Quantitative Biophotonics for Tissue Characterization and Function

Amir H. Gandjbakhche, PhD, Head, Section on Analytical and Functional Biophotonics
Franck Amyot, PhD, Research Fellow
Moinuddin Hassan, PhD, Research Fellow
Victor Chernomordik, PhD, Staff Scientist
Jason Riley, PhD, Postdoctoral Fellow
Alex Small, PhD, Postdoctoral Fellow
Alexander Sviridov, PhD, Guest Researcher¹
Jana Kainerstorfer, MS, Predoctoral Fellow
Abby Vogel, MS, Predoctoral Fellow²
Nika Bejou, Summer Student³
Zachary Ulissi, Summer Student⁴

Our objectives are to devise quantitative biophotonics methodologies and associated instrumentations in order to study (1) biological phenomena at different length scales—from nanoscopy to microscopy—and (2) diffuse biophotonics. We take advantage of our expertise in stochastic modeling to study complex biological phenomena whose properties are characterized by elements of randomness in both time and space, such as light-tissue interactions. We explore various properties of light-matter interactions as sources of optical contrasts, such as polarization properties, endogenous or exogenous fluorescent labels, absorption (e.g., hemoglobin or chromophore concentration), and/or scattering. We have used these contrast mechanisms for tomographic and spectroscopic methods to develop benchtop instrumentation for preclinical and clinical uses. We are identifying physiological sites where optical techniques might be clinically practical and offer new diagnostic knowledge and/or less morbidity than existing diagnostic methods.

Quantitative characterization of tissue

Sviridov, Ulissi, Chernomordik, Gandjbakhche; in collaboration with Hebden, Weiss, Smith

Many components of biological tissues—such as collagen, muscle fibers, keratin, retina, and glucose—demonstrate polarization properties attributable to their anisotropy. As a result, depolarization of initially polarized light is observable and strongly dependent on the tissue’s bulk optical properties. Mapping the degree of polarization may yield valuable diagnostic information about the superficial and subsurface structures of the skin and other tissues. A statistical analysis of data can exclude distortions in the polarization degree maps and enhance the observability of hidden structures, permitting quantification of the structures’ characteristic scales and directionality.

We use the pattern of the surface distribution of the degree of polarization to detect and analyze structures of initial skin fibrosis at the early stages of abnormality when it is barely visible. We devised a new data-processing algorithm that involves Pearson correlation analysis (Sviridov et al., J Biomed Opt 2005;10:051706). The technique reduces blurring of features by unpolarized backscattered light, permitting visualization of regions of high statistical similarities within the noisy tissue images. We are developing a hand-held camera for polarization imaging of biological tissue structures. Phantom experiments have demonstrated the ability of the designed system to visualize the collagen network at the scale of 10 to 20 µ. One potential application of the methodology is for characterizing the state of the uterus to predict preterm birth. In collaboration with physicians, we plan to analyze the results of in vivo tests to optimize the potential diagnostic content of the data.

Sviridov AP, Ulissi Z, Chernomordik V, Hassan M, Boccara AC, Gandjbakhche AH. Compact polarization camera with liquid-crystal retarder for patterning of biological textures. Biomedical Optics, OSA Technical Digest (CD) (Optical Society of America, 2008), Paper BTuF48.
Sviridov AP, Ulissi Z, Chernomordik V, Hassan M, Gandjbakhche AH. Visualization of biological texture using correlation coefficient images. J Biomed Opt Lett 2006;11:060504.

Preclinical studies on fluorescence imaging of HER2-positive breast cancer

Hassan, Riley, Chernomordik, Gandjbakhche; in collaboration with Capala, Lee, Gannot, Smith

Advances in the molecular biology of disease processes, new immunohistopathological techniques, and the development of fluorescently labeled cell surface markers have led to a revolution in (1) specific molecular diagnosis of disease by histopathology and (2) research into the molecular origins of disease processes. As a result, an exceptional level of specificity is now possible in the design of exogenous markers, permitting molecules to be engineered to bind only to specific receptor sites in the body. Such receptor sites may be antibodies or other biologically interesting molecules. Fluorophores bound to the engineered molecules preferentially concentrate at the sites of interest after injection. Such concentration parallels the delivery of some novel drugs, e.g., Trastuzumab, to the diseased site and opens a new era of the “treat and image” paradigm.

