Optical imaging in medicine: I. Experimental techniques

Comparison of methods for reducing the effects of scattering in spectrophotometry

Light scattering provides a problem in optical spectroscopy as the relationship between attenuation and absorption in the presence of scattering is nonlinear. Three simple methods of reducing the effects of scattering and hence returning to an approximately linear relationship are considered in this paper, namely, extracting light that has maintained its original polarization state through subtraction of orthogonal polarization states, use of an added absorber, and spatial filtering. These can all be applied relatively easily to conventional spectrophotometers. However, there is an inevitable trade-off between the accuracy of the measurement and the signal-to-noise ratio as scattered light is rejected from the detector. It is demonstrated that polarization subtraction is the most efficient technique at selecting weakly scattered photons from a scattered light background as it enables the relationship between attenuation and absorption coefficient to become more linear while maintaining a higher number of detected photons. In practical use, the drawback of polarization subtraction over added absorber and spatial filtering methods is that a large dc background light level is maintained, which contributes to a higher shot noise. This means that when the scattering coefficient is high (micros > or = 7 mm(-1)) the added absorber method offers better performance for shot noise limited detection.

Multiplicative optical tomography of cardiac electrical activity

Cardiac electrical activity can be mapped today through the response of voltage-sensitive dyes; but poor transparency of muscle tissue has enforced shallow-depth imaging. We present a three-dimensional (3D) reconstruction method for electrical activity deep inside the myocardial wall. Our approach is nonlinear and differs substantially from standard diffusive optical tomography. It does not require matrix inversion, data regularization or a priori information concerning the original object. Opposite sides of a slab-shaped preparation are scanned in parallel by detection and illumination points with a constant vector offset between illumination and detection axes (biaxial scanning). Scanning is performed in two perpendicular directions. In each direction, a pair of 2D images is obtained under offsets of opposite signs. These two pairs are the input for a multiplicative reconstruction algorithm, whose output is a 3D image. The overall procedure was successfully tested on computer-generated sources that include points, lines and hemispheres, patterned after actual electrophysiological excitations. The algorithm is computationally efficient and stable with respect to varying noise levels in the raw data.

Time-resolved optical tomography using short-pulse laser for tumor detection

Our objective is to perform a comprehensive experimental and numerical analysis of the short-pulse laser interaction with a tissue medium with the goal of tumor-cancer diagnostics. For a short-pulse laser source, the shape of the output signal is a function of the optical properties of the medium, and hence the scattered temporal optical signal helps in understanding the medium characteristics. Initially experiments are performed on tissue phantoms embedded with inhomogeneities to optimize the time-resolved optical detection scheme. Both the temporal and the spatial profiles of the scattered reflected and transmitted optical signals are compared with the numerical modeling results obtained by solving the transient radiative transport equation using the discrete ordinates technique. Next experiments are performed on in vitro rat tissue samples to characterize the interaction of light with skin layers and to validate the time-varying optical signatures with the numerical model. The numerical modeling results and the experimental measurements are in excellent agreement for the different parameters studied. The final step is to perform in vivo imaging of anesthetized rats with tumor-promoting agents injected inside skin tissues and of an anesthetized mouse with mammary tumors to demonstrate the feasibility of the technique for detecting tumors in an animal model.

Time-gated optical system for depth-resolved functional brain imaging

We present a time-domain optical system for functional imaging of the adult head. We first describe the instrument, which is based on a Ti:Sapphire pulsed laser (wavelength 750-850 nm) and an intensified CCD camera enabling parallel detection of multiple fibers. We characterize the system in terms of sensitivity and signal-to-noise ratio, instrument response function, cross-talk, stability, and reproducibility. We then describe two applications of the instrument: the characterization of baseline optical properties of homogeneous scattering media, and functional brain imaging. For the second application, we developed a two-part probe consisting in two squares of 4 x 4 sources and 3 x 3 detectors. The laser source is time-multiplexed to define 4 states of 8 sources that can be turned on during the same camera frame while minimizing cross-talk. On the detection side, we use for each detector 7 fibers of different lengths creating an optical delay, and enabling simultaneous detection in 7 windows (by steps of 500 ps) for each detector. This multiple window detection allows depth sensitivity. The imaging probe was tested on dynamic phantoms and a preliminary result on an adult performing a motor task shows discrimination between superficial and cortical responses to the stimulus on both hemispheres.

Focusing optics of a parallel beam CCD optical tomography apparatus for 3D radiation gel dosimetry

Optical tomography of gel dosimeters is a promising and cost-effective avenue for quality control of radiotherapy treatments such as intensity-modulated radiotherapy (IMRT). Systems based on a laser coupled to a photodiode have so far shown the best results within the context of optical scanning of radiosensitive gels, but are very slow ( approximately 9 min per slice) and poorly suited to measurements that require many slices. Here, we describe a fast, three-dimensional (3D) optical computed tomography (optical-CT) apparatus, based on a broad, collimated beam, obtained from a high power LED and detected by a charged coupled detector (CCD). The main advantages of such a system are (i) an acquisition speed approximately two orders of magnitude higher than a laser-based system when 3D data are required, and (ii) a greater simplicity of design. This paper advances our previous work by introducing a new design of focusing optics, which take information from a suitably positioned focal plane and project an image onto the CCD. An analysis of the ray optics is presented, which explains the roles of telecentricity, focusing, acceptance angle and depth-of-field (DOF) in the formation of projections. A discussion of the approximation involved in measuring the line integrals required for filtered backprojection reconstruction is given. Experimental results demonstrate (i) the effect on projections of changing the position of the focal plane of the apparatus, (ii) how to measure the acceptance angle of the optics, and (iii) the ability of the new scanner to image both absorbing and scattering gel phantoms. The quality of reconstructed images is very promising and suggests that the new apparatus may be useful in a clinical setting for fast and accurate 3D dosimetry.

