Non-invasive neuroimaging using near-infrared light

Abstract

This article reviews diffuse optical brain imaging, a technique that employs near-infrared light to non-invasively probe the brain for changes in parameters relating to brain function. We describe the general methodology, including types of measurements and instrumentation (including the tradeoffs inherent in the various instrument components), and the basic theory required to interpret the recorded data. A brief review of diffuse optical applications is included, with an emphasis on research that has been done with psychiatric populations. Finally, we discuss some practical issues and limitations that are relevant when conducting diffuse optical experiments. We find that, while diffuse optics can provide substantial advantages to the psychiatric researcher relative to the alternative brain imaging methods, the method remains substantially underutilized in this field.

Introduction

Several non- or minimally-invasive neuromonitoring techniques for examining functional brain activity are currently available to the psychiatric researcher and clinician. Historically, electroencephalography (EEG) was the first to appear on the scene, followed by other technologies including positron emission tomography (PET) and single-positron emission computed tomography (SPECT), magnetoencephalography (MEG), and most recently functional magnetic resonance imaging (fMRI). These methods are often categorized in terms of whether they provide direct or indirect information about brain function. Direct methods include MEG, EEG, and event-triggered EEG (also called event related potentials; ERPs), each of which monitors a direct consequence of brain electromagnetic activity. In particular, EEG and ERP record the electrical fields generated by neuronal activity, while MEG records the magnetic fields induced by such activity. PET, SPECT and fMRI, on the other hand, are indirect methods in that they generally monitor hemodynamic changes consequent to brain electrical activity. PET and SPECT brain imaging operate by monitoring the decay of blood-borne radioactive isotopes as they pass through the brain. FMRI, in contrast, detects changes in the local concentration of deoxyhemoglobin via its effect on imposed magnetic fields. While each of these techniques has its own distinct advantages, at present the direct methods tend to have limited spatial resolution, whereas the indirect methods can only detect neuronal activity after it has been filtered by a complex and poorly-understood neurovascular coupling function.

A lesser-known technology for monitoring brain function capitalizes on the absorption and scattering properties of near-infrared light to provide information about brain activity. It was long thought that the scattering of light by tissue made it impossible to recover information from anything but the most superficial layers of tissue (e.g., microscopy). Indeed, most long-standing optical techniques in use in the neurosciences are for superficial tissues only (Villringer and Chance 1997). Some 25 years ago, however, it was discovered that useful information could be obtained from thick tissue samples, including brain monitoring using light applied to and detected from the scalp (Jobsis 1977). This finding spurred the development of diffuse optics as a technique for human brain monitoring. The technique goes variously by the names of near-infrared spectroscopy (NIRS), diffuse optical tomography (or topography; DOT) and/or near-infrared imaging (NIRI). All of the techniques are based on essentially the same concept—shine light onto the scalp, detect it as it exits the head, and use the absorption spectra of the light absorbing molecules (chromophores) present in tissue to interpret the detected light levels as changes in chromophore concentrations.

To compare the spatial and temporal sensitivity of these various brain imaging techniques, we present Figure 1, modeled after a similar figure by (Churchland and Sejnowski 1988). It is clear that MEG and ERPs are strong in temporal sensitivity but relatively weak in terms of spatial sensitivity. In contrast, fMRI, PET and SPECT are stronger in spatial sensitivity but weak in terms of temporal resolution. Diffuse optical techniques, in comparison, can provide excellent temporal sensitivity as well as reasonable spatial sensitivity. When multiple colors of light are used, moreover, spectroscopic information about the sampled tissue also becomes available, thereby affording the promise of quantifying the concentrations of the various hemoglobin species-oxyhemoglobin, deoxyhemoglobin, and the sum of these (total hemoglobin, which is proportional to blood volume).

In addition to hemoglobin-based measures of brain activity, diffuse optical techniques also make non-hemoglobin-based measures feasible. For example, by recording data from several wavelengths simultaneously, one can measure other tissue chromophores, including cytochrome oxidase. As a marker of metabolic demands, cytochrome oxidase measurements can provide more direct information about neuronal activity than hemoglobin changes Heekeren et al 1999, Jobsis et al 1977. There is also evidence to suggest that diffuse optical methods can detect cell swelling that occurs in the 50–200 milliseconds following neuronal firing, which would be an even more direct measure of neuronal activity than the hemodynamic or metabolic markers Gratton and Fabiani 2001, Steinbrink et al 2000, Stepnoski et al 1991. This type of “fast” signal appears to be significantly smaller than the hemodynamic signals (on the order of a .01% signal change). With sufficiently fast and sensitive electronics, however, such signals could be feasibly recorded by the same equipment as the hemodynamic signals. (The spatial correspondences between these various signals remain to be investigated.) Thus, diffuse optical techniques may be simultaneously capable of providing both indirect and more direct methods of neuronal activity monitoring-complementary sources of information about brain function.

The primary advantages of the optical approach, however, lie in areas not explicitly represented by Figure 1. In particular, the instrumentation—which is completely non-invasive—can be made portable, unobtrusive, low-cost, low-power, and can even be made robust to motion artifacts (e.g., Totaro et al 1998). For the psychiatric researcher, these additional strengths can bring otherwise previously unthinkable projects into the realm of possibility. For example, with proper fiber coupling, extensive movement can be tolerated, opening up the possibility of studying infants, small children, patients with severe movement disorders, or other highly animated subjects, without sedation. The portability and near-zero run-time cost of the instrument affords bedside (or home) monitoring for extended periods, which could be useful for monitoring the effects of slowly acting drugs, or slowly evolving pathologies. And, the fact that near infrared light is non-ionizing means that there is no limit to the number of scans one can undergo.

Diffuse optical recordings depend on two critical characteristics of the electromagnetic spectrum as it interacts with biological tissue. First, while biological tissue is relatively opaque to visible light, it is not totally opaque, as demonstrated by a simple experiment. Darken a room, shine a flashlight through your hand and notice that the white light from the flashlight exits with a red hue, indicating that the tissue least absorbs the redder wavelengths. Near-infrared (NIR) light—from approximately 650–950 nm—is even more weakly absorbed by tissue than the red wavelengths (Figure 2). As a result, this range of wavelengths is often called an “optical window” into biological tissue. This property allows light of these wavelengths to penetrate several centimeters through tissue and still be detected.

The second critical characteristic of NIR light as it interacts with biological tissue is also apparent in Figure 2. The two dominant chromophores for the NIR wavelength range just happen to be two biologically relevant markers for brain activity: oxyhemoglobin (HbO₂) and deoxyhemoglobin (HbR). Thus, NIR wavelengths pass relatively easily through tissue, and their absorption can provide information relevant to brain function.

Having described what makes the technique possible, we next discuss the methodology details, including the types of measurements one can make, the nature of diffuse optical monitoring equipment, and the basic underlying theory necessary for data interpretation. We will then review brain research applications of diffuse optical techniques—with an emphasis on psychiatric patient groups—and will finally discuss the practical issues presently faced by optical researchers.