To optimize the course of the treatment of human epidermal growth factor receptor 2 (HER2)–positive breast cancer (and exclude from HE R2-based chemotherapy those patients with other forms of breast carcinoma), there is a need for molecular probes capable of in vivo monitoring of a receptor downregulated as an immediate response to such therapy. In collaboration with Jacek Capala, we developed fluorescent markers, based on HER2-specific affibody molecules, for in vivo monitoring of the level of overexpression of HER2. These small molecules possess high affinity for HER2 and do not compete with trastuzumab because they target a different epitope of the receptor, making them an appropriate choice as a contrast agent. The potential application is based on fluorescence intensity distributions originating from specific markers in the tumor area. The measured time course of marker variation after injection of the contrast agent can provide information on the status of HER2 expression in the tumor cells—information that may guide adjustments to the treatment regimen. For preclinical studies, we used a mouse model with xenografts of various carcinomas implanted into the mouse flank and tested two types of affibodies: albumin-binding domain (ABD) and monomer. Results showed that ABD performed better based on both marker accumulation in the tumor and reasonable clearance times. The developed methodology allows us to investigate the pharmacokinetics of the contrast agent and potentially assess the status of HER2 receptors. We have initiated a comparative study in the mouse model for several types of breast carcinoma (BT474, MDA-MB361, U251, and MCF7) that are characterized by different levels of amplification and overexpression of HER2.

Hassan M, Riley J, Chernomordik V, Smith P, Pursley R, Lee SB, Capala J, Gandjbakhche A. Fluorescence lifetime imaging system for in-vivo studies. Mol Imaging 2007;6:229-236.
Lee SB, Hassan M, Fisher R, Chertov O, Chernomordik V, Marek GK, Gandjbakhche A, Capala J. Affibody molecules for in vivo characterization of HER2-positive tumors by near-infrared imaging. Clin Cancer Res 2008;14:3840-3849.

Fluorescence life-time imaging

Hassan, Riley, Chernomordik, Gandjbakhche; in collaboration with Capala, Gannot, Smith

With specially designed fluorophores, it is possible to probe environmental conditions (e.g., pH) at a given site, based on local variations in fluorophore life time (the time required for an electron to return from an excited state to its initial state). To use fluorophores to target abnormalities that are likely to be located deep inside biological tissues, we developed reconstruction algorithms that take into account the highly scattering nature of the medium. Using our well-established small animal time-resolved imaging system, along with a quantitative time-resolved fluorescence diffuse inverse method, we showed that, for tissue-like phantoms, we could reconstruct intrinsic fluorescence life times from targets deeply embedded in a turbid medium, e.g., tissue. Three-dimensional localization of the fluorophores with the life-time mapping may be used to characterize tumor status; the technique provides useful clinical information, e.g., pH maps obtained with pH-sensitive dyes, and can help localize cancer cells and monitor treatment.

To investigate three-dimensional reconstruction algorithms, we used generalized analytical formulas for excitation/emission photon migration as a forward model. Experimental data obtained with our time-resolved instrumentation confirmed limitations of data analysis because of noise. For example, a low signal-to-noise ratio for photon times of flight greater than 5 to 6 ns precludes use of this time range to quantify fluorescence characteristics. The accuracy of the three-dimensional reconstruction critically depends on the noise sensitivity of input parameters. We have suggested that a new local set of data types is likely to provide more stability to noise than classical statistical moments (global data types). The equivalent forms of the moments in the “local” data types, based on the geometry (shape) of the temporal point spread function (TPSF), are the peak value, peak point, full width at half maximum (FWHM), and slope of the curve. To compare various data types, we created a numeric model of time-resolved fluorescence from fluorophores inside a turbid medium and showed that the local data types were much more resilient to noise than their standard global counterparts. After demonstrating this as the appropriate noise type versus the expected Poisson noise, we performed the analysis by using Gaussian noise.

