This PDF file contains the front matter associated with SPIE Proceedings Volume 6434, including the Title Page, Copyright information, Table of Contents, and the Conference Committee listing.

In this paper, we propose a novel method to solve the forward and inverse problems in diffuse optical tomography. Our forward solution is based on the diffusion approximation equation and is constructed using the Feynman-Kac formula with an interacting particle method. It can be implemented using Monte-Carlo (MC) method and thus provides great flexibility in modeling complex geometries. But different from conventional MC approaches, it uses excursions of the photons' random walks and produces a transfer kernel so that only one round of MC-based forward simulation (using an arbitrarily known optical distribution) is required in order to get observations associated with different optical distributions. Based on these properties, we develop a perturbation-based method to solve the inverse problem in a discretized parameter space. We validate our methods using simulated 2D examples. We compare our forward solutions with those obtained using the finite element method and find good consistency. We solve the inverse problem using the maximum likelihood method with a greedy optimization approach. Numerical results show that if we start from multiple initial points in a constrained searching space, our method can locate the abnormality correctly.

Cramer-Rao Bounds (CRB) for the expected variance in the parameter space were examined for Diffuse Optical Tomography (DOT), to define the lower bound (CRLB) of an ideal system. The results show that the relative standard deviation in the optical parameter estimate follows an inverse quadratic function with respect to signal to noise ratio (SNR). The CRLB was estimated for three methods of including spatial constraints. The CRLB estimate decreased by a factor of 10 when parameter reduction using spatial constraints (hard-priors) was enforced whereas, inclusion of spatial-priors in the regularization matrix (soft-priors) decreased the CRLB estimate only by a factor of 4. The maximum reduction in variance from the use of spatial-priors, occurred in the background of the imaging domain as opposed to localized target regions. As expected, the variance in the recovered properties increased as the number of parameters to be estimated increased. Additionally, increasing SNR beyond a certain point did not influence the outcome of the optical property estimation when prior information was available.

The underlying mathematical model for bioluminescence tomography (BLT) is an ill-posed inverse source problem, which can be cast into an optimization problem of a χ²-error function. The error function quantifies the discrepancy between the measured and predicted partial current of bioluminescent light at the tissue boundary. In this work, the predicted partial current has been calculated prior to image reconstruction by solving the equation of radiative transfer (ERT) for a set of source basis functions of the entire tissue domain. A global optimization technique based on stochastic sampling principles probes the global parameter space of bioluminescent source distributions composed of source basis functions. The stochastic approach avoids a premature convergence, which can be caused, e.g., by a narrow or a shallow surface landscape of the χ²-error function. Initial reconstruction results are compared to gradient-based image reconstructions.

Radio surgical interventions such as Gamma Knife and Cyberknife have become attractive as therapeutic interventions. However, one of the drawbacks of cyberknife is radionecrosis, which is caused by excessive radiation to surrounding normal tissues. Radionecrosis occurs in about 10-15% of cases and could have adverse effects leading to death. Currently available imaging techniques have failed to reliably distinguish radionecrosis from tumor growth. Development of imaging techniques that could provide distinction between tumor growth and radionecrosis would give us ability to monitor effects of radiation therapy non-invasively. This paper investigates the use of near infrared spectroscopy (NIRS) as a new technique to monitor the growth of brain tumors. Brain tumors (9L glioma cell line) were implanted in right caudate nucleus of rats (250-300 gms, Male Fisher C) through a guide screw. A new algorithm was developed, which used broadband steady-state reflectance measurements made using a single source-detector pair, to quantify absolute concentrations of hemoglobin derivatives and reduced scattering coefficients. Preliminary results from the brain tumors indicated decreases in oxygen saturation, oxygenated hemoglobin concentrations and increases in deoxygenated hemoglobin concentrations with tumor growth. The study demonstrates that NIRS technology could provide an efficient, noninvasive means of monitoring vascular oxygenation dynamics of brain tumors and further facilitate investigations of efficacy of tumor treatments.

Endoscopic near-infrared (NIR) optical tomography is a novel approach that allows the blood-based high intrinsic optical contrast to be imaged for the detection of cancer in internal organs. In endoscopic NIR tomography, the imaging array is arranged within the interior of the medium as opposed to the exterior as seen in conventional NIR tomography approaches. The source illuminates outward from the circular NIR probe, and the detector collects the diffused light from the medium surrounding the NIR probe. This new imaging geometry may involve forward and inverse approaches that are significantly different from those used in conventional NIR tomography. The implementation of a hollow-centered forward mesh within the context of conventional NIR tomography reconstruction has already led to the first demonstration of endoscopic NIR optical tomography. This paper presents some fundamental computational aspects regarding the performance and sensitivity of this endoscopic NIR tomography configuration. The NIRFAST modeling and image reconstruction package developed for conventional circular NIR geometry is used for endoscopic NIR tomography, and initial quantitative analysis has been conducted to investigate the "effective" imaging depth, required mesh resolution, and limit in contrast resolution, among other parameters. This study will define the performance expected and may provide insights into hardware requirements needed for revision of NIRFAST for the endoscopic NIR tomography geometry.

Shallow breast lesions less than 1.0 cm deep are frequently seen in patients when they are scanned in a supine position by a hand-held combined ultrasound and optical probe. Reflection boundary condition is suitable for imaging shallow lesions, but it is not sensitive enough for imaging lesions deeper than 1.0 cm. The absorption boundary condition is more desirable for imaging deeper lesions. However, it is less sensitive to shallower lesions unless the optical sources are positioned right on top of the lesion which is unknown in general. To solve this problem, we have designed a new probe which incorporates three angled sources to improve the illumination of the shallow region underneath the co-registered ultrasound probe. Monte Carlo method is extended to include an absorber in the medium and used to evaluate the analytical results. Simulations have shown that the combination of tilted and peripherally located sources can improve reconstruction compared with the probe with no tilted sources. Phantom experiments agree with simulation results.

