Diagnostic imaging with light

This paper reviews the evolution of optical imaging in diagnostic radiology and examines recent progress. Although the idea has been around for many decades, interest in the development of an effective method has never been so great. Optical imaging presents several potential advantages over existing radiological techniques. First, the radiation is non-ionizing and therefore reasonable doses can be repeatedly employed without harm to the patient. Second, optical methods offer the potential to differentiate between soft tissues with different optical absorption or scatter, but which are indistinguishable using other modalities. And third, specific absorption by natural chromophores (such as haemoglobin) allows functional information to be obtained. Principal clinical applications include a means of detecting breast disease and a cerebral imaging modality for mapping oxygenation and haemodynamics in the brain of newborn infants or cortical functional activity in adults. Past attempts to image tissues with light have been severely restricted by the overwhelming scatter which occurs when optical radiation spreads through tissue: however, recent innovations in technology have suggested once again that it may be a practical possibility.

Imaging very-low-contrast objects in breastlike scattering media with a time-resolved method

An investigation was performed of the effectiveness of a time-resolved method for imaging very-low-contrast features embedded in highly scattering media. Experiments employed slabs of breastlike material into which were inserted small cylindrical objects having either a scattering or an absorption coefficient of 4, 2, 1.5, and 1.1 times greater than the surrounding medium. An attempt was made to quantify the degree of contrast produced by each object. The results indicate that time-gating is far more effective at enhancing the contrast of the scattering inhomogeneities than of the absorbing inhomogeneities. This observation is shown to agree with a diffusion-based model, which also predicts that time-gating can decrease the contrast of absorbing inhomogeneities unless very short time-gates can be employed.

Parallel operation of Monte Carlo simulations on a diverse network of computers

Monte Carlo simulation methods are frequently used to determine light propagation in tissue and x-ray propagation as well as for solving other non-medically related problems. Such techniques are computationally slow, with the signal to noise ratio improving only as the square root of computation time. We present a method for the design of a Monte Carlo program that is capable of running on up to 24 computers simultaneously, with there being very few restrictions on the computer types as long as they run on a common network. This parallel operation is useful when the run time is expected to be long. A mixture of PCs and Sun workstations have been successfully used. The program as described was designed for the simulation of light transport in tissue, but the technique of achieving simple simultaneous execution on a number of different computers could be used wherever Monte Carlo techniques are used.

Evaluation of the temporally extrapolated absorbance method for dual-wavelength imaging through tissuelike scattering media

An independent assessment is described of a dual-wavelength imaging technique, known as the temporally extrapolated absorbance method (TEAM), proposed by Yamada et al. [Opt. Eng. 32, 634-641 (1993)]. The technique involves recording the temporal distribution of light transmitted across a scattering medium at two carefully chosen wavelengths at which the scattering properties of the medium are assumed to be identical. The objective is to image internal structure that absorbs more strongly at one wavelength than it does at the other. A simple theoretical treatment of TEAM is presented that employs a perturbation model of photon transport. This indicates that despite the lack of a secure theoretical basis, the technique may provide a potentially effective ad hoc method of generating images of highly scattering media. The method was also evaluated experimentally by using, for the first time to our knowledge, a single object and two wavelengths. A single-projection, two-dimensional image was obtained of a solid phantom with optical properties representative of breast tissue. The results exhibited good agreement with the theoretical model, and a small embedded feature that absorbs 3.5 times as strongly as the surrounding medium at one wavelength was revealed successfully.

Optical imaging in medicine: I. Experimental techniques

The overwhelming scatter which occurs when optical radiation propagates through tissue severely limits the ability to image internal structure using measurements of transmitted intensity. A broad range of methods has been proposed during the past decade or so in order to improve imaging performance. Direct methods involve isolating an unscattered or least-scattered component of transmitted scattered light. Indirect methods generally involve measuring some characteristic of the temporal distribution of transmitted light, or an equivalent in the frequency domain, and obtaining a computational solution to the inverse problem. In this paper, we review the experimental techniques which have been proposed in order to explore both direct and indirect imaging. The relative merits and limitations of the various experimental methods are discussed, and we consider the future directions and likelihood of success of optical imaging in medicine.