To test further the relative merits of the two sets of data types (global and local), we developed a generic reconstruction algorithm that allowed substitution of the data types. Though initial results for mathematical models and phantoms are encouraging, we need to investigate distributed fluorescent dyes. When life times were comparable to photon migration times, we have found scaling relationships for time-of-flight distributions of emission photons at different depths. Our phantom data substantiate the findings. Taking into account that background optical parameters of the medium may be estimated from multiple-source detector data sets, we suggested a multistep methodology of intrinsic life-time quantification based on the scaling relationships.

Chernomordik V, Hassan M, Amyot F, Riley J, Gandjbakhche A. Use of scaling relations to extract intrinsic fluorescence lifetime of targets embedded in turbid media. J Biomed Opt 2008;13:024025.
Hassan M, Riley J, Chernomordik V, Smith P, Pursley R, Lee SB, Capala J, Gandjbakhche A. Fluorescence lifetime imaging system for in-vivo studies. Mol Imaging 2007;6:229-236.
Riley J, Hassan M, Chernomordik V, Gandjbakhche A. Choice of data-types in fluorescence enhanced diffuse optical tomography. Med Phys 2007;34:4890-4900.

Using multimodality imaging techniques to monitor tissue vasculature in Kaposi’s sarcoma lesions

Hassan, Vogel, Riley, Chernomordik, Kainerstorfer, Gandjbakhche; in collaboration with Demos, Yarchoan, Binzoni

An intense effort is currently under way to develop novel targeted therapies for use in a variety of diseases, with a strong focus on adequate methods to monitor responses that correlate with or even predict the outcome of illnesses of interest. We have developed and are using three non-invasive imaging techniques: near-infrared multispectral imaging, thermography, and laser Doppler imaging (LDI). Combining multispectral imaging and LDI allows us to determine changes in blood volume, oxygenation state, and blood velocity of the microvasculature and location of an abnormality.

The near-infrared multispectral imaging system used in our laboratory was designed and built in collaboration with Lawrence Livermore National Laboratory. To exclude specular reflection from the skin, we detect photons in cross-polarized mode by using a pair of polarizers with orthogonal orientations in the illumination and detection arms. A spectral filter placed in front of a CCD⁵ camera permits the consecutive acquisition of six wavelength images (700, 750, 800, 850, 900, and 1,000 nm). With our two-layer skin model, we then use the images in a reconstruction of the distribution of analytes such as oxy- and deoxy-hemoglobin and melanin in the tissue. To exclude non-physiological contributions to the image, we developed a curvature-correction, rigid-registration algorithm that uses a Canny edge detector to extract the edges of the imaged body part and to make alignment more accurate. Our reconstruction algorithm allows us to extract physiological information from spectral images and has undergone substantiation by Monte Carlo simulations of photon migration. We have shown that expected differences in photon trajectories attributable to variations in tissue optical parameters at considered wavelengths result in minor errors in the quantification of physiological parameters such as blood volume/oxygenation. We have demonstrated that combining the three non-invasive imaging systems (multispectral imaging, thermography, and LDI) provides clinically useful information to evaluate follow-up treatment efficiency in Kaposi’s sarcoma patients.

Binzoni T, Vogel A, Gandjbakhche A, Marchesini R. Detection limits of multi-spectral optical imaging under the skin surface. Phys Med Biol 2008;53:617-636.
Lin B, Chernomordik V, Gandjbakhche A, Matthews D, Demos S. Investigation of signal dependence on tissue thickness in near infrared spectral imaging. Optics Express 2007;15:16581-16595.
Vogel A, Chernomordik VV, Hassan M, Amyot F, Dasgeb B, Demos SG , Pursley R, Little RF, Yarchoan R, Tao Y, Gandjbakhche AH. Using noninvasive multispectral imaging to quantitatively assess tissue vasculature. J Biomed Opt 2007;12:051604.
Vogel A, Dasgeb B, Hassan M, Amyot F, Chernomordik V, Tao Y, Demos SG, Wyvill K, Aleman K, Little R, Yarchoan R, Gandjbakhche AH. Using quantitative imaging techniques to assess vascularity in AIDS-related Kaposi’s sarcoma. Conf Proc IEEE Eng Med Biol Soc 2006;1:232-235.