Fluorescence molecular tomography (FMT) suffers from inherent ill-posedness due to the vast number of possible solutions to the reconstruction problem. To increase the robustness of such a problem one need prior information. We present here a method for rendering a priori information of the position of a fluorescent inclusion inside turbid media. The method utilizes solely two spectral bands within the fluorescence spectrum emitted from the fluorophore. The method is presented and verified using experimental data from a tissue phantom. The confinement is also used to impose weights onto the voxels before the inversion of the linear set of equations describing the FMT problem.

A theoretical framework is presented that allows a lifetime based analysis of the entire temporal diffuse fluorescence response curve from a turbid medium. Optimization studies using singular value decomposition analysis show that direct time domain fluorescence reconstructions are optimally performed using a few points near the peak and rise portions of the temporal response. It is also shown that the initial portion of the fluorescent response curve offers superior contrast-to-noise performance, while the late decay portions offer minimal cross-talk between multiple lifetime components.

We report on the reconstruction of absorption and fluorescence from measured time-domain diffuse reflectance and transmittance of laser and fluorescence radiation. Measurements were taken on slab-like, diffusely scattering and fluorescent phantoms containing fluorescent inhomogeneities, using fs laser pulses (&lgr; = 730 nm) and time correlated single photon counting. The source was scanned across the entrance face of the phantom, and at each source position data were collected in transmission and reflection at various detector positions. These measurements simulate in vivo data that will be obtained employing a scanning, time-domain fluorescence mammograph, where the breast is gently compressed between two parallel glass plates, and source and detector optical fibers scan synchronously at various source-detector offsets, allowing to record laser and fluorescence mammograms. The diffusion equations for the propagation of the laser and fluorescence radiation were solved in frequency domain by the finite element method. Measured time-resolved phantom data were Fourier-transformed to frequency domain prior to image reconstruction. Signal-to-noise ratios were high enough to use several data sets simultaneously in the reconstruction process belonging to various modulation frequencies up to several hundred MHz. To obtain the spatial distribution of the fluorescent contrast agent the Born approximation of the fluorescence diffusion equation was used.

We have carried out phantom studies for optimizing the design of a fluorescence mammograph employing time-domain and cw measurements, for improving data analysis and methods of reconstruction. By scanning pulsed (100 fs) laser radiation across a fluorescent, rectangular breast-like phantom with a spherical inhomogeneity simulating a tumor bearing breast slightly compressed between two parallel glass plates, distributions of times of flight of laser and fluorescence photons were measured in transmission and reflection for various detector arrangements. Absorption coefficients and dye concentrations were reconstructed using perturbation solutions of the diffusion equation at the laser and fluorescence wavelengths. We additionally employed a CCD camera to measure time-integrated intensity of fluorescence and laser radiation transmitted through the phantom. The increased number of projection angles entering the reconstruction improved spatial resolution. Further improvements were obtained when combined cw data and time-resolved remission data were used in the reconstruction.

The backscattering of circularly polarized (CP) light has been investigated using experiments and an analytical cumulant solution of the vector radiative transfer equation. The expression of the exact spatial cumulants of light distribution function has been derived. Both experimental and theoretical studies show that the helicity of the incident circular polarization is maintained in the light backscattered from large particle suspensions. Reflection from an embedded target inside the turbid medium reverses the helicity of the incident circular polarization. Polarization memory imaging makes use of this difference in helicity between light reflected from the target and that from the scattering medium and significantly enhances the image contrast by selecting out the circularly cross-polarized light. We experimentally demonstrate the superior image quality for target inside large polystyrene particle suspensions in water.

A novel support vector machine (SVM) classifier incorporating the complexity of fluorescent spectral data is designed to reliably differentiate normal and malignant human breast cancer tissues. Analysis has been carried out with parallel and perpendicularly polarized fluorescence data using 36 normal and 36 cancerous tissue samples. In order to incorporate the complexity of fluorescence spectral profile into a SVM design, the curvature of phase space trajectory is extracted as a useful complexity feature. We found that the fluorescence intensity peaks at 541nm-620nm as well as the complexity features at 621nm-700nm are important discriminating features. By incorporating both features in SVM design, we can improve both sensitivity and specificity of the classifier.

A fully non-contact CCD-based approach to sub-surface fluorescence diffuse optical imaging is presented. An overview of CCD-noise sources are described and a possible solution for obtaining an adequate SNR in CCD-based diffuse optical imaging is implemented. To examine the impact of excitation and remission light attenuation in this geometry, the linearity of response in recovering object position was examined in simulations, with respect to changes in target size, target-to-background contrast, and depth. To provide insight regarding the technological complications of sub-surface imaging, liquid phantom experiments were performed for targets of size 4mm, 8mm and 14mm having 10:1 target-to-background contrast. Overall, the results indicate that steps must be taken to eliminate blooming artifacts, perhaps by physically blocking the active source as it is projected onto the CCD chip. In general, response linearity in the recovered target centroid position, size, and fluorophore concentration as well as complications arising due to partial volume sampling effects are expected to improve if prior structural images obtained from another modality are incorporated into the DOT reconstruction algorithm.

Advanced imaging systems and theoretical models have been developed to quantify fluorescence, and this theoretical framework involves numerical-based solutions of a set of coupled diffusion equations. One key to advancing this modality is the extension of the imaging into realistic tissue geometries, which can be dynamically updated from data from other high resolution modalities. Here we explore the quantification of fluorescence in a three-dimensional (3-D) mouse phantom tagged with heterogeneous optical properties. A finite element model for the diffusion equation was used to approximate light propagation along with Newton's method for image reconstruction, to recover 3-D images of fluorescent yield. Using measurements generated on a brain tumor in a mouse with 2% noise, our results show that only 11.4% of the expected fluorescent yield could be recovered without any prior knowledge about the spatial structure of the domain. Using a parameter reduction scheme based upon prior spatial information of the location and size of the tumor, 100% of the expected value could be estimated. These preliminary results indicate that image guided fluorescence spectroscopy has the ability to provide accurate fluorescence recovery, whereas diffuse imaging based recovery is limited in the ability to quantify.