Optical waveguiding in suspensions of dielectric particles

An optical waveguide formed by a suspension of dielectric nanoparticles in a microchannel is described. The suspensions, chosen for their guiding and scattering properties, are silica and polystyrene particles that have diameters of 30-900 nm and are dispersed in water with volume fractions up to 10%. Changing the diameter and concentration of the particles causes the suspensions to transition from Rayleigh to Mie scattering and from single to multiple scattering. The threshold for optical guiding in a waveguide core composed of these suspensions is set by the numerical aperture of the effective refractive-index difference introduced by the suspension and not by the average interparticle distance.

Optical transillumination tomography with tolerance against refraction mismatch

Optical transillumination tomography (OT) is a laser-based imaging modality where ballistic photons are used for projection generation. Image reconstruction is therefore similar to X-ray computed tomography. This modality promises fast image acquisition, good resolution and contrast, and inexpensive instrumentation for imaging of weakly scattering objects, such as for example tissue-engineered constructs. In spite of its advantages, OT is not widely used. One reason is its sensitivity towards changes in material refractive index along the light path. Beam refraction artefacts cause areas of overestimated tissue density and blur geometric details. A spatial filter, introduced into the beam path to eliminate scattered photons, will also remove refracted photons from the projections. In the projections, zones affected by refraction can be detected by thresholding. By using algebraic reconstruction techniques (ART) in conjunction with suitable interpolation algorithms, reconstruction artefacts can be partly avoided. Reconstructions from a test image were performed. Standard filtered backprojection (FBP) showed a round mean square (RMS) deviation from the original image of 9.9. RMS deviation with refraction-tolerant ART reconstruction was 0.33 and 0.24, depending on the algorithm, compared to 0.57 (FBP) and 0.06 (ART) in a non-refracting case. In addition, modified ART reconstruction allowed detection of small geometric details that were invisible in standard reconstructions. Refraction-tolerant ART may be the key to eliminating one of the major challenges of OT.

Imaging in breast cancer: diffuse optics in breast cancer: detecting tumors in pre-menopausal women and monitoring neoadjuvant chemotherapy

Diffuse optical spectroscopy (DOS) and diffuse optical imaging (DOI) are non-invasive diagnostic techniques that employ near-infrared (NIR) light to quantitatively characterize the optical properties of centimeter-thick, multiple-scattering tissues. Although NIR was first applied to breast diaphanography more than 70 years ago, quantitative optical methods employing time- or frequency-domain 'photon migration' technologies have only recently been used for breast imaging. Because their performance is not limited by mammographic density, optical methods can provide new insight regarding tissue functional changes associated with the appearance, progression, and treatment of breast cancer, particularly for younger women and high-risk subjects who may not benefit from conventional imaging methods. This paper reviews the principles of diffuse optics and describes the development of broadband DOS for quantitatively measuring the optical and physiological properties of thick tissues. Clinical results are shown highlighting the sensitivity of diffuse optics to malignant breast tumors in 12 pre-menopausal subjects ranging in age from 30 to 39 years and a patient undergoing neoadjuvant chemotherapy for locally advanced breast cancer. Significant contrast was observed between normal and tumor regions of tissue for deoxy-hemoglobin (p = 0.005), oxy-hemoglobin (p = 0.002), water (p = 0.014), and lipids (p = 0.0003). Tissue hemoglobin saturation was not found to be a reliable parameter for distinguishing between tumor and normal tissues. Optical data were converted into a tissue optical index that decreased 50% within 1 week in response to neoadjuvant chemotherapy. These results suggest a potential role for diffuse optics as a bedside monitoring tool that could aid the development of new strategies for individualized patient care.

Lasers in cancer detection and diagnosis research: enabling characteristics with illustrative examples

The salient properties of laser light and the way light interacts with biological tissues and molecular constituents of tissues offer possibilities for detection and diagnosis of cancer. In particular, the wavelength selectivity of tunable lasers, narrow bandwidth around the selected wavelength, and spectral brightness enable probing of key molecular constituents of tissues, and endow laser-based techniques with much desired diagnostic potential. This article presents an overview of some recent developments in optical imaging and optical biopsy of different types of cancers, and illustrates the diagnostic role of the color of light.

Spectral and temporal near-infrared imaging of ex vivo cancerous and normal human breast tissues

Cancerous and normal ex vivo human breast tissues were investigated using spectroscopic and time-sliced two-dimensional (2-D) transillumination imaging methods in order to demonstrate the importance and potential of spectral and temporal measurements in breast cancer detection and diagnosis. The experimental arrangement for time-sliced optical imaging used 120 fs, 1 kHz repetition-rate, 800 nm light pulses from a Ti:sapphire laser system for sample illumination, and a 80 ps resolution ultrafast gated intensified camera system for recording 2-D time-sliced images. The spectroscopic imaging arrangement used 1225-1300 nm tunable output of a Cr: forsterite laser for sample illumination, a Fourier space gate to discriminate against multiple-scattered light, and a near-infrared area camera to record 2-D images. Images recorded with earlier temporal slices of transmitted light highlighted tumors, while those recorded with later slices accentuated normal tissues. When light was tuned closer to the 1203 nm absorption resonance of adipose tissues, a marked enhancement in contrast between the images of adipose and fibrous tissues was observed. A similar wavelength-dependent difference between normal and cancerous tissues was observed. These results correlate well with pathology and nuclear magnetic resonance based analyses of the samples.

