Technology and medicine have always been close partners. Diagnostic methods such as x-ray radiography, magnetic resonance imaging, ultrasound, and computed tomographies provide incredibly detailed anatomic information in demanding clinical settings. As these technologies progress and become standard-of-care, requirements for new methods based upon different technologies are emerging.
There is a pressing need for rapid, portable, and inexpensive devices that monitor tissue function noninvasively. Commercial emphasis has been placed upon rapid analysis instrumentation such as blood-gas analyzers, which recover biochemical information from drawn blood. But what if similar information could be obtained in real time without drawing blood? What if doctors could characterize suspicious lesions in mammograms without surgery, detect the effects of traumatic injury and progression to shock well before it occurs, or monitor the effects of tumor chemotherapies during treatment? Although this sounds too much like science fiction to be true, optical technologies now exist that could introduce these concepts into clinical practice within the next five years. Broad implementation of these optical diagnostic methods at the bedside would provide clinicians with critical feedback that could impact patient care. optics then and now
In the past, researchers have used optical techniques to look for shadows caused by hemoglobin absorption. In the 1920s, they worked with conventional sources and eventually progressed to laser sources in the 1980s. Initial optical breast-cancer detection devices lacked the required sensitivity and specificity to locate and classify breast lesions, however.
Today the situation is radically different. Coupled with exciting new concepts in tissue optics, telecom-driven technological advances have yielded noninvasive methods that provide completely new information from tissues. Photon diffusion techniques have painted a novel physiological portrait of the breast that may characterize lesions, chart the progression of disease, and monitor the effectiveness of anticancer therapies. Quantitative NIR optical spectroscopy provides an opportunity for revealing physiological information that is unobtainable by other radiological techniques. Instrumentation is compact and relatively economical, which suggests that the technology will be available to everyone.
Optics plays a role in other areas of medicine. Pulse oximeters are standard-of-care devices that monitor arterial hemoglobin oxygen saturation (the fraction of oxygen to total hemoglobin), usually during anesthesia. Technology of the 1960s and 1970s combined with an important clinical need made pulse oximetry broadly available. NIR spectroscopy is similarly poised for growth.
Figure 1. NIR radiation penetrates deeply and diffusively into tissues.network analysis
At the Beckman Laser Institute and Medical Clinic, we use an optical network analysis approach to extract quantitative information out of tissues using frequency-domain photon migration (FDPM).1 For example, to examine a device under test, a network analyzer sweeps over a range of source frequencies S (see figure 2). A fraction of S splits off as a reference R. The analyzer measures the electronic response of the device under test as a function of frequency by comparing the detected signal, either a reflection or a transmission, to R. The analyzer determines the amplitude and phase of the signal.
Figure 2. To analyze a device under test, a network analyzer sweeps a range of source frequencies S, splitting off a reference signal R and comparing the reflected signal A or transmission signal B to R.
The same approach can be applied to measurements of tissue absorption and scattering. In such a case, the device under test is a tissue, such as breast tissue. Instead of monitoring the electrical response, we monitor the diffuse optical response, either as a reflection or transmittance. The phase and amplitude of the diffuse optical signal fit to an appropriate mathematical model provide enough information to determine the tissue absorption and scattering properties.
What information does this analysis provide? The frequency response of the tissue is determined by the photon path length L (see sidebar). Increased tissue absorption yields a shorter L, or a reduced average time of travel through the tissue, which yields a smaller amplitude and phase shift. Tissue absorption in essence acts like a filter: High absorption weakens the high-frequency response. Scattering, on the other hand, generally increases L, so that increased scattering will decrease amplitude but increase the phase shift. This is how optical network analysis can pick out differences in L, based upon the absorption and scattering of the tissue. system design
The FDPM instrument begins and ends with a network analyzer (NA) (see figure 3).2 The analyzer provides the radio frequency (RF) current, which is combined with a separate bias current source to form an amplitude-modulated electronic signal that, in turn, modulates the intensity of a diode laser source at frequencies ranging from 50 to 1000 MHz. Thermoelectric coolers stabilize the diode temperature. Relevant diode laser wavelengths lie in the range of 660 to 1000 nm.
Figure 3. Off-the-shelf components such as a network analyzer, APD, and diode lasers yield compact and potentially economical optical diagnostic system for cancer detection.
The laser output is launched into graded-index optical fibers with 100 µm cores by either physical coupling (pigtailing) or optical coupling. Each of the individual lasers sends one optical fiber into a fully multiplexible 8 X 8 optical switch such that only one laser illuminates the tissue at a time. A computer controls both the RF switch network and the optical switch to ensure only one modulated laser signal reaches the tissue at any given moment.
A source fiber directs light from the switch to the tissue. An avalanche photodiode (APD) encased inside an RF-shielded module is placed 20 to 30 mm away and picks up the diffusely reflected signals. The APD module casing has machined attachments that hold the source fiber in place. In this configuration, the mean optical penetration depth is about 10 to 15 mm below the skin. The NA measures the amplitude and phase of the analog electronic signal from the APD. After correction for instrumental artifacts such as delays from electrical cables, the amplitude and phase are fit to a diffusive light transport model.
The FDPM instrument currently uses seven diode lasers that provide visible and NIR light at wavelengths ranging from 672 to 978 nm, allowing us to build an absorption spectrum; each wavelength provides sensitivity to different tissue chromophores. The optical power launched into the tissue ranges from 5 to 25 mW for each wavelength. Sweeping over all seven wavelengths requires 35 s. The NA acquires data in less than 3 s per wavelength, but there is a 2 s delay between each wavelength because of switching considerations. The entire system can be wheeled into a medical clinic and placed at the bedside.
