1.

Introduction

The near-infrared (NIR) optical properties of tissue, namely absorption and reduced scattering coefficients, can provide information on a variety of tissue physiology processes. Wavelength-dependent absorption is widely used to quantify the concentration of biologically important chromophores, such as hemoglobin, water, and lipid. To date, scattering properties are not fully understood. Basically, the wavelength dependence of scattering is associated with tissue structure.

In the past decades, increasing use of light either as a diagnostic or therapeutic tool has demanded a better knowledge of the optical properties of human tissues.^{1 2 3 4} The properties of the head, especially the neonatal brain, are not particularly well known. In recent years, great effort has been devoted to evaluating the optical properties of brain tissue.

Using integrating-sphere measurements, van der Zee, Essenpreis, and Delpy⁵ determined the optical properties of postmortem neonate and adult human brain tissue over the wavelength range of 500 to 1000 nm. Yaroslavsky et al.⁶ reported the optical properties of brain tissue in vitro from 360 to 1100 nm. However, due to significant alternation that may occur after the death of the organism, to obtain reliable information, it is important to estimate the optical properties in vivo.

The time-resolved method has proven to be a valuable tool to assess tissue optical properties in vivo noninvasively. Recently, in vivo optical properties of piglet brain⁷ and adult forehead^{8 9} were measured by time-resolved systems. Hebden et al.¹⁰ obtained the first 3-D images of a premature infant brain using a 32-channel time-resolved imaging system. However, these measurements were obtained with a complex time-resolved system not suitable for clinical use. A frequency-domain apparatus, on the contrary, is relatively inexpensive and can easily be made portable, which are two important considerations for a clinically useful instrument.

Fantini et al.¹¹ and Hueber et al.¹² reported the optical properties of newborn piglet brain using a frequency-domain multidistance spectroscopy. Bevilacqua et al.¹³ measured the local optical properties of normal and malignant human brain tissue in vivo during brain surgery with short source-detector separations. More recently, Choi et al.¹⁴ published the measurements of absorption and reduced scattering coefficients through the forehead on 30 adult volunteers using a multidistance frequency-domain method. Up to now, however, published data on in vivo optical properties of neonatal brain in large populations, to the best of our knowledge, is relatively scarce.

Recently, we have developed a portable, multiwavelength, frequency-domain near-infrared spectroscopy (NIRS) instrument for quantitative, noninvasive measurement of tissue optical properties.^{15 16} A calibration procedure was employed to quantify the absorption and reduced scattering coefficients with single source-detector separation. This instrument has been validated by extensive tissue equivalent phantom and animal experimental studies.^{17 18} We investigated the optical properties of 23 infant heads using our frequency-domain NIRS instrument. The physiological parameters were also derived from the absorption coefficients and compared with those available in the literature.

2.

Theory and Method

3.

Instrumentation and Calibration

3.1.

Frequency-Domain NIRS Instrument

This study has been performed with a homodyne frequency-domain NIRS instrument (see Fig. 1), which was described in detail elsewhere.^{15 16} Briefly, this frequency-domain NIRS instrument uses three laser diodes at 788-, 814-, and 832-nm wavelength. The three wavelengths are time multiplexed at a rate of 1 Hz. The light is modulated at 140 MHz and directed to the forehead of neonates by a 1-mm-diam fiber guide. The diffused light is collected to the photomultiplier tube (PMT, Hamamatsu, H7683-02) by another 3-mm-diam fiber bundle, placed several centimeters apart from the source fiber. The amplitude and phase shift of the detected photon density wave can be further derived from the output of the in-phase and quadrature demodulator (Mini-Circuit, MIQY-140D).

Eq. (7)

$$\theta ={\text{tan}}^{-1}\frac{Q_{dc}}{I_{dc}},$$

Eq. (8)

$$A={\left(I_{dc}^2+Q_{dc}^2\right)}^{1/2}.$$

Figure 1

Schematic diagram of the frequency-domain NIRS instrument. OSC is the oscillator; PMT is the photomultiplier tube; LNA is the low noise amplifier; BPF is the bandpass filter; I and Q are the in phase and quadrature phase; LPF is the low pass filter; D/A is the digital-to-analog converter; and A/D is the analog-to-digital converter.