Time-resolved optical mammography using a liquid coupled interface

A method has been devised for generating three-dimensional optical images of the breast using a 32-channel time-resolved system and a liquid-coupled interface. The breast is placed in a hemispherical cup surrounded by sources and detectors, and the remaining space is filled with a fluid with tissue-like optical properties. This approach has three significant benefits. First, cups can accommodate a large range of breast sizes, enabling the entire volume of the breast to be sampled. Second, the coupling of the source and detector optics at the surface is constant and independent of the subject, enabling intensity measurements to be employed in the image reconstruction. Third, the external geometry of the reconstructed volume is known exactly. Images of isolated targets with contrasting absorbing and scattering properties have been acquired, and the performance of the system has been evaluated in terms of the contrast, spatial resolution, and localization accuracy. These parameters were strongly dependent on the location of the targets within the imaged volume. Preliminary images of a healthy human subject are also presented, which reveal subtle heterogeneity, particularly in the distribution of scatter. The ability to detect an absorbing target adjacent to the breast is also demonstrated.

Optical imaging of turbid media using independent component analysis: theory and simulation

A new imaging approach for 3-D localization and characterization of objects in a turbid medium using independent component analysis (ICA) from information theory is developed and demonstrated using simulated data. This approach uses a multisource and multidetector signal acquisition scheme. ICA of the perturbations in the spatial intensity distribution measured on the medium boundary sorts out the embedded objects. The locations and optical characteristics of the embedded objects are obtained from a Green's function analysis based on any appropriate model for light propagation in the background medium. This approach is shown to locate and characterize absorptive and scattering inhomogeneities within highly scattering medium to a high degree of accuracy. In particular, we show this approach can discriminate between absorptive and scattering inhomogeneities, and can locate and characterize complex inhomogeneities, which are both absorptive and scattering. The influence of noise and uncertainty in background absorption or scattering on the performance of this approach is investigated.

Dynamic time-resolved diffuse spectroscopy based on supercontinuum light pulses

We present a detailed characterization of a system for fast time-resolved spectroscopy of turbid media based on supercontinuum generation in a photonic crystal fiber. The light source provides subpicosecond pulses in the 550-1000-nm spectral range, at 85 MHz, at an average power of up to 50 mW. Wavelength-resolved detection is accomplished by means of a spectrometer coupled to a 16-channel, multianode photomultiplier tube, giving a resolution of 4.5-35 nm/channel, depending on the grating. Time-dispersion curves are acquired with time-correlated single-photon counting, and absorption and reduced scattering coefficients are determined by fitting the data to the diffusion equation. We characterized the system by measuring the time-resolved diffuse reflectance of epoxy phantoms and by assessing the performance in terms of accuracy, linearity, noise sensitivity, stability, and reproducibility. The results were similar to those from previous systems, whereas the full-spectrum (610-810 nm) acquisition time was as short as 1 s owing to the parallel acquisition. We also present the first in vivo real-time dynamic spectral measurements showing tissue oxygenation changes in the arm of a human subject.

Time-domain optical mammography SoftScan: initial results

RATIONALE AND OBJECTIVES

Near-infrared (NIR) technology appears promising as a noninvasive technique for breast cancer screening and diagnosis. The technology capitalizes on the relative transparency of human tissue in this spectral range and its sensitivity to the main components of the breast: water, lipid, and hemoglobin. In this study, the authors report quantitative measurements of these components and the functional contrast between healthy and diseased tissue.

MATERIALS AND METHODS

A four-wavelength time domain optical imaging system was used to perform noninvasive NIR measurements in the breast of 49 women both pre- and postmenopausal, ages 24-80. Algorithms based on a diffusive model of light transport provided absolute bulk and local values of breast constituent concentrations.

RESULTS

Important variations in the functional and structural NIR properties of the breast were observed. Demographics trend were noticed in accordance with breast physiology. In the 23 cases imaged with suspicious masses, the optical images were consistent with the mammographic findings. Substantial contrast between masses and adjacent tissue is observed. Moreover, consistent differences between malign and benign cases are found with optical imaging.

CONCLUSION

The results of this pilot study illustrate the sensitivity of optical techniques to the composition of the breast. In addition, preliminary data suggest that benign and malignant tumors can potentially be noninvasively differentiated with optical imaging. Moreover, statistically significant discrimination based on deoxy-hemoglobin content between malign and benign cases was found with optical imaging (P = .0184, one-tailed t test).

Linear and nonlinear reconstruction for optical tomography of phantoms with nonscattering regions

Most research in optical imaging incorrectly assumes that light transport in nonscattering regions in the head may be modeled by use of the diffusion approximation. The effect of this assumption is examined in a series of experiments on tissue-equivalent phantoms. Images from cylindrical and head-shaped phantoms with and without clear regions [simulating the cerebrospinal fluid (CSF) filled ventricles] and a clear layer (simulating the CSF layer surrounding the brain) are reconstructed with linear and nonlinear reconstruction techniques. The results suggest that absorbing and scattering perturbations can be identified reliably with nonlinear reconstruction methods when the clear regions are also present in the reference data but that the quality of the image degrades considerably if the reference data does not contain these features. Linear reconstruction performs similarly to nonlinear reconstruction, provided the clear regions are present in the reference data, but otherwise linear reconstruction fails. This study supports the use of linear reconstruction for dynamic imaging but suggests that, in all cases, image quality is likely to improve if the clear regions are modeled correctly.

