Light echoes image the human body

17 Jun 2002

The exceptional resolution achieved using optical coherence tomography enables diseases such as cancer and glaucoma to be diagnosed at a very early stage. Rob van den Berg talks to the pioneers of this technique.

From Opto & Laser Europe October 2001

Most medical-imaging technologies allow visualization of tissue anatomy in the human body at resolutions ranging from a hundred microns to several millimetres. These technologies are therefore not sensitive enough to diagnose early-stage tissue abnormalities that are associated with diseases such as cancer and atherosclerosis, which require a resolution of microns.In 1991, however, James Fujimoto's group at the Massachusetts Institute of Technology (MIT) in Boston, US, developed a technique called optical coherence tomography (OCT) - the optical analogue of the well known ultrasound imaging technique. Using the time delay of optical echoes the group succeeded in looking at the internal microstructure of animal tissue.

Since those early days the technique has progressed considerably and, with the advent of ultrashort and ultrabroadband femtosecond lasers, it is now possible to attain micron resolution. Its main field of application is still in ophthalmology, but several groups of researchers are working on extending the technique and developing new applications in medicine.

The most obvious way of improving the technique is by making ever finer details visible. OCT is based on low-coherence interferometry (see below), so its axial (depth) resolution is inversely proportional to the bandwidth of the light source and proportional to the square of the centre wavelength of the light. Standard OCT systems use superluminescent diodes (SLDs) with a bandwidth of a few tens of nanometres as light sources. Typically, this leads to a resolution of 10 to 15 µm, which is still not sufficient to image individual cells and structures within cells, such as nuclei.

With the advent of femtosecond Kerr-lens modelocked lasers, high-resolution and high-speed OCT imaging became possible. This was demonstrated by Wolfgang Drexler from MIT who achieved a resolution of 1 µm and succeeded in imaging the structure inside cells: membranes and nuclei appeared highly backscattering compared with the weakly scattering cytoplasm. OCT could even identify cells in different stages of cell division (Optics Letters 24 1221-3).

Fujimoto said: "A major challenge in OCT is to find a source with a short coherence length, but with enough brightness in a single mode. The ultrafast lasers that generate pulses of less than 5 fs are powerful systems, but they are not practical for clinical use. Recently, special photonic-bandgap fibres have been developed with a broadband continuum - from 1.5 µm to the ultraviolet - that can be generated using self-phase modulation. The advantage of this is that you can use a commercially available Ti:sapphire laser as the input."

Ingmar Hartl of the Institut für Medizinische Optik, Munich, Germany and the MIT group have used this white-light source for the in vivo imaging of biological tissue at a resolution of 2.5 µm and at a centre wavelength of 1.3 µm (Optics Letters 26 608-10).

Although it seems advisable to go to shorter wavelengths to increase the resolution further, infrared wavelengths are generally advantageous because optical scattering is reduced and image penetration depths are improved.

"These photonic fibres will enable imaging to be performed at arbitrary wavelengths," said Fujimoto. "The extremely wide bandwidths also open up the possibility of spectroscopic OCT, thus permitting the differentiation of tissue pathologies via the tissue's spectroscopic properties. Uwe Morgner and colleagues at the University of Karlsruhe, Germany, were the first to demonstrate this in vivo on an African frog tadpole (Optics Letters 25 111-13).

When asked about these developments, Ton van Leeuwen of the Academic Medical Centre (AMC) in Amsterdam, the Netherlands, said: "We will never be able to compete with those technical guys from MIT, but we have the advantage of being in a hospital where we have people around with clinical experience in every field."In the laser centre of the AMC, Van Leeuwen leads a small group that develops new applications for OCT. In collaboration with the group of Joe Izatt of Case Western Reserve University in Cleveland, in the US, Van Leeuwen found a way to determine the velocity-flow profile of blood in an artery.

"We do this using the Doppler shift of the OCT spectrum of the reflected light, and we can measure velocities of up to several hundred millimetres per second," said Van Leeuwen. "The spatial resolution that OCT offers allows us to determine the shear rate of blood flow - the variation of the velocity as a function of distance - under different conditions, for example, in a bent artery. This shear rate is an important indicator of atherosclerosis," he added.

These kinds of experiments may become important in determining the effect of certain forms of treatment, such as balloon angioplasty. Van Leeuwen said: "Most research in this field consists of numerical modelling. There is hardly anyone that is able to carry out measurements."

Another unconventional application under development in Amsterdam is the determination of the oxygen content in blood samples using spectral information from OCT scans. This became possible after the arrival of a femtosecond laser system.Van Leeuwen and his students are also working on a method to detect apoptosis - a regulatory mechanism in which cells are killed by the body. "Apoptosis does not lead to the inflammation of cells: the cell nuclei are cleanly chopped up into smaller fragments. These fragments scatter the light intensely, so we are able to detect the process in its early stages," said Van Leeuwen.