In this study, time-domain fluorescence diffuse optical tomography (FDOT) in biological tissue is investigated by solving the inverse problem using a convolution and deconvolution of the zero-lifetime emission light intensity and the exponential function for a finite lifetime, respectively. We firstly formulate the fundamental equations in time-domain assuming that the fluorescence lifetime is equal to zero, and then the solution including the lifetime is obtained by convolving the emission light intensity and the lifetime function. The model is a 2-D 10 mm-radius circle with the optical properties simulating biological tissue for the near infrared light, and contains some regions with fluorophores. Temporal and spatial profiles of excitation and emission light intensities are calculated and discussed for several models. The inverse problem of fluorescence diffuse optical tomography is solved using simulated measurement emission intensities for reconstructing fluorophore concentration. A time-domain measurement system uses ultra-short pulsed laser for excitation and measures the temporal and spatial distributions of fluorescence emitting from the tissue surface. To improve image quality, we propose implementation of a FDOT algorithm using full time-resolved (TR) data.

Recent interest in modeling and reconstruction algorithms for Bioluminescence Tomography (BLT) has increased and led to the general consensus that non-spectrally resolved intensity-based BLT results in a non-unique problem. However, the light emitted from, for example firefly Luciferase, is widely distributed over the band of wavelengths from 500 nm to 650 nm and above, with the dominant fraction emitted from tissue being above 550 nm. This paper demonstrates the development of an algorithm used for multi-wavelength 3D spectrally resolved BLT image reconstruction in a mouse model. It is shown that using a single view data, bioluminescence sources of up to 15 mm deep can be successfully recovered given correct information about the underlying tissue absorption and scatter.

We propose a self-normalized scanning fluorescence diffuse optical tomography technique. The method requires a single modulated light source for excitation and multiple detector pairs symmetrically located at positions beside the source. The amplitude ratio and phase difference between the two detectors of each pair are measured as the source and detectors are scanned through the region of interest. It has been observed that the phase difference profile along a scanning line depends on target depth rather than fluorophore concentration enabling estimation of the target depth from the phase difference before reconstruction of the fluorophore concentration. The depth of a cylindrical target with a target-to-background contrast of 0.2:0.01 (Cy5.5) was varied from 0.7cm to 1.8cm, and the estimated depths were very close to their expected values with 15% maximum error. Based on the estimated target depth, the imaging volume can be segmented into a smaller region surrounding the target and a larger background region enabling the use of a dual-zone mesh based image reconstruction. Fluorophore concentration was reconstructed by using both amplitude ratio and phase difference between the detector pairs. The reconstructed fluorophore concentration has achieved 90% accuracy. Because the amplitude ratio is dimensionless and the phase difference is a relative value, this technique is self-normalized and sensitive to low-contrast heterogeneity of fluorophore concentration in a turbid medium. In addition, based on the estimated target position and depth, the dual-zone mesh reconstruction scheme has significantly improved the reconstruction accuracy of fluorophore concentration.

Two techniques to regularize the diffuse optical tomography inverse problem were compared for a variety of simulated test domains. One method repeats the single-step Tikhonov approach until a stopping criteria is reached, regularizing the inverse problem by scaling the maximum of the diagonal of the inversion matrix with a factor held constant throughout the iterative reconstruction. The second method, a modified Levenberg-Marquardt formulation, uses an identical implementation but reduces the factor at each iteration. Four test geometries of increasing complexity were used to test the performance of the two techniques under a variety of conditions including varying amounts of data noise, different initial parameter estimates, and different initial values of the regularization factor. It was found that for most cases tested, holding the scaling factor constant provided images that were more robust to both data noise and initial homogeneous parameter estimates. However, the results for a complex test domain that most resembled realistic tissue geometries were less conclusive.

We developed an eight-channel scanning time-domain fluorescence mammograph capable of imaging the distribution of a non-specific fluorescent contrast agent in the female breast, besides imaging intrinsic absorption and scattering properties of healthy breast tissue and tumors. The apparatus is based on the PTB multi-channel laser pulse mammograph, originally designed for measurements of absorption and scattering coefficients at four selected wavelengths (&lgr; = 652 nm, 684 nm, 797nm, and 830 nm). It was upgraded for time-resolved detection of fluorescence, excited at 735 nm by a ps diode laser with 10 mW output power and detected at wavelengths &lgr; ⩾ 780 nm. Cooled PMTs with GaAs photocathodes are used to detect laser and fluorescence photons at five positions in transmission and three positions in reflection. Measurements are performed with the breast being slightly compressed between two parallel glass plates. The transmitting and receiving fiber bundles are scanned synchronously over the breast in steps of typically 2.5 mm. At each scan position, distributions of times of flight of laser photons are measured by time-correlated single photon counting at eight detector positions, followed by measurements of distributions of times of arrival of fluorescence photons. The performance of the fluorescence mammograph was investigated by using breast-like phantoms with a fluorescent inhomogeneity with dye enrichment varying between 2:1 and 10:1 over background values.

