Fluorescence Diffuse Optical Tomography Research Articles

Experimental evaluation of the resolution and quantitative accuracy of temperature-modulated fluorescence tomography.

Previously, we reported on the spatial resolution and quantitative accuracy of temperature-modulated fluorescence tomography (TM-FT) using simulation studies. TM-FT is a novel fully integrated multimodality imaging technique that combines fluorescence diffuse optical tomography (FT) with focused ultrasound. Utilizing unique thermo-reversible fluorescent nanocapsules (ThermoDots), TM-FT provides high-resolution cross-sectional fluorescence images in thick tissue (up to 6cm). Focused ultrasound and temperature-sensitive ThermoDots are combined to provide accurate localization of these fluorescent probes and functional a priori information to constrain the conventional FT reconstruction algorithm. Our previous simulation studies evaluated the performance of TM-FT using synthetic phantoms with multiple fluorescence targets of various sizes located at different depths. In this follow-up work, we perform experimental studies to evaluate the performance of this hybrid imaging system, in particular, the effect of size, depth, and concentration of the fluorescence target. While FT alone is unable to accurately locate and resolve the fluorophore target in many cases, TM-FT is able to resolve the size and concentration of the ThermoDots within a thick turbid medium with high accuracy for all cases. The maximum error in the recovered ThermoDots concentration and target sizes with TM-FT are 12% and 25%, respectively.

Fluorescence Diffusion in the Presence of Optically Clear Tissues in a Mouse Head Model.

Diffuse Optical Tomography commonly neglects or assumes as insignificant the presence of optically clear regions in biological tissues, estimating their contribution as a small perturbation to light transport. The inaccuracy introduced by this practice is examined in detail in the context of a complete, based on realistic geometry, virtual fluorescence Diffuse Optical Tomography experiment where a mouse head is imaged in the presence of cerebral spinal fluid. Despite the small thickness of such layer, we point out that an error is introduced when neglecting it from the model with possibly reduction in the accuracy of the reconstruction and localization of the fluorescence distribution within the brain. The results obtained in the extensive study presented here suggest that fluorescence diffuse neuroimaging studies can be improved in terms of quantitative and qualitative reconstruction by accurately taking into account optically transparent regions especially in the cases where the reconstruction is aided by the prior knowledge of the structural geometry of the specimen. Thus, this has only recently become an affordable choice, thanks to novel computation paradigms that allow to run Monte Carlo photon propagation on a simple graphic card, hence speeding up the process a thousand folds compared to CPU-based solutions.

Printed optics: phantoms for quantitative deep tissue fluorescence imaging.

Three-dimensional (3D) printing allows for complex or physiologically realistic phantoms, useful, for example, in developing biomedical imaging methods and for calibrating measured data. However, available 3D printing materials provide a limited range of static optical properties. We overcome this limitation with a new method using stereolithography that allows tuning of the printed phantom's optical properties to match that of target tissues, accomplished by printing a mixture of polystyrene microspheres and clear photopolymer resin. We show that Mie theory can be used to design the optical properties, and demonstrate the method by fabricating a mouse phantom and imaging it using fluorescence optical diffusion tomography.

In Vivo Follow-up of Brain Tumor Growth via Bioluminescence Imaging and Fluorescence Tomography.

Reporter gene-based strategies are widely used in experimental oncology. Bioluminescence imaging (BLI) using the firefly luciferase (Fluc) as a reporter gene and d-luciferin as a substrate is currently the most widely employed technique. The present paper compares the performances of BLI imaging with fluorescence imaging using the near infrared fluorescent protein (iRFP) to monitor brain tumor growth in mice. Fluorescence imaging includes fluorescence reflectance imaging (FRI), fluorescence diffuse optical tomography (fDOT), and fluorescence molecular Imaging (FMT®). A U87 cell line was genetically modified for constitutive expression of both the encoding Fluc and iRFP reporter genes and assayed for cell, subcutaneous tumor and brain tumor imaging. On cultured cells, BLI was more sensitive than FRI; in vivo, tumors were first detected by BLI. Fluorescence of iRFP provided convenient tools such as flux cytometry, direct detection of the fluorescent protein on histological slices, and fluorescent tomography that allowed for 3D localization and absolute quantification of the fluorescent signal in brain tumors.

