical coherence domain reflectometry. ⁹ After the group’s work validated the utility of optical coherence domain reflectometry for measuring structures throughout the eye via individual A-scans, another student working on the project, David Huang, then a Health Sciences Technology Student at Harvard Medical School and Massachusetts Institute of Technology, had the insight that the clear next step was transitioning from multiple A-scans to a two-dimensional B-scan analogous to B-mode ultrasound. ⁹ This transition marked the invention of OCT; the inventors named the technology optical coherence tomography and published the first OCT images in Science. ^{1 , 9}

The 1991 Science paper demonstrated features of postmortem, in vitro human retina with detailed depictions of the optic nerve head and nerve fiber layer. ¹ The paper also featured images of a fibrocalcific plaque in a postmortem, in vivo human coronary artery specimen. ¹ To generate the images, multiple A-scans were combined using a logarithmic false color or gray scale. ¹ Following the generation of these images, the tissues underwent histologic evaluation, which recapitulated the structural findings and validated the technique. ¹ This publication was not only scientifically interesting but also demonstrated promising clinical applications of OCT in the fields of ophthalmology and cardiology. Of note, a 1994 Japanese patent described a similar concept, though there was no corresponding publication in the literature to catalogue the work of that group. ¹⁰

At this point, the multidisciplinary team of scientists, engineers, clinicians, and students who contributed to OCT’s development saw the culmination of their efforts in the publication of the first OCT images. The contributions of this team, and Swanson’s engineering expertise, yielded a sound hardware platform for OCT. ⁹ However, the software was not yet adequate for imaging in vivo specimens. ⁹