Recent advances in diffuse optical imaging

We review the current state-of-the-art of diffuse optical imaging, which is an emerging technique for functional imaging of biological tissue. It involves generating images using measurements of visible or near-infrared light scattered across large (greater than several centimetres) thicknesses of tissue. We discuss recent advances in experimental methods and instrumentation, and examine new theoretical techniques applied to modelling and image reconstruction. We review recent work on in vivo applications including imaging the breast and brain, and examine future challenges.

Simulation of voltage-sensitive optical signals in three-dimensional slabs of cardiac tissue: application to transillumination and coaxial imaging methods

Voltage-sensitive dyes are an important tool in visualizing electrical activity in cardiac tissue. Until today, they have mainly been applied in cardiac electrophysiology to subsurface imaging. In the present study, we assess different imaging methods used in optical tomography with respect to their effectiveness in visualizing 3D cardiac activity. To achieve this goal, we simulate optical signals produced by excitation fronts initiated at different depths inside the myocardial wall and compare their properties for various imaging modes. Specifically, we consider scanning and broad-field illumination, including trans- and epi-illumination. We focus on the lateral optical resolution and signal intensity, as a function of the source depth. Optical diffusion theory is applied to derive a computationally efficient approximation of the point-spread function and to predict voltage-sensitive signals. Computations were performed both for fluorescent and absorptive voltage-sensitive dyes. Among all the above-mentioned methods, fluorescent coaxial scanning yields the best resolution (<2.5 mm) and gives the most information about the intramural cardiac activity.

Gauss-Newton method for image reconstruction in diffuse optical tomography

We present a regularized Gauss-Newton method for solving the inverse problem of parameter reconstruction from boundary data in frequency-domain diffuse optical tomography. To avoid the explicit formation and inversion of the Hessian which is often prohibitively expensive in terms of memory resources and runtime for large-scale problems, we propose to solve the normal equation at each Newton step by means of an iterative Krylov method, which accesses the Hessian only in the form of matrix-vector products. This allows us to represent the Hessian implicitly by the Jacobian and regularization term. Further we introduce transformation strategies for data and parameter space to improve the reconstruction performance. We present simultaneous reconstructions of absorption and scattering distributions using this method for a simulated test case and experimental phantom data.

Estimation of anisotropic optical parameters of tissue in a slab geometry

The scattering and absorption coefficients of many homogeneous biological tissues such as muscle, skin, white matter in the brain, and dentin are often anisotropically oriented with respect to their bounding interface. In consequence the curves of equal intensity of re-emitted light on the surface of the slab will no longer be circular. We here consider the problem of determining the parameters allowing one to estimate the angles defining anisotropy, directional bias of diffusive spreading, and scattering and absorbing coefficients from data obtained from time-gated measurements of light intensity transmitted through a slab of the tissue. Our model can be solved exactly and leads to accurate approximations in which measured values of the surface intensity are shown to be elliptical. The parameters of the ellipses suffice to estimate the anisotropy of the tissue interior. A summary of the parameter estimates with the observables from which they are found is given in a table. Our analysis is based on a diffusion model.

Improvement of transcutaneous fluorescent images with a depth-dependent point-spread function

The point-spread function (PSF) for transcutaneous fluorescent imaging was obtained as an analytical solution in a closed form. It is applicable to cases in which the optical property of the image-blurring turbid medium is considered to be fairly homogeneous. We proposed a technique to improve a transcutaneous image by using depth-dependent PSF. Contrast of the fluorescent image was improved for depths of 1-15 mm in a scattering medium (micro(s)' = 1/mm). The visible depth was more than doubled with this technique. An experiment with a rat demonstrated considerable improvement of a transcutaneous image of the cerebral vein at a specified depth. The spread image of the heart was reduced to the correct size by use of the PSF with the actual depth of the heart.

Three-dimensional localization and optical imaging of objects in turbid media with independent component analysis

A new approach for optical imaging and localization of objects in turbid media that makes use of the independent component analysis (ICA) from information theory is demonstrated. Experimental arrangement realizes a multisource illumination of a turbid medium with embedded objects and a multidetector acquisition of transmitted light on the medium boundary. The resulting spatial diversity and multiple angular observations provide robust data for three-dimensional localization and characterization of absorbing and scattering inhomogeneities embedded in a turbid medium. ICA of the perturbations in the spatial intensity distribution on the medium boundary sorts out the embedded objects, and their locations are obtained from Green's function analysis based on any appropriate light propagation model. Imaging experiments were carried out on two highly scattering samples of thickness approximately 50 times the transport mean-free path of the respective medium. One turbid medium had two embedded absorptive objects, and the other had four scattering objects. An independent component separation of the signal, in conjunction with diffusive photon migration theory, was used to locate the embedded inhomogeneities. In both cases, improved lateral and axial localizations of the objects over the result obtained by use of common photon migration reconstruction algorithms were achieved. The approach is applicable to different medium geometries, can be used with any suitable photon propagation model, and is amenable to near-real-time imaging applications.