Reconstruction of an optical inhomogeneity map improves fluorescence diffuse optical tomography

We propose a new reconstruction algorithm for fluorescence diffuse optical tomography, which is designed for highly heterogeneous objects, such as biological tissues. It is a two-step algorithm that exploits continuous-wave measurements acquired at both excitation and fluorescence wavelengths. First, an optical inhomogeneity map, which depends on both absorption and diffusion coefficients, is obtained from excitation measurements. Second, the fluorescence distribution is reconstructed considering the recovered optical inhomogeneity map. The algorithm includes dimensionality reduction techniques, namely measurement compression and structured illumination, which significantly reduce acquisition and reconstruction times. The algorithm has been tested on experimental data acquired from tissue-mimicking phantoms considering sparsity priors. We demonstrate the feasibility and effectiveness of this new approach that provides improved reconstruction quality when compared with the standard normalized Born method.

Approximate marginalization of absorption and scattering in fluorescence diffuse optical tomography

In fluorescence diffuse optical tomography (fDOT), the reconstruction of the fluorophoreconcentration inside the target body is usually carried out using a normalized Born approximation model wherethe measured fluorescent emission data is scaled by measured excitation data. One of the benefits of the model is that it can tolerate inaccuracy in the absorption and scattering distributions that are used in the construction of the forward model to some extent. In this paper, we employ the recently proposed Bayesian approximation error approach to fDOT for compensating for themodeling errors caused by the inaccurately known optical properties of the target in combination with the normalized Born approximation model. The approach is evaluated using a simulated test case withdifferent amount of error in the optical properties. The results show that the Bayesian approximation error approach improvesthe tolerance of fDOT imaging against modeling errors caused by inaccurately known absorption and scattering of the target.

Patch-based anisotropic diffusion scheme for fluorescence diffuse optical tomography—part 1: technical principles

Fluorescence diffuse optical tomography (fDOT) provides 3D images of fluorescence distributions in biological tissue, which represent molecular and cellular processes. The image reconstruction problem is highly ill-posed and requires regularisation techniques to stabilise and find meaningful solutions. Quadratic regularisation tends to either oversmooth or generate very noisy reconstructions, depending on the regularisation strength. Edge preserving methods, such as anisotropic diffusion regularisation (AD), can preserve important features in the fluorescence image and smooth out noise. However, AD has limited ability to distinguish an edge from noise. In this two-part paper, we propose a patch-based anisotropic diffusion regularisation (PAD), where regularisation strength is determined by a weighted average according to the similarity between patches around voxels within a search window, instead of a simple local neighbourhood strategy. However, this method has higher computational complexity and, hence, we wavelet compress the patches (PAD-WT) to speed it up, while simultaneously taking advantage of the denoising properties of wavelet thresholding. The proposed method combines the nonlocal means (NLM), AD and wavelet shrinkage methods, which are image processing methods. Therefore, in this first paper, we used a denoising test problem to analyse the performance of the new method. Our results show that the proposed PAD-WT method provides better results than the AD or NLM methods alone. The efficacy of the method for fDOT image reconstruction problem is evaluated in part 2.

Patch-based anisotropic diffusion scheme for fluorescence diffuse optical tomography—part 2: image reconstruction