Fluorescence enhanced diffuse optical tomography (fDOT) is envisioned to be useful to collect functional information from small animal models. For oncology applications, cancer-targeted fluorescent markers can be used as a surrogate of the cancer activity. We are developing a continuous wave fDOT bench intended to be integrated in systems dedicated to whole body small animal fluorescence analyses. The focus is currently put on the reconstruction of non immersed small animals imaged by a CCD camera. The reconstruction stage already corrects the tissue heterogeneity artifacts through the computation of an optical heterogeneity map. We will show how this formalism coupled with the determination of the animal boundaries performed by a laser scanner, can be used to manage non contact acquisitions. The time of reconstruction for a 10 × 9 laser source positions, 45 × 40 detector elements and 14 × 11 × 14 mesh voxels is typically 10 minutes on a 3GHz PCs corresponding to the acquisition time allowing the two tasks to be performed in parallel. The system is validated on an in vivo experiment performed on three healthy nude mice and a mouse bearing a lung tumor at 10, 12 and 14 days after implantation allowing the follow up of the disease. The 3D fluorescence reconstructions of this mouse are presented and the total fluorescence amounts are compared.

We describe a Monte Carlo test of a new method for the calculation of the relative concentrations of two localized absorbers in a highly scattering medium. The method, previously proposed by us, is based on a property of the intensity changes caused by the localized absorbers, which is strictly verified by diffusion theory within first order perturbation, and remains approximately verified beyond the limits of first order perturbation. The applicability of the method was theoretically and experimentally validated in an infinite medium geometry for "small size" perturbations, as well as for larger perturbations. The purpose of the Monte Carlo test reported here is to show that the method is applicable also to a slab geometry for a relatively large (both in size and optical contrast) perturbation. This test indicates that the dual wavelength approach to the estimation of the relative concentration of two absorbers can be insensitive to the medium geometry as well as the size and geometry of the localized inclusion.

In the course of our experiments imaging the compressed breast in conjunction with digital tomosynthesis, we have noted that significant changes in tissue optical properties, on the order of 5%, occur during our imaging protocol. These changes seem to consistent with changes both in total Hemoglobin concentration as well as in oxygen saturation, as was the case for our standalone breast compression study, which made use of reflectance measurements. Simulation experiments show the importance of taking into account the temporal dynamics in the image reconstruction, and demonstrate the possibility of imaging the spatio-temporal dynamics of oxygen saturation and total Hemoglobin in the breast. In the image reconstruction, we make use of spatio-temporal basis functions, specifically a voxel basis for spatial imaging, and a cubic spline basis in time, and we reconstruct the spatio-temporal images using the entire data set simultaneously, making use of both absolute and relative measurements in the cost function. We have modified the sequence of sources used in our imaging acquisition protocol to improve our temporal resolution, and preliminary results are shown for normal subjects.

A typical perturbation-based image reconstruction technique requires a homogeneous reference for accurate estimation of target-caused perturbation and accurate reconstruction of target optical properties. Therefore reference selection is critical. In this report, we analyze the influence of different references on reconstructed optical images when a typical perturbation approach, such as Born method, is used for imaging reconstruction. A new photon tracking method using Monte Carlo simulation is developed and used to analyze the contributions of a target and a chest-wall layer underneath the breast tissue to the NIR measurements. We have found that a chest-wall layer has much larger contribution than the target when the breast-tissue layer thickness is less than 1.5 to 2.0 cm deep depending on bulk optical properties. Different references, such as optical property-matched and depth-unmatched reference, depth-matched and property-unmatched reference, property-unmatched and depth-unmatched reference, and property-matched and depth-matched reference, are compared and evaluated using the total perturbation and the reconstructed image quality as quantitative criteria. We have found that the property-matched and depth-matched reference provides the best result with error less than 5%, and the depth-matched and property-unmatched reference provides the worst results with error larger than 60%. Phantom experiments confirm with the simulation results. As results, we suggest using the measurements from a normal contralateral site of the same quadrant as the breast lesion with matched propagation depth as best property-matched and depth-matched reference with the help of co-registered ultrasound in clinical trials.

Optical imaging using independent component analysis (OPTICA) is enhanced to provide a high resolution cross section imaging of objects in a turbid medium by a backprojection technique. The performance is demonstrated by imaging a human breast model made of ex vivo human breast tissues. Cancerous site of 5mm size is detected at the midplane of the 33mm thick breast model. The reconstructed cross section image compares favorably with pathology findings.

We have developed a novel method for combining non-concurrent MR and DOT data, which integrates advanced multimodal registration and segmentation algorithms within a well-defined workflow. The method requires little user interaction, is computationally efficient for practical applications, and enables joint MR/DOT analysis. The method presents additional advantages: More flexibility than integrated MR/DOT imaging systems, The ability to independently develop a standalone DOT system without the stringent limitations imposed by the MRI device environment, Enhancement of sensitivity and specificity for breast tumor detection, Combined analysis of structural and functional data, Enhancement of DOT data reconstruction through the use of MR-derived a priori structural information. We have conducted an initial patient study which asks an important question: how can functional information on a tumor obtained from DOT data be combined with the anatomy of that tumor derived from MRI data? The study confirms that tumor areas in the patient breasts exhibit significantly higher total hemoglobin concentration (THC) than their surroundings. The results show significance in intra-patient THC variations, and justify the use of our normalized difference measure defined as the distance from the average THC inside the breast, to the average THC inside the tumor volume in terms of the THC standard deviation inside the breast. This method contributes to the long-term goal of enabling standardized direct comparison of MRI and DOT and facilitating validation of DOT imaging methods in clinical studies.

NIR tomography image reconstruction can be improved by incorporating spectral constraints and prior spatial information. The convergence of scattering power was studied based on the distribution of projection error with different parameters. The reason that scattering properties are harder to recover than chromophore concentrations was discussed. Using "hard prior" spectral reconstruction, the role of stopping criteria was found to be important. Multiple wavelength simulations were used to choose suitable stopping criteria. Preliminary tests using a wavelength tunable Ti-Sapphire laser shows promise for frequency domain measurements covering a wide range of wavelengths.