Optical tomographic imaging of small animals

Diffuse optical tomography is emerging as a viable new biomedical imaging modality. Using visible and near-infrared light this technique can probe the absorption and scattering properties of biological tissues. The main applications are currently in brain, breast, limb and joint imaging; however, optical tomographic imaging of small animals is attracting increasing attention. This interest is fuelled by recent advances in the transgenic manipulation of small animals that has led to many models of human disease. In addition, an ever increasing number of optically reactive biochemical markers has become available, which allow diseases to be detected at the molecular level long before macroscopic symptoms appear. The past three years have seen an array of novel technological developments that have led to the first optical tomographic studies of small animals in the areas of cerebral ischemia and cancer.

A general framework for nonlinear multigrid inversion

A variety of new imaging modalities, such as optical diffusion tomography, require the inversion of a forward problem that is modeled by the solution to a three-dimensional partial differential equation. For these applications, image reconstruction is particularly difficult because the forward problem is both nonlinear and computationally expensive to evaluate. In this paper, we propose a general framework for nonlinear multigrid inversion that is applicable to a wide variety of inverse problems. The multigrid inversion algorithm results from the application of recursive multigrid techniques to the solution of optimization problems arising from inverse problems. The method works by dynamically adjusting the cost functionals at different scales so that they are consistent with, and ultimately reduce, the finest scale cost functional. In this way, the multigrid inversion algorithm efficiently computes the solution to the desired fine-scale inversion problem. Importantly, the new algorithm can greatly reduce computation because both the forward and inverse problems are more coarsely discretized at lower resolutions. An application of our method to Bayesian optical diffusion tomography with a generalized Gaussian Markov random-field image prior model shows the potential for very large computational savings. Numerical data also indicates robust convergence with a range of initialization conditions for this nonconvex optimization problem.

Polarization visualization and selection of biotissue image two-layer scattering medium

We analyze and experimentally test the concept of laser polarization biotissue probing. The methods of increasing the SNR in coherent images of the optically anisotropic architectonics of the morphological biotissue structure are considered. The possibilities of polarization selection and contrasting of such images screened by other biotissues are examined. The influence of the depolarization degree of the scattered background on the SNR is investigated. The possibilities of polarization correction of the probing beam for contrasting biotissue images are analyzed.

Diffuse optical tomography using intensity measurements and the a priori acquired regions of interest: theory and simulations

Light transmission data collected around an object show large variation with source-detector separation owing to the presence of single or multiple inhomogeneous regions in the object. This variation in the measured intensity is made use of to reconstruct regions of the inhomogeneous inclusions. In addition, it is possible to select a set of data from the above which is most likely least affected by the presence of the inhomogeneity, and estimate reasonably accurately the background optical properties from it. The reconstructed region is found to always contain the inhomogeneity and is of size approximately 140% by area of the inhomogeneity. With the regions to be reconstructed a priori known, a model-based iterative reconstruction procedure for reconstructing the optical properties of the region converged five times faster than without such information. It is also shown that whereas for the full object, a view-based propagation-backpropagation reconstruction procedure failed to converge, owing to large underdeterminacy of the problem, a smaller problem attempting to reconstruct a priori specified regions of interest converged and did so faster than a non-view-based approach for similar objects. Reconstruction results are presented from simulated transmitted intensity data from the following objects with regions of inhomogeneity in both absorption and scattering: (i) single centrally located inhomogeneity, (ii) two off-centred inhomogeneous regions of equal size and contrast (iii) two off-centred inhomogeneous regions of unequal size and equal contrast and (iv) two off-centred inhomogeneous regions of unequal size and contrast. Whereas the model-based iterative image reconstruction procedure gave good convergence in the first, second and third cases, in the fourth case the reconstructions failed to recover the exact numerical value of the optical properties in the higher contrast region.

Optical biopsy: a new frontier in endoscopic detection and diagnosis

Endoscopic diagnosis currently relies on the ability of the operator to visualize abnormal patterns in the image created by light reflected from the mucosal surface of the gastrointestinal tract. Advances in fiber optics, light sources, detectors, and molecular biology have led to the development of several novel methods for tissue evaluation in situ. The term "optical biopsy" refers to methods that use the properties of light to enable the operator to make an instant diagnosis at endoscopy, previously possible only by using histological or cytological analysis. Promising imaging techniques include fluorescence endoscopy, optical coherence tomography, confocal microendoscopy, and molecular imaging. Point detection schemes under development include light scattering and Raman spectroscopy. Such advanced diagnostic methods go beyond standard endoscopic techniques by offering improved image resolution, contrast, and tissue penetration and providing biochemical and molecular information about mucosal disease. This review describes the basic biophysics of light-tissue interactions, assesses the strengths and weaknesses of each method, and examines clinical and preclinical evidence for each approach.

Tissue characterization by quantitative optical imaging methods

Optical methods have a long history in the field of medical diagnosis. The biomolecular specificity possible with optical methods has been particularly valuable in microscopy and histopathology while in vivo imaging of deep structures has traditionally been the domain of X-ray and MRI. The use of optical methods in deep tissue has been limited by multiple-scattering which blurs or distorts the optical signal. New stochastic methods which account for multiple scattering have been developed that are extending the usefulness of optical methods deep into tissue. In optical mammography, photons may travel through 10 cm of tissue before arriving at the detector. We have developed a method for quantifying parameters of anomalous sites in breast tissue that may be used for functional characterization of tumors. In other work presented here, we are developing fluorescence based methods to detect and monitor tumor status. The immune response to a tumor is a target for fluorescently labeled specific antibodies. We have developed a method to localize the tumor site using CW fluorescence. Additionally, we have developed a method which uses time-resolved data and capitalizes on probe lifetime sensitivity to metabolic parameters such as pH and temperature to obtain functional information from the tumor site.