Investigation of the effect of discrete absorbers upon the measurement of blood volume with near-infrared spectroscopy

This paper derives an analytical model for investigating the effect of the distribution of absorbers upon light attenuation in a scattering medium. Results from this are found to agree with those of Monte Carlo simulations. The implications of this model are then examined for their likely effect upon the measurement of cerebral blood volume (CBV) using near-infrared (NIR) spectroscopy. We conclude that, given the small diameter of the majority of cerebral blood vessels, the distribution of the blood will have little effect upon the measurement of CBV. Where changes to the blood volume occur in the larger (> 0.2 mm diameter) vessels on the surface of the brain, NIR spectroscopy is likely to underestimate the change.

Evaluation of spatial resolution as a function of thickness for time-resolved optical imaging of highly scattering media

Previous experimental and theoretical investigations of the utility of time-resolved methods as a means of optical imaging through the human breast have indicated that a spatial resolution of approximately 1 cm is achievable by isolating the shortest path length photons which propagate through the tissue. Studies have also shown that resolution may be improved further by extrapolating the measured distribution using an appropriate model of photon transport. The experiments described here were performed in order to observe the relationship between achievable spatial resolution and the thickness of the medium. For a given time gate, an improvement in the spatial resolution was observed as the object thickness was reduced. Overall, the results indicate that a breast compression of about 1 cm may improve the limiting spatial resolution by as much as 7 mm. Less encouraging is the implication that temporal extrapolation over several orders of magnitude in intensity is required to achieve a comparable improvement in spatial resolution.

Theoretical and experimental investigation of near-infrared light propagation in a model of the adult head

Near-infrared light propagation in various models of the adult head is analyzed by both time-of-flight measurements and mathematical prediction. The models consist of three- or four-layered slabs, the latter incorporating a clear cerebrospinal fluid (CSF) layer. The most sophisticated model also incorporates slots that imitate sulci on the brain surface. For each model, the experimentally measured mean optical path length as a function of source-detector spacing agrees well with predictions from either a Monte Carlo model or a finite-element method based on diffusion theory or a hybrid radiosity-diffusion theory. Light propagation in the adult head is shown to be highly affected by the presence of the clear CSF layer, and both the optical path length and the spatial sensitivity profile of the models with a CSF layer are quite different from those without the CSF layer. However, the geometry of the sulci and the boundary between the gray and the white matter have little effect on the detected light distribution.

In vivo measurements of the wavelength dependence of tissue-scattering coefficients between 760 and 900 nm measured with time-resolved spectroscopy

We present in vivo values for the optical transport coefficients (mu(a), mu(s)?) of the adult human forearm, calf, and head from 760 to 900 nm measured with time-resolved spectroscopy. The accuracy of the method is tested with tissue-simulating phantoms. We obtain mu(s)?(lambda) approximately 1.1 - (5.1 x 10(-4) lambda) mm(-1) (forearm), 1.6 - (8.9 x 10(-4) lambda) mm(-1) (calf), and 1.45 - (6.5 x 10(-4) lambda) mm(-1) (head), where lambda is measured in nanometers. At 800 nm we obtain mu(a) = 0.023 +/- 0.004 mm(-1) (forearm), 0.017 +/- 0.005 mm(-1) (calf), and 0.016 +/- 0.001 mm(-1) (head). Our values differ substantially from published in vitro data. In particular, our transport coefficients for the adult head are substantially lower than previously reported values for adult human cerebral matter and pig skull cortical bone measured in vitro.

The effect of scalp ischaemia on measurement of cerebral blood volume by near-infrared spectroscopy

Near-infrared spectroscopy (NIRS) is a noninvasive method of quantifying changes in cerebral haemodynamics from changes in the absorption of near-infrared light by oxyhaemoglobin and deoxyhaemoglobin. Measurement of neonatal cerebral blood volume (CBV) by NIRS was described in 1990 but it has been suggested that, in adults, scalp and skull blood content contribute a significant amount to the cerebral haemodynamic variables quantifiable by NIRS. To investigate this, CBV was measured in nine adult subjects, in the frontal region of the head, before and after inflating a pneumatic tourniquet proximal to the measurement site. Because a change in scalp blood content could potentially alter the pathlength of light passing through the head and hence affect the measured CBV, the optical pathlength factor was therefore also measured before and after tourniquet inflation. Blood flow occlusion was confirmed by laser Doppler velocimetry. The results showed that tourniquet inflation had no effect on the estimated value of CBV or the differential pathlength factor. We conclude that, provided the distance between light entry and exit on the surface of the scalp is sufficiently large, changes in scalp blood flow have no effect on NIRS measurement of cerebral haemodynamics.

Guest editorial: special section on near infrared spectroscopy and imaging of tissues

This Special Section Guest Editorial provides an overview of the topical area and an introduction to the articles featured in the special section.