It is expected that the optical signatures of physiological changes are biomarkers reacting faster to breast tumor evolution than structural changes, meaning that diffuse optical tomography (DOT) could be a promising modality for monitoring and detecting early changes of the lesion during neoadjuvant treatment. Numerous publications as well as our preliminary results revealed that the heterogeneity inside the breast and the variability within the population are challenging for such application. Moreover the sensitivity of the breast physiology to the external pressure applied during data acquisition is adding a significant variance to the process. In the present study we evaluate key factors that could make neoadjuvant treatment monitoring, using DOT, successfully: 1) sensitivity-the clue for earlier detection; 2) repeatability-minimizing the impact of the artificially induced variance (related with pressure, angle of the view, etc.); 3) accurate co-localization of the ROI within the sequential measurements performed during the neoadjuvant treatment. Non-clinical and clinical studies were performed using a multi-wavelength time-domain platform, with transmission detection configuration, and 3D images of optical and physiological properties were generated using diffuse propagation approximation. The results of non-clinical studies show that the sensitivity of the system allows detection and quantification of absorption changes equivalent to less than 1 micromole of blood. Clinical studies, involving more than 40 patients, revealed that with the appropriate precautions during patient positioning and compression adjustment, the repeatability of the results is very good and the similarities between the two breasts are high suggesting that the contra-lateral breast could be used as a reliable reference for DOT as well.

Differences in tissue water state have been measured in normal and malignant breast tissues. Broadband Diffuse Optical Spectroscopy (DOS) has been used to acquire 650-1000 nm absorption spectra of normal and tumor breast tissues from 7 patients in vivo. The absolute values of spectral differences between normalized tissue water spectra and pure water spectra were combined and divided by the number of points in the sum to form the bound water index (BWI). In all subjects, the average BWIs of line scan points were significantly lower in tumor tissues (1.62 ± 0.27 x10^-3) than normal tissues (3.06 ±0.51 x10^-3, Wilcoxon Ranked Sum Test z= 0.003 and power=0.98). These results imply that the water in tumors behaves more like free water than the water in normal tissue.

Under body O₂ imbalance, the Autonomic Nervous System is responsible for redistribution of blood flow with preference to the most vital organs (brain, heart), while the less vital organs (intestine, GI tract) are hypoperfused. The aim of this study was to develop and use an animal model for real time monitoring of tissue viability in the brain, and the small intestine, under various levels of oxygen and blood supply. Male Wistar rats were anesthetized, the brain cortex and intestinal serosa were exposed and connected by optical fibers to the Multi-Site Multi-Parametric (MSMP) monitoring system. Tissue blood flow (TBF) and mitochondrial NADH redox state were monitored simultaneously in the two organs. The rats were subjected to short anoxia, 20 minutes hypoxia or epinephrine (2& 8&mgr;g/kg I.V.). Under oxygen deficiency, cerebral blood flow (CBF) was elevated, whereas intestinal TBF was reduced. Mitochondrial NADH was significantly elevated in both organs. Systemic injection of Adrenaline showed a dose-depended increase in systemic blood pressure and CBF response whereas, intestinal TBF similarly decreased in both doses. In addition, NADH was elevated (reduced form) in the intestine whereas oxidation was observed in the brain. In conclusion, our preliminary results may imply the ability of using of the MSMP for monitoring non-vital organs in order to detect early changes in the balance between oxygen supply and demand in the body.

To investigate the temperature dependence of intrinsic optical signals (IOSs) relating to brain tissue viability, we performed simultaneous measurement of light absorption due to the redox states of cytochrome c oxidase and light scattering, which reflects morphological characteristics of cells in tissue, for rat brains after blood removal by saline infusion at different infusion temperatures. To determine IOSs at each temperature, we first examined an isosbestic wavelength of the redox states of cytochtome c oxidase for each rat based on multiwavelength diffuse reflectance measurement. We then measured diffuse reflectance intensity at the isosbestic wavelength as a scattering signal, while diffuse reflectance intensity at 800 nm was detected to monitor the reduction of CuA in cytochrome c oxidase. At all temperatures, the scattering signal was steady in an early phase but showed a drastic, triphasic change in a certain time range of infusion; during this scattering change, the reduction of CuA started and proceeded rapidly. The start time of triphasic scattering change as well as the start time of the reduction of CuA was extended for more than 2 min by lowering infusion temperature from 30 to 23°C and we found that there was a linear correlation between these two start times. These results suggest that tissue metabolic activity can be maintained for longer time by keeping the brain at lower temperature, and triphasic scattering change can be used as an optical signal indicating the reduction of CuA in cytochrome c oxidase, and hence loss of tissue viability for brain.

We demonstrate the use of diffuse optical imaging via transillumination to detect cancerous tissue in a rat animal model. In this imaging modality infrared radiation is transmitted through whole animal tissue. The radiation is nonionizing and uses endogenous contrast: namely deoxyhemoglobin (Hb) and oxyhemoglobin (HbO). Differential image analysis is performed to visualize the presence of cancerous tissue. Varying levels of inspired air and carbogen gases ensure a differential response in absorption by blood due to changing levels of Hb and HbO. We believe that this response may be sufficient to provide contrast in differential image analysis. The present method also sheds light on physiological challenges in whole animal imaging especially with respect to significant optical signals from healthy tissue. Specifically, we have seen strong signals from abdominal regions in normal rats indicative of diet related anomalous transmission. We have also been able to track the changes in optical signal during animal death.

In this paper we present a novel application of digital detection and data-acquisition techniques to a prototype dynamic optical tomography system. The core component is a digital signal processor (DSP) that is responsible for collecting and processing the digitized data set. Utilizing the processing power of the DSP, real-time data rates for this 16-source, 32-detector system, can be achieved at rates as high as 140Hz per tomographic frame. Many of the synchronously-timed processes are controlled by a complex programmable logic device (CPLD) that is used in conjunction with the DSP to orchestrate data flow. The operation of the instrument is managed through a comprehensive graphical user interface, which was designed using the LabVIEW software package. Performance analysis demonstrates very low system noise (~.60pW RMS noise equivalent power) and excellent signal precision (<0.1%) for most practical cases. First experiments on tissue phantoms show that dynamic behavior can be accurately captured using this system.