NOISE 2 imaging system: seeing through scattering tissue with a reference point

We propose a fly-eye-like imaging system for seeing objects embedded in scattering media. Objects are recovered from many speckled images observed by a digital camera through a microlens array. Each microlens in the array generates a speckle image of the object buried between two layers of chicken breast tissue. In the computer each image is Fourier transformed jointly with an image of the speckled pointlike source captured under the same conditions. A set of the squared magnitudes of the Fourier-transformed pictures is accumulated to form a single average picture. This final picture is again Fourier transformed, resulting in the reconstruction of the hidden object.

Noninvasive optical imaging by speckle ensemble

We propose a new method imaging through scattering media. An object hidden between two biological tissues (chicken breast) is reconstructed from any speckled images obtained from the output of a multichannel optical imaging system. The effect of multiple imaging is achieved with a microlens array. Each lens is the array projects a different speckled image onto a digital camera. The set of speckled images from the entire array is first shifted to a common center and then accumulated into a single average picture.

Parallel programming of gradient-based iterative image reconstruction schemes for optical tomography

Optical tomography (OT) is a fast developing novel imaging modality that uses near-infrared (NIR) light to obtain cross-sectional views of optical properties inside the human body. A major challenge remains the time-consuming, computational-intensive image reconstruction problem that converts NIR transmission measurements into cross-sectional images. To increase the speed of iterative image reconstruction schemes that are commonly applied for OT, we have developed and implemented several parallel algorithms on a cluster of workstations. Static process distribution as well as dynamic load balancing schemes suitable for heterogeneous clusters and varying machine performances are introduced and tested. The resulting algorithms are shown to accelerate the reconstruction process to various degrees, substantially reducing the computation times for clinically relevant problems.

Temporal response of a random medium from speckle intensity frequency correlations

We reconstruct the temporal response of a random medium by using speckle intensity frequency correlations. When the scattered field from a random medium is described by circular complex Gaussian statistics, we show that third-order correlations permit retrieval of the Fourier phase of the temporal response with bispectral techniques. Our experimental results for random media samples in the diffusion regime are in excellent agreement with the intensity temporal response measured directly with an ultrafast pulse laser and a streak camera. Our speckle correlation measurements also demonstrate sensitivity to inhomogeneous samples, highlighting the potential application for imaging within a scattering medium.

Validity conditions for the radiative transfer equation

We compare the radiative transfer equation for media with constant refractive index with the radiative transfer equation for media with spatially varying refractive indices [J. Opt. A Pure App. Opt. 1, L1 (1999)] and obtain approximate conditions under which the former equation is accurate for modeling light propagation in scattering media with spatially varying refractive indices. These conditions impose restrictions on the variations of the refractive index, the gradient of the refractive index, the divergence of the rays, the frequency of modulation, and the widths of light pulses, which are related to the mean refractive index, the absorption coefficient, and the reduced scattering coefficient of the medium. Each condition is geometrically interpreted. Some implications of the results are discussed.

Optimum filtering for EROS measurements

The event-related optical signal (EROS) is a relatively new technology that provides noninvasive data about the time course of neural activity in circumscribed cortical areas. However, much still remains to be learned about the physiology and physics underlying the observed signals. We examined the instrumental and physiological noise observed in the intensity modulation and phase-delay measurements produced by a frequency domain oxymeter in response to steady-state auditory stimuli. We present here data on the effects of different filters on the between-subjects response consistency for amplitude and phase measurements. The results of these analyses may help explain some apparently discrepant results previously reported from different laboratories (Steinbrink et al., 2000), and illustrate differences between data from different types of measurements.

Time-series estimation of biological factors in optical diffusion tomography

We apply state space estimation techniques to the time-varying reconstruction problem in optical tomography. We develop a stochastic model for describing the evolution of quasi-sinusoidal medical signals such as the heartbeat, assuming these are represented as a known frequency with randomly varying amplitude and phase. We use the extended Kalman filter in combination with spatial regularization techniques to reconstruct images from highly under-determined time-series data. This system also naturally segments activity belonging to different biological processes. We present reconstructions of simulated data and of real data recorded from the human motor cortex (Franceschini et al 2000 Optics Express 6 49-57). It is argued that the application of these time-series techniques improves both the fidelity and temporal resolution of reconstruction in optical tomography.

In vivo quantitative three-dimensional localization of tumor labeled with exogenous specific fluorescence markers

We introduce a diffused optical detection system based on the administration of a fluorophore-antibody conjugate to diseased tissue. The conjugate interacts with the antigens expressed by the diseased tissue, resulting in fluorescent labeling of the antigen. By combining an optical detection system with a reconstruction algorithm developed on the basis of the random-walk model, we were able to determine the position of the fluorophore (and, thus, of the diseased cells) in the tissue. We present three-dimensional reconstructions of the location of a fluorophore (FITC-fluorescein isothiocyanate) in the tongues of mice. Measurements were performed with the fluorophore embedded at various simulated depths. The simulations were performed with agarose-based gel slabs applied to the tongue as tissuelike phantoms. Reconstructed fluorophore locations agree well with the actual values.

