PerspectiveIMAGING

Tissue Optics

See allHide authors and affiliations

Science  27 Jun 1997:
Vol. 276, Issue 5321, pp. 2002-2003
DOI: 10.1126/science.276.5321.2002

Our bodies are not impervious to light. When you step outside on a sunny day, the inside of your head is illuminated—in fact, sufficient light penetrates your skull to comfortably read this page (1). Shine a flashlight at your finger, and your finger will glow red, a demonstration that red and near-infrared light penetrate deeply into tissue. Just as farmers once candled eggs to check for chick embryos, transillumination has been used to screen for scrotal tumors, breast cancer, and blood oxygenation (2). Although the resolution and diagnostic accuracy of these early attempts at imaging with light were poor, this is changing. In this issue on page 2037, Tearney et al. (3) use light-based optical coherence tomography (OCT) to form tomographic images of human tissue in situ with a resolution exceeding that of magnetic resonance imaging (MRI), computerized tomography (CT), or ultrasound.

In tissue, light is both absorbed and scattered; scattering dominates in the red and near-infrared spectrum (see the figure). This native scattering renders us opaque and blurs transmission images. However, those photons surviving passage through the tissue emerge bearing clues about their voyage. Study of their behavior has spawned a field now known as tissue (or biomedical) optics (see related News story on page 1991). Optical approaches to tissue imaging have general advantages: They are inexpensive, noninvasive, transportable, and nontoxic. In addition, biological materials have native contrast that allows optical assessment of tissue chemistry and cellular structure.

Path of light.

Light moves randomly through tissue because of scattering, which dominates over absorbance for red and near-infrared light. At shallow depths, some photons are coherently reflected by a single scattering event (A), permitting detection by interferometric techniques. Images using this coherently reflected signal can be constructed, such as in OCT. At deeper depths, light is multiply scattered and can either be absorbed (B) or survive tissue transit to escape (C) and be detected. Images can be constructed using this diffuse light, as in optical diffusion tomography. Light can also be generated within the tissue (D) and then externally detected.

OCT takes advantage of the scattering inherent in tissue to generate in vivo images. As photons meander through tissue, a small amount of light that has bounced off internal structures only once is coherently back-reflected from various boundaries within the tissue. Using interferometric techniques, this coherently backscattered light can be detected at intensities down to 1 part in 1011, while simultaneously revealing the depth of the scattering event. At the heart of OCT is a Michelson interferometer, which detects interference between halves of a single light beam split along a sample arm, containing low-coherence light backscattered from multiple depths within the tissue; and a reference arm, containing light passing through an air or water gap of variable length (4). [HN2], [HN3], [HN4] When sample and reference light are recombined, and the time delay through both paths matches within the coherence length of the source, they interfere. This interference indicates that a component of light backreflected by the tissue traversed a path with a time delay equal to the delay through the reference path. By varying the length of the reference path, multiple depths into the tissue can be sampled. As long as the speed of light in tissue (c divided by the refractive index) is constant, the length of the reference path is linearly related to tissue depth probed. This allows 10- to 20-μm spatial resolution into the tissue along the beam axis. Finer image resolution can be achieved by using short coherence-length sources. Images may be formed by scanning this depth-probing beam either in an x-y grid pattern or in a rotational circular scan.

OCT has been used before as an external or ex vivo imaging approach—for skin, cardiac vessels, embryos, eye, and other sites (5-9). Clinically, it has been the most useful in ophthalmology, where optical imaging of cross sections of the retina and other optic substructures has assisted diagnosis and therapy (8) of diseases such as macular degeneration and choroidal neovascularization.

In the present study, Tearney et al. have converted their device into a catheter-based system. The result is finely resolved images of esophageal tissue layers. This report is significant in several ways. First, the 10-μm spatial resolution of OCT outperforms that of MRI, ultrasound, and confocal microscopy, which image with 200-μm resolution. [HN5] Even high-frequency (100 MHz) intraocular ultrasound produces only 20- to 40-μm resolution. Second, OCT can be used for biopsies in situations where excisional biopsies are difficult. Biopsy that requires tissue sampling is less than ideal for the eye, coronary vessels, brain, or nervous tissue, all of which can be irreversibly damaged by biopsy or are difficult to access. A catheter-based OCT system such as that reported by Tearney et al. lends itself to use during endoscopy, opening much of the body to view via minimally invasive approaches. (Although for this approach to have medical value, cross-validation with conventional pathology approaches will be needed.) Third, even if excisional biopsy is required, OCT may be useful in guiding the surgeon in selecting the optimal site for maximal biopsy yield. Last, extensions of the technique are possible, including multispectral analysis, optical Doppler tomography (ODT), and intravascular imaging. Each of these areas is under study.