In this study, we report the development of a near infrared broadband, steady-state, multi-channel imaging system to quantify hemodynamic parameters, which is to be used for measuring the rat brain tumors in vivo. The imager was calibrated with laboratory phantoms to eliminate spectral effects of the CCD spectrometers, optical fibers, and multiplexer channels. The calibration procedures also help determine the source strength in order to obtain accurate image reconstructions. A multi-channel, multi-wavelength, spectrally constrained reconstruction algorithm is under development to obtain tomographic maps of concentrations of hemoglobin derivatives and reduced scattering coefficients. The developed imaging system and reconstruction algorithm were validated with dynamic multi-tube phantoms.

We have developed an efficient Levenberg-Marquardt iterative algorithm utilizing a three-dimensional field measurements coupled to a two-dimensional optical property reconstruction scheme. This technique takes advantage of accurate estimation of light distribution in 3D forward calculation and reduced problem size and less computation time in 2D inversion. Important advances in terms of improving algorithm efficiency and accuracy include use of an iterative general minimum residual method (GMRES) for computing the field solutions, application of the dual mesh scheme and adjoint method for Jacobian construction, and implementation of normalization scheme to reduce the absorption-scattering cross talk. The synthetic measurement data were calculated for a cubic phantom containing a single absorption anomaly and a single scattering anomaly. The model had a background of &mgr;_a=0.03mm^-1 and &mgr;_s=1.4mm^-1. The absorption and scattering anomalies have the &mgr;_a = 0.06 mm^-1 and &mgr;_s' = 2.0 mm^-1. Five sources and 72 detectors are used per slice. A typical human prostate is composed of 6 slices. The reconstruction images successfully recover the both anomalies with good localization. Experiment data from tissue simulated phantom are also presented. The clinical DOT imaging was performed before photodynamic therapy based on the protocol. The preliminary results showed the reconstructed prostate &mgr;_a varied between 0.025 and 0.07 mm^-1 and &mgr;_s' ranged from 1.1 to 2 mm^-1. These results show that this new 2D-3D hybrid algorithm consistently outperform the 2D-2D or 3D-3D counterparts.

The cardiovascular system is designed to deliver oxygen to every cell in the body through the microcirculation. Optical Reflectance Spectroscopy (ORS) is a powerful tool used to study oxygen delivery through vessels less than 50 &mgr;m in diameter. Depth analysis can be achieved by varying the geometry of the incident light source and the detector of the back-scattered light. A fibre optic probe has been designed with spacings to study the capillary loops and microvessels of the skin. Similarly, Near Infrared Spectroscopy (NIRS) can directly measure haemodynamics in muscle. A combined study of ORS and NIRS is currently investigating the relationship of vasomotion in the skin and underlying muscle. Vasomotion is usually defined as rhythmic changes in the diameter of the small blood vessels and has been linked to both endothelial and sympathetic activity. It has been suggested that vasomotion in the muscle preserves nutritive perfusion not only in the muscle itself but also to neighbouring tissue i.e. skin. ORS and NIRS can provide a direct measure of these changes in blood volume. At frequencies linked with endothelial and sympathetic activity, rhythmical oscillations in blood volume of the same magnitude, were demonstrated in both skin and muscle, 15.3(4.0)% skin vs 16.3(5.3)% muscle for endothelial frequencies, (mean(SD), t-test, p=0.633) and 10.9(3.8)% skin and 12.4(5.5)% muscle for sympathetic frequencies (p=0.354). These data demonstrate the potential of these optical techniques to enable simultaneous examination of microvascular haemodynamics in two tissue types.

In this study, we showed that exercise type- and intensity-dependent regional differences in muscle oxygenation and oxygen consumption rate (Vo₂) of the knee extensor muscles could be imaged in real time with a multi-channel spatially resolved near-infrared spectroscopy (SR-NIRS) imaging device. Healthy subjects performed isometric knee extension exercise for 30 s (without- or with-leg-press action) at different exercise intensities [10%, 40% and 70% of maximum voluntary contraction (MVC)]. "Separation-type" probes were attached to the skin over the major knee extensor muscles: vastus lateralis (VL), rectus femoris (RF) and vastus medialis (VM). Placement of the probes enabled simultaneously measurement of 12 sites over a skin area of about 30 cm² (temporal resolution = 0.25 s). Local Vo₂ of each muscle, resting Vo₂ (Vo₂, rest) and recovery Vo₂ (Vo₂, rec ), were determined with arterial occlusion before the start and after the end of contraction, respectively. There was no significant difference between the values of Vo₂ rest, in the muscles. However, during knee extension exercise without-leg-press action, Vo₂ rec, value of the RF was significantly greater than the values of the VL and VM at all exercise intensities. In contrast, during exercise with-leg-press action, Vo₂ rec, values of the RF and VM were greater than those of the VL, especially during exercise at 40% and 70% MVC. In summary, the regional differences in muscle oxygenation and Vo₂ of the knee extensor muscles, probably due to the differences in relative contributions of muscles to exercise and in muscle architecture, were imaged using SR-NIRS.

When numerically implementing the equation of radiative transfer (ERT) to calculate light propagation in biological tissue, one has several choices on how to perform the spatial discretization. In this study we investigate the performance of two commonly employed differencing schemes ("step" and "weighted diamond"). Using a discrete-ordinates finite-volume method in a two-dimensional absorbing and scattering medium, the code performances are evaluated in terms of accuracy and computational requirement. We find that, compared to the step-differencing scheme, the weighted diamond differencing scheme provides more accurate solutions of the radiation intensity over a wide range of optical properties. Furthermore, the weighted diamond scheme is computationally more efficient than the step method. When used in conjunction with tomographic reconstruction algorithms, we observe that using the weighted-diamond differencing scheme leads to more accurate reconstructions of the optical properties.