4.

Experimental Protocol

23 infants were enrolled in this study. They had a median gestional age of 39 (range 35 to 42.5) weeks, a body weight of 3200 (2400 to 4600) g, and a postnatal age of 5 (1 to 17) days. Two of them were preterm infants with a gestional age of 35 and 36 weeks, respectively. The subjects were admitted due some diseases but were all in recovery at the time of study and had normal cranial ultrasound scans: mild jaundice (n=12), pneumonia (n=3), hypoglycemia (n=2), and healthy infants (n=6). Informed parental consent was obtained before each investigation.

The subjects were in a supine, comfortable position during the entire experiment. Measurements were performed on each subject by gently placing the probe on the forehead. Just enough pressure was applied to ensure optical contact between skin and probe. At least two measurements were performed successively at a given location to exclude the problem of optical coupling. Sometimes we changed the position a little bit for better contact.

The source-detector separation of the probe was fixed at 4 cm, which means the frequency-domain NIRS instrument views a banana-shaped tissue volume between the emitter and detector located approximately 1 to 3 cm beneath the skin.^{3 9} With the probe on the forehead, this instrument appears to monitor both gray and white matter in the frontal neocortex.

The calibration procedures were performed right before or after the measurements. The top and bottom sides of the cylindrical silicon resin phantom were covered with aluminum foil to simulate the semi-infinite boundary condition.

5.

Results and Discussion

5.1.

Optical Properties

Figure 2 shows the histograms of the fitted absorption coefficient μ_a and reduced scattering coefficient μ_s ^′ in 23 infant foreheads at wavelengths of 788 and 832 nm, respectively. The vertical axis in each histogram is the number of subjects. Mean and standard deviation of the fitted optical properties are listed in Table 1.

Figure 2

Histograms of the fitted μ_a and μ_s ^′ in 23 infant foreheads at wavelengths of 788 and 832 nm, respectively. The vertical axis in each histogram is the number of subjects. The source-detector separation is 4 cm.

Table 1

Average optical properties, physiological parameters, and standard deviation of brain in 23 infants.
λ (nm)	788	832
μa (cm −1)	0.078±0.014	0.089±0.019
μs ′ (cm −1)	9.16±1.22	8.42±1.23
[Hb] (μM)	16.0±5.1
[HbO 2] (μM)	23.7±8.4
THC (μM)	39.7±9.8
StO 2(%)	58.7±11.2

Note that the absorption coefficients of the infant foreheads are lower than the values reported in adults.^{8 9 14} This difference may partly be due to the brain itself, and partly due to the influence of the superficial layer. NIRS monitors the tissue beneath the optical fiber, which includes scalp (skin, subcutaneous fat, muscle), skull, cerebrospinal fluid (CSF), and the brain. The measurement of an adult head may be contaminated by these layered structures. In infants, however, the thin extracerebral tissues (<5 mm) do not significantly contaminate the measurement of optical properties from the surface of the head.³⁰

The broadly distributed μ_s ^′ in Fig. 2 may be due to the large intersubject variation. In other words, the variation of μ_s ^′ should be taken into account when comparing data among subjects. However, the current commercially available continuous-wave NIRS cannot separate the effect of scattering from the attenuation of light and thus simply assumes a constant μ_s ^′. Frequency-domain NIRS allows separation of μ_a and μ_s ^′, which make it possible to provide absolute concentration of Hb and HbO ₂.

5.2.

Physiological Parameters

Histograms of the physiological properties extracted from μ_a values are summarized in Fig. 3. Note that we have ignored the background absorption and used absorption coefficients at two wavelengths to derive the hemoglobin concentration. Despite the simplistic assumptions involved, the physiological properties derived from the absorption coefficients (Table 1) agree reasonably well with the results available in the literature.