Near-infrared light propagation in an adult head model. I. Modeling of low-level scattering in the cerebrospinal fluid layer

Adequate modeling of light propagation in a human head is important for quantitative near-infrared spectroscopy and optical imaging. The presence of a nonscattering cerebrospinal fluid (CSF) that surrounds the brain has been previously shown to have a strong effect on light propagation in the head. However, in reality, a small amount of scattering is caused by the arachnoid trabeculae in the CSF layer. In this study, light propagation in an adult head model with discrete scatterers distributed within the CSF layer has been predicted by Monte Carlo simulation to investigate the effect of the small amount of scattering caused by the arachnoid trabeculae in the CSF layer. This low scattering in the CSF layer is found to have little effect on the mean optical path length, a parameter that can be directly measured by a time-resolved experiment. However, the partial optical path length in brain tissue that relates the sensitivity of the detected signal to absorption changes in the brain is strongly affected by the presence of scattering within the CSF layer. The sensitivity of the near-infrared signal to hemoglobin changes induced by brain activation is improved by the effect of a low-scattering CSF layer.

Streak camera: a multidetector for diffuse optical tomography

We describe an experimental setup for time-resolved diffuse optical tomography that uses a seven-channel light guide to transmit scattered light to a streak camera. This setup permits the simultaneous measurement of the time profiles of photons reemitted at different boundary sites of the objects studied. The instrument, its main specifications, and detector-specific data analysis before image reconstruction are described. The instrumentation was tested with phantoms simulating biological tissue, and the absorption and reduced scattering images that were obtained are discussed.

Hybrid Monte Carlo-diffusion method for light propagation in tissue with a low-scattering region

The heterogeneity of the tissues in a head, especially the low-scattering cerebrospinal fluid (CSF) layer surrounding the brain has previously been shown to strongly affect light propagation in the brain. The radiosity-diffusion method, in which the light propagation in the CSF layer is assumed to obey the radiosity theory, has been employed to predict the light propagation in head models. Although the CSF layer is assumed to be a nonscattering region in the radiosity-diffusion method, fine arachnoid trabeculae cause faint scattering in the CSF layer in real heads. A novel approach, the hybrid Monte Carlo-diffusion method, is proposed to calculate the head models, including the low-scattering region in which the light propagation does not obey neither the diffusion approximation nor the radiosity theory. The light propagation in the high-scattering region is calculated by means of the diffusion approximation solved by the finite-element method and that in the low-scattering region is predicted by the Monte Carlo method. The intensity and mean time of flight of the detected light for the head model with a low-scattering CSF layer calculated by the hybrid method agreed well with those by the Monte Carlo method, whereas the results calculated by means of the diffusion approximation included considerable error caused by the effect of the CSF layer. In the hybrid method, the time-consuming Monte Carlo calculation is employed only for the thin CSF layer, and hence, the computation time of the hybrid method is dramatically shorter than that of the Monte Carlo method.

State-estimation approach to the nonstationary optical tomography problem

We propose a new numerical approach to the nonstationary optical (diffusion) tomography (OT) problem. The assumption in the method is that the absorption and/or diffusion coefficients are nonstationary in the sense that they may exhibit significant changes during the time that is needed to measure data for one traditional image frame. In the proposed method, the OT problem is formulated as a state-estimation problem. Within the state-estimation formulation, the absorption and/or diffusion coefficients are considered a stochastic process. The objective is to estimate a sequence of states for the process when the state evolution model for the process, the observation model for OT experiments, and data on the exterior boundary are given. In the proposed method, the state estimates are computed by using Kalman filtering techniques. The performance of the proposed method is evaluated on the basis of synthetic data. The simulations also illustrate that further improvements to the results in nonstationary applications can be obtained by adjustment of the measurement protocol.

Assessment of an in situ temporal calibration method for time-resolved optical tomography

A 32-channel time-resolved optical imaging device is developed at University College London to produce functional images of the neonatal brain and the female breast. Reconstruction of images using time-resolved measurements of transmitted light requires careful calibration of the temporal characteristics of the measurement system. Since they can often vary over a period of time, it is desirable to evaluate these characteristics immediately after, or prior to, the acquisition of image data. A calibration technique is investigated that is based on the measurement of light back-reflected from the surface of the object being imaged. This is facilitated by coupling each detector channel with an individual source fiber. A Monte Carlo model is employed to investigate the influence of the optical properties of the object on the back-reflected signal. The results of simulations indicate that their influence may be small enough to be ignored in some cases, or could be largely accounted for by a small adjustment to the calibrated data. The effectiveness of the method is briefly demonstrated by imaging a solid object with tissue-equivalent optical properties.