In this report, a phase-contrast diffuse optical tomography system, which can measure the refractive indices of human breast masses in vivo, is described. To investigate the utility of phase-contrast diffuse optical tomography (PCDOT) for differentiation of malignant and benign breast masses in humans, and to compare PCDOT with conventional diffuse optical tomography (DOT) for analysis of breast masses in humans. 35 breast masses were imaged in 33 patients (mean age = 51 years; range 22-80 years) using PCDOT. Images characterizing the tissue refractive index, absorption and scattering of breast masses were obtained with a finite element-based reconstruction algorithm. The accuracies of absorption and scattering images were compared with images of refractive index in light of the pathology results. Absorption and scattering images were unable to accurately discriminate benign from malignant lesions. Malignant lesions tended to have decreased refractive index allowing them to discriminate from benign lesions in most cases. The sensitivity, specificity, false positive value, and overall accuracy for refractive index were 81.8%, 70.8%, 29.2%, and 74.3%, respectively. Overall we show that benign and malignant breast masses in humans demonstrate different refractive index and differences in refractive index properties can be used to discriminate benign from malignant masses in patients with high accuracy. This opens up a new avenue for improved breast cancer detection using NIR diffusing light.

The ability to determine the age of a bruise of unknown age mechanism is important in matters of domestic and child abuse and forensics. While physicians are asked to make clinical judgment on the age of a bruise using color and tenderness, studies have shown that a physicians estimate is highly inaccurate and in cases no better than chance alone. We present here the temporal progression of reflection spectrum collected from accidentally inflicted contusions in adult and child study participants with a synopsis of the observed phenomena. Reflection spectra collected using a portable fiber optic reflection spectrometer can track the increase in extravasated hemoglobin from trauma caused blood vessel rupture and subsequent removal of this hemoglobin occurring concurrent with an increase in the absorption attributed to the breakdown product bilirubin. We hypothesize that this time dependent pattern can be used to determine the age of an unknown bruise in an individual provided rate constant information for the patient can be determined in a controlled calibration bruise. Using reflection spectra to estimate bruise age can provide a rapid and noninvasive method to improve the ability of physicians in dating the age of a contusion.

In this study, we measured the optical properties of optical tissue phantoms and human tissues by measuring the photon density distribution on the boundary of the phantoms and tissues and by using the Frequency-Domain DOT (Diffuse optical tomography) system. A trust region nonlinear optimization method is used for the image reconstruction in DOT. This method is a more elaborated Newton method which takes advantage of the classical Newton method and the gradient descent method. The validity of this method is verified by reconstructing the optical properties from phantom experiment and simulated data using the Frequency-Domain DOT system.

Functional Near-Infrared Spectroscopy (fNIRS) is a powerful non-invasive method to measure hemodynamic changes in tissues, while electroencephalography (EEG) provides excellent neuronal-electrical information in functional brain mapping. We developed an fNIRS/ERP instrument which can concurrently monitor the electrical neuronal activation and the hemodynamic response for the study of neurovascular coupling of the human brain. The probe of the instrument consists of 1 LED operating at three different wavelengths (735 nm & 805 nm & 850 nm) and 2 photodiodes (PDs) with the interoptode distance up to 3 cm, and an Ag-AgCl electrode lies in the center of the LED-PD pairs. The four components are encapsulated in a black sponge to decrease the interference of outside light as well as facilitate the placement on the forehead. The signals from the PDs and the electrode separately pass though two adjustable amplify-filter circuits which amplify the weak signals and block the high frequency interferences. A high speed data acquisition board samples the modulated signals under the control of a home-made software. The time resolution of the instrument achieves less than 10ms, which makes it realizable to compare the fNIRS data with the ERP data.

It is well know that the inverse problem in optical tomography is highly ill-posed. The image reconstruction process is often unstable and non-unique, because the number of the boundary measurements data is far fewer than the number of the unknown parameters (optical properties) to be reconstructed. To overcome this problem one can either increase the number of measurement data (e.g. multi-spectral or multi-frequency methods), or reduce the number of unknows (e.g. using prior structural information from other imaging modalities). In this paper, we introduce a novel approach for reducing the unknown parameters in the reconstruction process. The discrete cosine transform (DCT), which has long been used in image compression, is here employed to parameterize the reconstructed image. In general, only a few DCT coefficient are needed to describe the main features in an image, and the number of unknowns in the image reconstruction process can be drastically reduced. Numerical as well as experimental examples are shown that illustrate the performance of the new code.

In this paper, a Monte Carlo based model of diffuse reflectance is introduced to retrieve the light absorption and scattering coefficients of multi-layered tissue. The model consists of a forward part and an inverse one. The forward refers to a multi-layered Monte Carlo simulation at a certain source-detector separation, which lead to a time-resolved profile. The inverse one is a fitting procedure based on Levenberg-Marquardt algorithm using forward model in iteration, and the intensity in time bins is fitted. The model was tested with independent Monte Carlo simulations and fiber-optic-based measurement in the deduced absorption and reduced scattering coefficients. Although a high amount of computation time is needed, this model is developed as a verification standard to these investigation methods using diffuse approximation to the transport equation. Furthermore, it is adaptable to the complex fiber-optic probe geometries.

In the framework of perturbation approach to diffusion equation analytical expressions are derived to describe the effects on the time-resolved transmittance due to the presence of an inclusion that can be purely absorptive or diffusive. The formula assume the optical properties of the inclusion are spatially Gaussian distributed. The accuracy and the application range of the perturbed transmittance are investigated through comparisons with the numerical solutions of the time-dependent diffusion equation given by using the Finite Element Method. A case of practical interest for two-dimensional breast imaging is considered.