Experimental validation of Monte Carlo and finite-element methods for the estimation of the optical path length in inhomogeneous tissue

To validate models of light propagation in biological tissue, experiments to measure the mean time of flight have been carried out on several solid cylindrical layered phantoms. The optical properties of the inner cylinders of the phantoms were close to those of adult brain white matter, whereas a range of scattering or absorption coefficients was chosen for the outer layer. Experimental results for the mean optical path length have been compared with the predictions of both an exact Monte Carlo (MC) model and a diffusion equation, with two differing boundary conditions implemented in a finite-element method (PEM). The MC and experimental results are in good agreement despite poor statistics for large fiber spacings, whereas good agreement with the FEM prediction requires a careful choice of proper boundary conditions.

Optical properties of multicellular tumour spheroids

Multicellular tumour spheroids grown in vitro have been widely used in cancer research as an experimental preparation with many of the characteristics of tumours. They provide a model system for understanding the optical behaviour of tumour tissue, which is of interest in novel diagnostic and therapeutic procedures. Optical measurements on fresh spheroids in the wavelength range of 600-1000 nm yielded scattering coefficients, absorption coefficients and g values (mean cosine of scatter) of 160-90 mm-1, 0.1-0.4 mm-1 and 0.99, respectively. Following fixation, considerably higher values of scattering and absorption coefficients were seen. The values are compared with those reported elsewhere for excised tumour tissue and interpretations of the optical behaviour are suggested.

Measurement of cranial optical path length as a function of age using phase resolved near infrared spectroscopy

Near infrared spectroscopy (NIRS) has been used to measure concentration changes of cerebral hemoglobin and cytochrome in neonates, children, and adults, to study cerebral oxygenation and hemodynamics. To derive quantitative concentration changes from measurements of light attenuation, the optical path length must be known. This is obtained by multiplying the source/ detector separation by a laboratory measured differential path length factor (DPF) which accounts for the increased distance traveled by light due to scattering. DPF has been measured by time of flight techniques on small populations of adults and postmortem infants. The values for adults are greater than those for newborns, and it is not clear how to interpolate the present data for studies on children. Recent developments in instrumentation using phase resolved spectroscopy techniques have produced a bedside unit which can measure optical path length on any subject. We have developed an intensity modulated optical spectrometer which measures path length at four wavelengths. Two hundred and eighty three subjects from 1 d of age to 50 y were studied. Measurements were made at a fixed frequency of 200 MHz and a source detector separation of 4.5 cm. Results suggest a slowly varying age dependence of DPF, following the relation DPF690 = 5.38 + 0.049A0.877, DPF744 = 5.11 + 0.106A0.723, DPF807 = 4.99 + 0.067A0.814, and DPF832 = 4.67 + 0.062A0.819, where DPF690 is the DPF measured at 690 nm and A is age is expressed in years from full term. There was a wide scatter of values, however, implying that ideally DPF should be measured at the time of each study.

Measurement of tissue temporal point spread function (TPSF) by use of a gain-modulated avalanche photodiode detector

This paper describes an opto-electronic cross-correlator designed for measurement of the temporal point spread function (TPSF) of light at the bedside. Ultra-short (< 2 ps) pulses of light from a mode-locked laser were used to illuminate a tissue phantom. The light exiting from the tissue phantom was coupled by an optical fibre to a small-area (200 microns diameter, Hamamatsu S2381) avalanche photodiode (APD). The gain of the photodiode was modulated at the repetition rate of the pulsed laser (82 MHz). Usually the gain was approximately 100, but for a period of approximately 130 ps (FWHM) the gain was increased to approximately 105. A lock-in amplifier, which sampled the integrated APD current, gave an output proportional to the difference between the current in the low- and high-gain states. Hence a small section of the TPSF was selectively sampled. An overall temporal resolution of 275 ps FWHM was achieved. As the timing of the gain modulation was controlled by an all-electronic variable-time-delay system, the whole of the TPSF could be sampled without requiring any moving prism or mirror which is typical of many cross-correlators. Hence the system is mechanically very rugged, which enhances its durability in a portable instrument.

Near-infrared spectroscopy: theory and applications

In conclusion, NIRS appears to offer both a new monitoring modality and new information about cerebral oxygenation. Technical problems in the application of this technology persist, most notably determination of pathlength and the volume of tissue interrogated. Those familiar with the history of pulse oximetry will recall that although Millikan developed an ear oximeter in 1947, it was not until Aoyagi combined recognition of the pulse signal with spectroscopy in the 1970s that oximetry was transformed into a clinically applicable monitor. In much the same way, NIRS may find the same tremendous usefulness as a noninvasive monitor of cerebral oxygen utilization, pending resolution of the remaining technical problems.