Figure 3

Histograms of [Hb], [HbO ₂], total hemoglobin concentration (THC), and tissue oxygen saturation (StO ₂) derived from absorption coefficients. The vertical axis in each histogram is the number of subjects. The source-detector separation is 4 cm.

Cooper et al.³¹ measured 19 newborn infants with second differential spectroscopy. Their preliminary results gave a mean deoxyhemoglobin concentration of 14.6±4.0 μM, which is consistent with our result (16.0±5.1 μM). NIRS measurements of absolute cerebral hemoglobin concentrations in neonates by Wolf et al.³² found an average tissue oxygen saturation (StO ₂) of about 66±4%, while we obtain a somewhat lower value of 58.7±11.2%. Wyatt et al.³³ reported the mean cerebral blood volume of the infant brain (2.2±0.4 ml per 100 g of tissue). This corresponds to an estimated total hemoglobin concentration (THC) of 36±11 μM of tissue, which agrees well with our average of 39.7±9.8 μM.

6.

Discussion

The main clinical interest in NIRS was initially for monitoring brain oxygenation in newborn babies. NIRS has been used for perinatal application since 1985, and hundreds of studies have been performed.^{34 35} Used in a research context, NIRS has improved understanding of cerebral circulation. However, NIRS is used infrequently to guide clinical care because of quantitation problems related to the biophysics of measurement.

The improved NIRS technologies, such as time- and frequency-resolved NIRS enables experimenters to separate tissue scattering effects from tissue absorption effects, and thus extract more quantitative information about tissue chromophores. In this work, a novel frequency-domain method was employed to quantify the optical properties of infant brains. However, several points should be considered during the interpretation of the experimental data.

First, the measurements were obtained with a single source-detector separation (4 cm). The NIRS instrument views a banana-shaped tissue volume including the scalp, skull, cerebrospinal fluid (CSF), and the brain. In newborn infants, the thin extracerebral tissues make it possible to detect more information from both gray and white matter in the frontal neocortex.³⁰ In adults, however, the interference from the thicker extracranial tissues cannot be ignored.³⁶ To obtain depth information on different layers, the multiple-distance method could be introduced in adult brain applications.^{14 37}

Secondly, a simplified model of light propagation (homogeneity and semi-infinite boundary condition) was employed to recover the optical properties of tissue from the experimental data. The neonatal head, however, was both highly inhomogeneous and curved. The difference between the actual anatomy and model may induce systematic errors. Analytical solutions of the diffusion equation for more accurate sphere and cylinder geometries are also available in the literature.^{25 38} Unfortunately, they are time consuming and not suitable for real-time application.

Finally, we assumed that oxy- and deoxyhemoglobin are the primary chromophores in the brain when deriving the physiological properties from μ_a at two wavelengths (788 and 832 nm). In brain tissue, hemoglobin, water, and cytochrome-c oxidase are known chromophores in the near-infrared region.²⁷ The accuracy of quantitative hemoglobin concentration may be hindered by ignoring the nonhemoglobin absorption.

There are also some practical limitations to obtaining the quantitative measurements of infant brains, which have not been discussed here. The main concern of this work is to investigate the potential of this novel frequency-domain NIRS in pediatric applications. The baseline of the optical and physiological properties of infant brains was obtained. Work is in progress to improve the instrument and to design more complex measurement protocols for assessment of infant brain injury and neurodevelopment.

7.

Conclusions

We present determination of in vivo optical properties of 23 neonatal foreheads using a portable frequency-domain NIRS instrument. A calibration phantom of known optical properties is employed to quantify the absorption coefficient and reduced scattering coefficient with a single source-detector separation. The ability to separate tissue scattering effects from tissue absorption effects makes it possible to obtain quantitative measurements. Physiological parameters derived from the measured absorption coefficients are consistent with those in the literature. The factors affecting the interpretation of the experimental data are also discussed, which suggests considerable work remains to assess the utility of this approach in clinical practice.