Early detection seems to be a common promise of virtually all OCT applications. Drexler and the MIT group recently succeeded in imaging for the first time the internal microarchitecture of the human retina in vivo (Nature Medicine 7 502-7). The 350 nm bandwidth of the state-of-the-art Ti:sapphire used in this study was centred around 800 nm, which is necessary to avoid absorption in the ocular medium.Fujimoto said: "Visualizing the various layers of the retina and the subsequent processing of the images to identify and measure these layers is important in the early diagnosis and monitoring of diseases such as glaucoma, which is a leading cause of blindness, and macular oedema, which is associated with diabetes.

Symptoms of these diseases only present themselves at an advanced stage. However, by mapping the retina in a sufficient number of patients at different stages of a disease, you can follow the progression of the condition and study the effect of drug treatment. OCT provides information that is not available with any other technique."

A prototype of the MIT OCT instrument has been in use in the New England Eye Center for 10 years and has been used to treat more than 10 000 patients. In 1996 a commercial instrument became available from Zeiss Humphrey, a California-based company that acquired the rights for OCT imaging in ophthalmology. According to OCT business manager Rick Torney: "More than 400 instruments have been sold all over the world." They are almost all based on SLDs, but the company has also looked into the possibilities of ultrabroadband lasers. At this point, however, "the results are not promising for a viable commercial product: the cost of these laser systems far exceeds that of the OCT scanner."

David Kolstad of LightLab Imaging - a start-up firm in Westford, Massachusetts, US, that develops OCT applications outside ophthalmology - agreed: "These table-mounted femtosecond systems are too large and too unstable. We also focus on SLDs with an OCT resolution of between 10 and 15 µm - which is still 10 times as good as that for ultrasound imaging."

LightLab Imaging is active in several areas. Together with Japanese endoscope manufacturer Asahi Pentax, a method is being developed to differentiate between tissue types - for example, normal versus malignant - and to guide biopsies.

Kolstad said: "Currently, many tissue samples have to be taken virtually 'in the blind', and there is a high probability that pre-cancerous tissue is missed. OCT could reduce the false-negative rate and one day eliminate biopsies altogether."

No wonder that the technique is sometimes referred to as optical biopsy. An elegant illustration of this is an imaging needle for OCT, developed by Xingde Li at MIT, which can be inserted into tissue to look for abnormalities before an ordinary needle biopsy is taken of that specific area.

The needle is less than 0.5 mm in diameter and contains all of the optics required for imaging: a lens, a microprism and a tiny window. The optical beam can be scanned radially by rotating the needle with the optics. The 125 nm bandwidth of the Cr:forsterite laser gives an axial resolution of almost 6 µm (Optics Letters 25 1520-2).

A promising application area in cardiology is the diagnosis of so-called vulnerable plaques (pockets of lipid that are ingrained in the arterial wall and covered by a thin cap). Kolstad said: "[The lipid pockets] do not block the arteries and therefore do not cause any pain. However, once they rupture, they form occlusions within one or two hours, which ultimately results in a heart attack. OCT is able to image the structure under the arterial wall, which is invisible using conventional techniques."

Kolstad expects to be able start the first in vivo human studies this autumn, pending FDA approval. The list of potential applications seems endless: in dentistry, pulmonary disease, cervix imaging and gastro intestinal imaging. But, warns Fujimoto: "Most applications are far from mature, and still require a lot of hard work." In OCT and ultrasound imaging the time of flight and the intensity of reflected waves are measured. The main difference between these two techniques is one of scale, because the propagation speeds for light and sound are different. With ultrasound, distances can be measured with a resolution of approximately 100 µm, which corresponds to a time resolution of 100 ns. For light, the echo time delays are much smaller: typically, a distance of 10 µm corresponds to a time resolution of only 30 fs. This means that completely different detection techniques are required.

Today, interferometry is almost always applied to OCT. The easiest implementation is the Michelson interferometer, in which the light is split into a reference and a signal beam. While the former is reflected by a mirror, the latter is backscattered by the material or tissue that is being investigated. By moving the reference mirror an interference pattern is obtained, provided that the difference in path length does not exceed the coherence length of the light source used. The amplitude of the interference signal for different path lengths provides depth information. The OCT contrast is determined by transitions between layers with different indices of refraction - for example, between fatty and muscle tissue - and differences in reflectivity of cellular components, such as membranes and nuclei. James Fujimoto, MIT