Cellular dynamics of angiogenesis

Amyot, Small, Ulissi, Gandjbakhche; in collaboration with Sackett, Boukari, Plant, Camphausen, Neagu

Much debate remains on which form of vascular endothelial growth factor (VEGF) guides chemotactic migration of cells to form a capillary network and on the roles played by the extracellular matrix and matrix metalloproteases (MMP). Given that growth factor gradients guide chemotactic migration, we devised a system of reaction-diffusion equations to model VEGF’s diffusion, binding, and cleavage in vivo. Our simulations showed that rapid binding to the matrix led to short-range gradients of matrix-bound VEG F, whereas slower binding led to longer-range gradients. Our findings are consistent with in vivo observations that the vasculature is disorganized around tumors that produce the fast-binding isoforms of VEGF. Only the slowly binding forms of VEGF produce long-range gradients that can guide cell migration to form an efficient and organized network. We also found that the cleaved form of VEGF, removed from the matrix by the action of MMPs, is distributed with a gradient that points away from the tumor, calling into question whether the cleaved form of VEGF plays a chemotactic role in tumor-induced angiogenesis. Our observation is consistent with the findings that cleaved VEGF molecules have a weaker chemotactic effect on receptors. Our simulations showed that, without MMPs, a gradient of matrix-bound VEGF cannot be sustained, consistent with findings that the production of MMPs by cells near the parent capillary is necessary to initiate the formation of new vasculature.

Growth factor distribution, MMP release, and cell migration, differentiation, and aggregation are major components of tumor-induced angiogenesis. We have developed a model that includes these phenomena and the interaction of endothelial cells (EC) with the extracellular matrix (ECM). ECs switch among growth, differentiation, motility, and apoptotic behavior in response to the ECM’s local topology and composition. Based on the assumption that the ECM medium is a statistically inhomogeneous medium (not all areas support sprout growth), we showed that the ECM can be a natural barrier to angiogenesis. We studied vascular network formation for several ECM distributions and topologies and found an analogy with percolation theory in network formation. A threshold exists under which sprouts cannot reach the tumor. During the growth of the vascular network, attraction exerted by the tumor competes with the preferred path created by the ECM. We also examined the influence of branching on tumor vascularization; branching is a natural phenomenon that helps the tumor become vascularized. By increasing the number of sprouts, the vascular network increases the probability of reaching the tumor, as it can explore more pathways. Our simulations showed that, after two branching events, the vascular network is highly likely to reach the tumor.

Amyot F, Small A, Boukari H, Camphausen K, Gandjbakhche A. Topology of the heterogeneous nature of the extracellular matrix on stochastic modeling of tumor-induced angiogenesis. Microvasc Res 2008 [E-pub ahead of print].
Amyot F, Small A, Boukari H, Sackett D, Elliott J, McDaniel D, Plant A, Gandjbakhche A. Thin films of oriented collagen fibrils for cell motility studies. J Biomed Mater Res B Appl Biomater 2008;86B:438-443.
Small A, Amyot F, Neagu A, Sackett D, Chernomordik V, Gandjbakhche A. Spatial distribution of VEGF isoforms and chemotactic signals in the vicinity of a tumor. J Theor Biol 2008;252:593-607.

Virtual quantification of confocal microscopy systems

Riley, Gandjbakhche; in collaboration with Combs, Smirnov, Knutson, Balaban, Nossal, Boukari

In collaboration with Robert Balaban and colleagues, we have been working on quantification of photonic behavior in two-photon microscopy. That work, in turn, has led to a collaboration with Hacène Boukari to quantify photon behavior in fluorescence correlation spectroscopy.