OCT has several limitations. Excised tissue can be stained for the presence of antigenic or structural components, cultured for infectious agents, and viewed under a microscope with a resolution approaching 1 μm or better using near-field or electron microscopy. In contrast, OCT cannot identify the molecular components of tissue, normal or pathogenic. The spatial resolution of OCT is considerably less than that of an excision biopsy. Short-pulse OCT may increase axial resolution to 2 to 4 μm, but improvements in lateral resolution are more difficult. Scattering degrades beam coherence and limits the effective spot size inside tissue, while variations in refractive index over the sampled region further reduce image quality (10). Thus, the theoretical diffraction limit for resolution is never achieved in practice.

In addition, the depth of imaging for OCT is limited to 2 to 4 mm. At depths beyond several millimeters, essentially all photons have undergone multiple scattering and lost coherence. Here, diffusion theory, which treats photon migration through tissue as a diffusive process similar to the diffusion of heat or gases, works well (see News story on page 1991). Such an approach has allowed imaging of stroke, brain hemorrhages, and breast tumors in human subjects, although at a lower resolution than OCT (11), and has allowed chemical characterization of tissue.

Finally, illumination-based approaches work only if there is a distinctive and detectable optical signature. Subtle differences between tissues (such as between cancerous and normal cells) demand a high degree of spectral sensitivity and may not be detectable with OCT. Key molecules in many biological processes—those associated with gene activation or infection, for example—may possess no distinguishing spectral features, thus limiting assays of such events to in vitro tests removed from the relevant influences of the intact organism. Here, use of an internal light beacon, conceptually parallel to PET or nuclear medicine scanning, has allowed for real-time, noninvasive monitoring and imaging of infection and gene expression in vivo (12).

In the future, OCT and other optical techniques may usher in the use of light-assisted medical diagnosis, or “optical biopsy,” leading to automated surgical tools or portable diagnostic devices that can identify tumors, provide feedback to surgical interventions such as tissue welding or electrocautery, and detect surgical errors in real time. Because each tissue has a unique spectral signature (for example, liver looks distinct from bowel because of differences in both absorbance and in the way the tissue scatters light), automated discrimination among tissue types is feasible. Biocompatible dyes, fluorescent markers, or other contrast agents will improve accuracy and resolution for many optical techniques, similar to the way such approaches improved CT and MRI. Such dye-based optical techniques would be the in vivo equivalent of the histologic stains available to the pathologist.

Of all of the light-based techniques, OCT provides the highest resolution structural images at shallow depths, but each approach is likely to find an application in which it excels. We are now leaving the dark ages of biology and medicine.

HyperNotes Related Resources on the World Wide Web

The Optical Coherence Tomography Database, developed by the New York Eye & Ear Infirmary, describes OCT and provides OCT images of the retina.

The Oregon Medical Laser Center, which develops medical applications for light and lasers, describes optical diagnostics, including optical coherence tomography.

The Introduction to MRI, developed at the VA MRI Facility, covers the basics of magnetic resonance imaging, instrumentation, image characteristics, contrast agents, and other topics related to MRI.

MRI World Wide Web Sites is a list of Web resources for magnetic resonance imaging.

The NMR Information Server at the University of Florida provides information on magnetic resonance imaging and nuclear magnetic resonance and links to related Web resources.

Dalhousie University Division of Oral and Maxillofacial Radiology maintains Magnetic Resonance Imaging Resources on the Internet.

The Basics of MRI, by Joseph P. Hornak, describes MRI and NMR, the mathematics and physics of NMR, and imaging principles. A glossary of terms is included.

The Whole Brain Atlas provides striking interactive magnetic resonance images of the human brain. A Neuroimaging Primer describes computerized tomography and MRI.

ULTRANET: Ultrasound Technology Graphic Hot Links & More, maintained by Mike Davis, is a list of Web sites for ultrasonic imaging and includes links to a large number of images.

The Medical Ultrasound Imaging WWW Directory is a list of links and e-mail addresses for people, organizations, and companies involved in medical ultrasound imaging. It offers General Ultrasound Information, which includes Ultrasound Basics, a patient information page developed by Harvard University and Beth Israel Hospital. Also included is a link to What is Ultrasound?, a brief description of ultrasonic imaging.

Medical Imaging Resources on the Internet, maintained by the Centre of Medical Imaging Research (CoMIR) at the University of Leeds, offers links to Web sites for magnetic resonance imaging, computerized tomography, ultrasonic imaging, and other diagnostic imaging technologies.

The RSNA Launch Pad offers a list of links to other Web sites of interest to members of the Radiological Society of North America. This list includes resources for medical imaging.

The World Wide Web Virtual Library: Microscopy is a comprehensive list of Web resources for microscopy.

1. David A. Benaron's research is described in the Faculty Research Directory in Biomedical & Biological Sciences at Stanford University.

2. Science Hypermedia provides a diagram and description of the Michelson interferometer.

3. Macalester College offers a brief description of the Michelson interferometer.

4. The Mt. Wilson Observatory Web site describes the application of the Michelson interferometer in astronomy.

5. The Principles of Confocal Microscopy are described and animated diagrams are included in the Integrated Microscopy Resource Web site.

6. Stanford University

7. Stanford University School of Medicine

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article