An investigation of light transport through scattering bodies with non-scattering regions

Near-infra-red (NIR) spectroscopy is increasingly being used for monitoring cerebral oxygenation and haemodynamics. One current concern is the effect of the clear cerebrospinal fluid upon the distribution of light in the head. There are difficulties in modelling clear layers in scattering systems. The Monte Carlo model should handle clear regions accurately, but is too slow to be used for realistic geometries. The diffusion equation can be solved quickly for realistic geometries, but is only valid in scattering regions. In this paper we describe experiments carried out on a solid slab phantom to investigate the effect of clear regions. The experimental results were compared with the different models of light propagation. We found that the presence of a clear layer had a significant effect upon the light distribution, which was modelled correctly by Monte Carlo techniques, but not by diffusion theory. A novel approach to calculating the light transport was developed, using diffusion theory to analyze the scattering regions combined with a radiosity approach to analyze the propagation through the clear region. Results from this approach were found to agree with both the Monte Carlo and experimental data.

Influence of respiration and changes in expiratory pressure on cerebral haemoglobin concentration measured by near infrared spectroscopy

Near infrared spectroscopy (NIRS) was used to measure the changes in concentration of cerebral oxy- and deoxygenated haemoglobin ([HbO2] and [Hb]) in six healthy adult volunteers spontaneously breathing against increased expiratory pressures (IEPs) between 0 and 20 cm H2O. During expiration, an increase in [HbO2] was recorded, accompanied by a smaller decrease in [Hb], producing a small increase in total cerebral haemoglobin concentration ([Hbsum]). The mean plus/minus SD change in [Hbsum] at the maximum 1EP of 20 cm H2O was 1.2 +/- 0.7 micromol L-1 (equivalent to 1.4%). Changes in [Hbsum] correlated with IEP level (r = 0.95) and changes in MABP (r = 0.96). The results suggest that homeostatic mechanism do not maintain cerebral blood volume or flow constant over the period of a single breath in normal adults.

Fetal heart rate changes and cerebral oxygenation measured by near-infrared spectroscopy during the first stage of labour

OBJECTIVE

To determine the relationship between contraction related changes in fetal heart rate and cerebral oxygenation measured by near-infrared spectroscopy during labour.

STUDY DESIGN

A specially designed optical probe was inserted through the dilated cervix and placed against the fetal head in 30 women during labour. Alterations in fetal heart rate during the final hour of the first stage of labour were compared with changes in the cerebral haemoglobin oxygenation index (delta oxyhaemoglobin concentration - delta deoxyhaemoglobin concentration) measured before, during and after uterine contractions.

RESULTS

Uterine contractions which were associated with either no alteration, accelerations or early decelerations of the fetal heart rate showed no significant changes in the haemoglobin oxygenation index. Variable, late and prolonged decelerations all showed significant decreases in the haemoglobin oxygenation index (P < 0.01) either during (variable) or after (variable, late and prolonged) the uterine contraction.

CONCLUSION

The association between variable, late and prolonged FHR decelerations and significant falls in cerebral oxygenation during late labour suggests that these fetal heart rate patterns are associated with an increased risk of fetal cerebral hypoxia.

Use of near infrared spectroscopy to estimate cerebral blood flow in conscious and anaesthetized adult subjects

Near infrared spectroscopy (NIRS) can be used to quantify cerebral haemodynamic states non-invasively and to estimate cerebral blood flow (CBF). In the first part of this study we have compared CBF measurements in conscious and anaesthetized subjects. In the second part we have compared paired measurements made during anaesthesia, first on the scalp and then the dura after craniotomy. Mean CBF was 17 (SD 7) ml 100 g-1 min-1 in the conscious subjects compared with 21 (8) ml 100 g-1 min-1 on the scalp during anaesthesia (P > 0.1). Mean CBF on the dura was 68 (21) ml 100 g-1 min-1 (P < 0.0001). Computer modelling suggests that the difference in magnitude between scalp and dura measurements of CBF is likely to be caused by the optical effect of extracerebral tissue which powerfully scatters light passing through it but does not contribute significantly to the measured CBF because it has only a small blood content itself. The results lend support to this method of estimating CBF although formal validation by comparison with an established technique is needed.