In this article, we briefly described a time-correlated single photon counting (TCSPC) system specifically designed for extracting the optical properties in turbid medium. This system was evaluated by use of two sets of liquid tissue-simulating phantoms containing different concentrations of Intralipid-10% as scatters and India ink as absorbers. With the distribution of times of flight (DTOF) of photons measured by the TCSPC system, some featured parameters, such as the mean time of flight and the variance of DTOF were calculated. Based on these parameters, we developed a simple and fast method to obtain the absorption coefficient &mgr;_a and reduced scattering coefficient &mgr;_s' of the turbid medium. Furthermore, the accuracy of the method was validated using the Monte Carlo simulations. It was found that the optical properties could be extracted with this method, which was much faster than the conventional curve-fitting procedure. Our method could be useful in on-line monitoring of optical properties and of the oxygen saturation (SaO₂) in forearm muscle.

Fluorescence diffuse optical tomography (DOT) has attracted many attentions from the community of biomedical imaging, since it provides effective enhancement in imaging contrast. This modality is now rapidly evolving as a potential means of monitoring molecular events in small living organisms with help of molecule-specific contrast agents, referred to as fluorescence molecular tomography (FMT). FMT could greatly promote pathogenesis research, drug development, and therapeutic intervention. Although FMT in steady-state and frequency-domain modes have been promisingly demonstrated, the extension to time-domain scheme is imminent for its several unique advantages over the others. By extending the previously developed generalized pulse spectrum technique (GPST) for time-domain DOT, we propose a linear, featured-data image reconstruction algorithm for time-domain FMT that can simultaneously reconstruct both fluorescent yield and lifetime images, and validate the methodology with simulated data.

A full time-resolved (TR) scheme that has been previously developed for diffuse optical tomography is extended to time-domain fluorescence diffuse optical tomography regime, based on the finite-element-finite-time-difference diffusion modeling and the Newtown-Raphson inversion method. We validate the proposed methodology using simulated data and demonstrate its capability of simultaneously recovering the fluorescent yield and lifetime, as well as its superiority of improving quantitative accuracy and spatial resolution of reconstruction to the featured-data one. The full time-resolved scheme helps set up the 'gold standard' for evaluating the performance of the other featured-data ones, and is more practically feasible in molecular imaging than in generic diffuse optical tomography due to availability of the difference measuring.

Noninvasive molecular imaging of amyloid plaques in murine Alzheimer's disease models would accelerate drug development and basic Alzheimer's research. Amyloid plaques differ from traditional fluorescent targets in size and spatial distribution and therefore present a unique challenge for biomarker development and tomography. To study imaging feasibility and establish biomarker criteria, we developed a digital mouse head model from a 100 &mgr;m-resolution, digital, segmented mouse atlas¹. The cortical region of the brain was filled with a spatially uniform distribution of plaques that had different fluorescent properties from the surrounding brain tissue, similar to current transgenic mouse models of Alzheimer's disease. Fluorescence was simulated with a Monte Carlo algorithm using different plaque densities, detection geometries, and background fluorescence. Our preliminary results demonstrated that shielding effects might require nonlinear reconstruction algorithms and that background fluorescence would seriously hinder quantitative burden estimation. The Monte Carlo based approach presented here offers a powerful way to study the feasibility of non-invasive imaging in murine Alzheimer's models and to optimize experimental conditions.

Indocyanine Green (ICG) is currently the only FDA-approved contrast agent suitable for imaging tumor vascularity and performing permeability measurements. However, it is non-specific; clearance (wash-out) by the liver is very rapid, and tumor to normal tissue contrast is not optimal for medical imaging applications. Therefore, new ICG derivatives are being developed with improved affinity tumor cell affinity, and prolonged circulation times.¹ Furthermore, several new contrast agents (molecular beacons) with specific tumor targeting have been reported. Tumor cells over-express certain receptors which results in an increased uptake of ligands specific to those receptors. Chemical conjugation of molecular beacons to such ligands allows for accumulation of the agents specifically at tumor cites. For example, Weissleder et al.² have developed protease-activated molecular beacons that achieved a 12-fold tumor to normal tissue contrast. New molecular beacons might have direct impact on therapy monitoring: with higher sensitivity and specificity, one can noninvasively monitor and therapeutically intervene tumors in their early stages, which should lead to better survival rates.

The interest in fluorescence imaging has increased steadily in the last decade. Using fluorescence techniques, it is feasible to visualize and quantify the function of genes and the expression of enzymes and proteins deep inside tissues. When applied to small animal research, optical imaging based on fluorescent marker probes can provide valuable information on the specificity and efficacy of drugs at reduced cost and with greater efficiency. Meanwhile, fluorescence techniques represent an important class of optical methods being applied to in vitro and in vivo biomedical diagnostics, towards noninvasive clinical applications, such as detecting and monitoring specific pathological and physiological processes. ART has developed a time domain in vivo small animal fluorescence imaging system, eXplore Optix. Using the measured time-resolved fluorescence signal, fluorophore location and concentration can be quickly estimated. Furthermore, the 3D distribution of fluorophore can be obtained by fluorescent diffusion tomography. To accurately analyze and interpret the measured fluorescent signals from tissue, complex theoretical models and algorithms are employed. We present here a numerical simulator of eXplore Optix. It generates virtual data under well-controlled conditions that enable us to test, verify, and improve our models and algorithms piecewise separately. The theoretical frame of the simulator is an analytical solution of the fluorescence diffusion equation. Compared to existing models, the coupling of fluorophores with finite volume size is taken into consideration. Also, the influences of fluorescent inclusions to excitation and emission light are both accounted for. The output results are compared to Monte-Carlo simulations.