The detector (Hamamatsu; Bridgewater, NJ) is an APD with a 1-mm-diameter active region, a 600 MHz cutoff frequency, and an intrinsic gain of 100. A miniature 1 GHz amplifier adds another factor of 100. The APD provides a unique combination of speed, sensitivity, and dynamic range, especially above 900 nm. In stark contrast, photomultipliers typically have much higher gain, but poorer sensitivity, especially above 900 nm, and slower response times.
There are many challenges associated with broadband electronics. Obvious problems include impedance mismatching and RF interference. Diode laser availability is another challenge, since manufacturers may stop producing certain sources. We have recently addressed this by integrating a complete steady-state spectrum (Ocean Optics; Dunedin, FL) together with our discrete laser diode measurements to produce a complete absorption spectrum from 650 to 1000 nm.3 We may extract absorption coefficients from the steady-state measurements only because FDPM provides the tissue-scattering spectrum. The additional information from the complete absorption spectrum will help quantify tissue chromophores even better than before.
The use of off-the-shelf components simplifies assembly but limits size and versatility. For example, the analyzer sweeps through 101 frequencies in as little as 21 ms, but there is a long internal delay between each sweep. The recent boon in cellular telephones could provide the impetus to miniaturize fast, high-frequency driver circuits, and perhaps even reduce our cart instrumentation to the size of a cellular telephone, capable of operating in real time.in the clinic
By testing the technique in a clinical setting, we can quantify the optical absorption differences between the breasts of premenopausal women and postmenopausal women and obtain spectra that yield quantitative concentrations of blood, water, and lipids.4 For example, the total hemoglobin concentrations are about 40 and 14 µM in the pre- and postmenopausal breast, respectively. The premenopausal breast is characterized by high blood concentration, high metabolism (low tissue saturation), high water, and low lipids, whereas the opposite is true of the postmenopausal breast. This noninvasive quantitative, physiological component analysis is unique to NIR optical methods.
We also have used the technique to characterize the absorption spectrum of a 1-cm diameter malignant tumor compared to normal breast tissue (see figure 4). This noninvasive measurement was obtained simply by placing the probe over the patient's suspicious lump and comparing the measurements to her opposite normal side. One can see distinctive quantitative differences between the spectra, which are mostly due to higher blood and water concentrations in the tumor relative to the normal tissue. Data that show the normal states of tissue eventually will lead to the quantitative classification of diseased tissue.
Figure 4. Functional differences between normal and cancerous tissue yield distinct absorption spectra, opening the door to the noninvasive identification of tumors.
None of the innovations we describe will ever jump from the research bench to the clinical bedside unless technology helps guide the way. Inexpensive point-of-care optical devices could help individualize health care and improve disease prevention, therapy, and overall quality of life without breaking the bank. oe
This work was supported by the following grants and gifts: National Institutes of Health Laser Microbeam and Medical Program, Department of Energy, Office of Naval Research, California Breast Cancer Research Program, the Chao Comprehensive Cancer Center, and the U. S. Army Medical Research and Material Command.
The authors also acknowledge the hard-working people of the Beckman Laser Institute and Medical Clinic who have contributed to this work: Andrew Berger (now with the University of Rochester, NY), Frederic Bevilacqua, Matt Brenner, John Butler, Randall Holcombe, Dorota Jakubowski, Ryan Lanning, Tuan Pham, and Natasha Shah.
1. Tromberg, BJ, Shah, N, Lanning, R et al., "Non-invasive in vivo characterization of breast tumors using photon migration spectroscopy." Neoplasia 2000; 2:1-15.
2. Pham, T, Coquoz, O, Fishkin, J, et al., "A Broad bandwidth frequency domain instrument for quantitative tissue optical spectroscopy." Rev Sci Instrum 2000; 71:1-14.
3. Bevilacqua, F, Berger, A, Cerussi, A, et al, "Broadband absorption spectroscopy in turbid media by combining frequency-domain and steady-state methods." Appl Opt; in press.
4. Cerussi, AE, Berger, AJ, Bevilacqua, F, et al. "Sources of absorption and scattering contrast for near-infrared optical mammography." Acad Radiol, in press.
Albert Cerussi, Bruce Tromberg
Albert Cerussi is a post-doctoral researcher, and Bruce Tromberg is a professor and director of the Laser Microbeam and Medical Program at the Beckman Laser Institute and Medical Clinic, University of California at Irvine, Irvine, CA.
Optics provides an inexpensive and noninvasive means to monitor blood physiology and more. For example, hold a flashlight to your finger; the red light is the only color transmitted. Hemoglobin absorbs blue and green very strongly, but it absorbs red very weakly.
The absorption of a chromophore is proportional to its concentration. Spectrophotometers quantify optical absorber concentrations in optically clear media. The strongest near-infrared (NIR) absorbers in tissues (approximately 630 to 1000 nm) are hemoglobin (in both oxygenated and reduced forms), water, and lipids.
Differences in the absorption spectra of oxygenated and reduced hemoglobins allow quantitative distinction between these two states. NIR light is inherently sensitive to tissue blood-oxygenation state without exogenous contrast agents. Low NIR tissue absorption permits photons to travel several centimeters deep into tissue (see figure 1). Optical methods can be rapid, noninvasive, portable, and inexpensive.
Complications arise from the fact that multiple light scattering in tissue distorts the photon path length (L). Absorption and scattering contribute to optical losses, and a single attenuation measurement cannot distinguish between these two loss mechanisms. Without some knowledge of L, it is impossible to determine chromophore concentrations.
The solution is to look at changes in L. The only way to determine L with fixed distance measurements is to measure the temporal behavior of a photon. Photons undergo a random walk (i.e. a diffusive process) through the tissue; hence the terms photon diffusion or photon migration are often used.
-- A.C. and B.T.