Recent developments in two-photon imaging have aimed at pushing back the depth limit of this remarkable imaging technique. In particular, we have been involved in the development and quantification of a new approach to enhance two-photon microscopy. The basic tenet of the approach is to obtain total emission detection (TED) as opposed to the usual epi/trans detection of two-photon imaging. Others have developed an instrument called “MoreLight” that uses optics to capture light from all directions; it represents an advance over the small solid angle usually used in two-photon imaging. To quantify the instrument, our group developed a simple emission-only fluorescence Monte Carlo model. Given that we are dealing with scattered light (light emitted through tissue), we use a photonic model rather than a wave model. We can safely ignore the diffraction blurring because we are at the boundary of photonic/wave behavior of light and are considering only the gain of the instrument. The initial results from our algorithm matched well with data from a phantom study (given a scaling factor based on known instrumental losses due to efficiency of optical components). Use of the virtual modeling tool also permitted further quantification of the effective gain in different tissue types.

We are now collaborating with Ralph Nossal and Hacène Boukari in the Program in Physical Biology to quantify the effects of scattering in fluorescence correlation spectroscopy (FCS). It is common to use microbeads in FCS to examine a molecule’s diffusion rates; however, the microbeads’ effect on imaging needs further consideration. It is generally noted that some blurring occurs, but no quantification is currently available. We have thus created a second Monte Carlo simulation that looks at the source field for a confocal instrument and noted that the effects of scatter on an FCS illumination field are more complex than just blurring. The scattering (and relative anisotropy) has significant effects on the shape and location of the illuminated region, thus biasing the monitored diffusion rate. We have shown how scattering can blur the confocal spot entirely and how relatively high anisotropy factors can reduce this effect. It is thus important to understand the size and refractive index of the microbeads and how they affect the illumination field in FCS.

Combs CA, Smirnov AV, Riley JD, Gandjbakhche AH, Knutson JR, Balaban RS. Optimization of multiphoton excitation microscopy by total emission detection using a parabolic light reflector. J Microsc 2007;228(Pt3):330-337.

¹Russian Academy of Science, Troitsk, Russia
²University of Maryland at College Park
³¹University of Virginia, Charlottesville, VA
⁴University of Delaware, Newark, DE
⁵charge-coupled device

Collaborators

Robert Balaban, PhD, Laboratory of Cardiac Energetics, NHLBI, Bethesda, MD
Tiziano Binzoni, PhD, Université de Genève, Geneva, Switzerland
Hacène Boukari, PhD, Program in Physical Biology, NICHD, Bethesda, MD
Kevin Camphausen, PhD, Radiation Oncology Branch, NCI, Bethesda, MD
Jacek Capala, PhD, Radiation Oncology Branch, NCI, Bethesda, MD
Christian Combs, PhD, Light Microscopy Facility, NHLBI, Bethesda, MD
Stavros Demos, PhD, Lawrence Livermore National Laboratory, Livermore, CA
Israel Gannot, PhD, Tel Aviv University, Tel Aviv, Israel, and The George Washington University, Washington, DC
Jeremy Hebden, PhD, University College London, London, UK
Ilko Ilev, PhD, Office of Science and Engineering Laboratories, FDA, Bethesda, MD
Jay Knutson, PhD, Laboratory of Molecular Biophysics, NHLBI, Bethesda, MD
Sang-Bong Lee, PhD, Radiation Oncology Branch, NCI, Bethesda, MD
Adrian Neagu, PhD, Victor Babes University of Medicine and Pharmacy, Vest Timis, Romania
Ralph Nossal, PhD, Program in Physical Biology, NICHD, Bethesda, MD
Anne Plant, PhD, NIST, Gaithersburg, MD
Dan Sackett, PhD, Program in Physical Biology, NICHD, Bethesda, MD
Alexsandr Smirnov, PhD, Light Microscopy Facility, NHLBI, Bethesda, MD
Paul Smith, PhD, Division of Bioengineering and Physical Science, ORS, Bethesda, MD
Ronald Waynant, PhD, Office of Science and Engineering Laboratories, FDA, Bethesda, MD
George Weiss, PhD, Division of Computational Bioscience, CIT, NIH, Bethesda, MD
Robert Yarchoan, MD, HIV and AIDS Malignancy Branch, NCI, Bethesda, MD

For further information, contact amir@helix.nih.gov or visit http://www.sbsp-limb.nichd.nih.gov.