Oxyphor R2 and G2: phosphors for measuring oxygen by oxygen-dependent quenching of phosphorescence

Oxygen-dependent quenching of phosphorescence is a useful and essentially noninvasive optical method for measuring oxygen in vivo and in vitro. Calibration of the phosphors is absolute, and once phosphors have been calibrated in one laboratory the same constants can be used by anyone else as long as the measurement is done under the same conditions. Two new phosphors, one based on Pd-meso-tetra-(4-carboxyphenyl)porphyrin and the other on Pd-meso-tetra-(4-carboxyphenyl)tetrabenzoporphyrin, are very well suited to in vivo oxygen measurements. Both phosphors are Generation 2 polyglutamic Pd-porphyrin-dendrimers, bearing 16 carboxylate groups on the outer layer. These phosphors are designated Oxyphor R2 and Oxyphor G2, respectively. Both are highly soluble in biological fluids such as blood plasma and their ability to penetrate biological membranes is very low. The maxima in the absorption spectra are at 415 and 524 nm for Oxyphor R2 and 440 and 632 nm for Oxyphor G2, while emissions are near 700 and 800 nm, respectively. The calibration constants of the phosphors are essentially independent of pH in the physiological range (6.4 to 7.8). In vivo application is demonstrated by using Oxyphor G2 to noninvasively determine the oxygen distribution in a subcutaneous tumor growing in rats.

Non-invasive neuroimaging using near-infrared light

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.

Spatial resolution in fast time-resolved transillumination imaging: an indeterministic Monte Carlo approach

The spatial resolution achievable in time-resolved optical transillumination imaging through a turbid (scattering and absorbing) medium has been reassessed theoretically. The temporal point spread function was constructed assuming a delta function input pulse, a approximately 50 mm thick medium and a small detector with zero risetime. Temporal profiles were derived from an indeterministic Monte Carlo simulation for different time scales. From the temporal point spread function (TPSF), an analytic edge response function from which the spatial resolution was determined was derived. Previous analytical methods for determining the spatial resolution are approximations for very short flight times (sub-100 ps time region). The results show that a spatial resolution of about two millimetres is possible under ideal signal-to-noise ratio conditions and with detector gate times of the order of ten picoseconds. If this predicted spatial resolution can be achieved in an imaging system, it may be possible to improve the diagnosis of breast tumours.

Number of distinct sites visited by a random walker trapped by an absorbing boundary

The number of distinct sites visited by a lattice random walker is a subject of continuing interest in both mathematics and physics. All previous investigations have used the assumption that the lattice is unbounded. An assessment of the amount of tissue interrogated by a photon in reflectance measurements for diagnostic purposes suggests analyzing properties of the average number of distinct sites visited by a random walker trapped by an absorbing plane at time t. We show that for sufficiently large t this number is the same as the average number of distinct sites visited for this time when the surface is not present. A more complete analysis is possible for a random walk on a line terminated by an absorbing point.

Analytical calculation of the mean time spent by photons inside an absorptive inclusion embedded in a highly scattering medium

The mean time spent by photons inside a nonlocalized optically abnormal embedded inclusion has been derived analytically. The accuracy of the results has been tested against Monte Carlo and experimental data. We show that for quantification of the absorption coefficient of absorptive inclusions, a corrective factor that takes into account the size of the inclusion is needed. This finding suggests that perturbation methods derived for very small inclusions which are used in inverse algorithms require a corrective factor to adequately quantify the differential absorption coefficient of nonlocalized targets embedded in optically turbid media.

Photon migration in turbid media using a cumulant approximation to radiative transfer

A photon transport model for light migration in turbid media based on a cumulant approximation to radiative transfer is presented for image reconstruction inside an infinite medium or a bounded medium with a planar geometry. This model treats weak inhomogeneities through a Born approximation of the Boltzmann radiative transfer equation and uses the second-order cumulant solution of photon density to the Boltzmann equation as the Green's function for the uniform background. It provides the correct behavior of photon migration at early times and reduces at long times to the center-moved diffusion approximation. At early times, it agrees much better with the result from the Monte Carlo simulation than the diffusion approximation. Both approximations agree well with the Monte Carlo simulation at later times. The weight function for image reconstruction under this proposed model is shown to have a strong dependence at both early and later times on absorption and/or scattering inhomogeneities located in the propagation direction of and close to the source, or in the field of view of and close to the detector. This effect originates from the initial ballistic motion of incident photons, which is substantially underestimated by the diffusion approximation.

Linear single-step image reconstruction in the presence of nonscattering regions

There is growing interest in the use of near-infrared spectroscopy for the noninvasive determination of the oxygenation level within biological tissue. Stemming from this application, there has been further research in using this technique for obtaining tomographic images of the neonatal head, with the view of determining the level of oxygenated and deoxygenated blood within the brain. Because of computational complexity, methods used for numerical modeling of photon transfer within tissue have usually been limited to the diffusion approximation of the Boltzmann transport equation. The diffusion approximation, however, is not valid in regions of low scatter, such as the cerebrospinal fluid. Methods have been proposed for dealing with nonscattering regions within diffusing materials through the use of a radiosity-diffusion model. Currently, this new model assumes prior knowledge of the void region; therefore it is instructive to examine the errors introduced in applying a simple diffusion-based reconstruction scheme in cases where a nonscattering region exists. We present reconstructed images, using linear algorithms, of models that contain a nonscattering region within a diffusing material. The forward data are calculated by using the radiosity-diffusion model, and the inverse problem is solved by using either the radiosity-diffusion model or the diffusion-only model. When using data from a model containing a clear layer and reconstructing with the correct model, one can reconstruct the anomaly, but the qualitative accuracy and the position of the reconstructed anomaly depend on the size and the position of the clear regions. If the inverse model has no information about the clear regions (i.e., it is a purely diffusing model), an anomaly can be reconstructed, but the resulting image has very poor qualitative accuracy and poor localization of the anomaly. The errors in quantitative and localization accuracies depend on the size and location of the clear regions.