Fluorescence diffuse optical tomography (fDOT) provides 3D images of fluorescence distributions in biological tissue, which represent molecular and cellular processes. The image reconstruction problem is highly ill-posed and requires regularisation techniques to stabilise and find meaningful solutions. Quadratic regularisation tends to either oversmooth or generate very noisy reconstructions, depending on the regularisation strength. Edge preserving methods, such as anisotropic diffusion regularisation (AD), can preserve important features in the fluorescence image and smooth out noise. However, AD has limited ability to distinguish an edge from noise. We propose a patch-based anisotropic diffusion regularisation (PAD), where regularisation strength is determined by a weighted average according to the similarity between patches around voxels within a search window, instead of a simple local neighbourhood strategy. However, this method has higher computational complexity and, hence, we wavelet compress the patches (PAD-WT) to speed it up, while simultaneously taking advantage of the denoising properties of wavelet thresholding. Furthermore, structural information can be incorporated into the image reconstruction with PAD-WT to improve image quality and resolution. In this case, the weights used to average voxels in the image are calculated using the structural image, instead of the fluorescence image. The regularisation strength depends on both structural and fluorescence images, which guarantees that the method can preserve fluorescence information even when it is not structurally visible in the anatomical images. In part 1, we tested the method using a denoising problem. Here, we use simulated and in vivo mouse fDOT data to assess the algorithm performance. Our results show that the proposed PAD-WT method provides high quality and noise free images, superior to those obtained using AD.

In Vitro and In Vivo Imaging of Fluorescent Aptamers.

Fluorescence imaging techniques could be used in different ways to study the interaction of aptamers with biological systems from cell culture to animal models. Here, we present the methods developed in our laboratory for fluorescently labeled aptamers, study their internalization inside living cells using time-lapse microscopy, and monitor their biodistribution in mice bearing subcutaneous xenograft tumors using planar fluorescence imaging and fluorescence diffuse optical tomography (fDOT).

Prospects on Time-Domain Diffuse Optical Tomography Based on Time-Correlated Single Photon Counting for Small Animal Imaging

This paper discusses instrumentation based on multiview parallel high temporal resolution (<50 ps) time-domain (TD) measurements for diffuse optical tomography (DOT) and a prospective view on the steps to undertake as regards such instrumentation to make TD-DOT a viable technology for small animal molecular imaging. TD measurements provide information-richest data, and we briefly review the interaction of light with biological tissues to provide an understanding of this. This data richness is yet to be exploited to its full potential to increase the spatial resolution of DOT imaging and to allow probing, via the fluorescence lifetime, tissue biochemical parameters, and processes that are otherwise not accessible in fluorescence DOT. TD data acquisition time is, however, the main factor that currently compromises the viability of TD-DOT. Current high temporal resolution TD-DOT scanners simply do not integrate sufficient detection channels. Based on our past experience in developing TD-DOT instrumentation, we review and discuss promising technologies to overcome this difficulty. These are single photon avalanche diode (SPAD) detectors and fully parallel highly integrated electronics for time-correlated single photon counting (TCSPC). We present experimental results obtained with such technologies demonstrating the feasibility of next-generation multiview TD-DOT therewith.

Accelerating fDOT image reconstruction based on path-history fluorescence Monte Carlo model by using three-level parallel architecture.

The excessive time required by fluorescence diffuse optical tomography (fDOT) image reconstruction based on path-history fluorescence Monte Carlo model is its primary limiting factor. Herein, we present a method that accelerates fDOT image reconstruction. We employ three-level parallel architecture including multiple nodes in cluster, multiple cores in central processing unit (CPU), and multiple streaming multiprocessors in graphics processing unit (GPU). Different GPU memories are selectively used, the data-writing time is effectively eliminated, and the data transport per iteration is minimized. Simulation experiments demonstrated that this method can utilize general-purpose computing platforms to efficiently implement and accelerate fDOT image reconstruction, thus providing a practical means of using path-history-based fluorescence Monte Carlo model for fDOT imaging.

Accurate quantification of fluorescent targets within turbid media based on a decoupled fluorescence Monte Carlo model.

We propose a method based on a decoupled fluorescence Monte Carlo model for constructing fluorescence Jacobians to enable accurate quantification of fluorescence targets within turbid media. The effectiveness of the proposed method is validated using two cylindrical phantoms enclosing fluorescent targets within homogeneous and heterogeneous background media. The results demonstrate that our method can recover relative concentrations of the fluorescent targets with higher accuracy than the perturbation fluorescence Monte Carlo method. This suggests that our method is suitable for quantitative fluorescence diffuse optical tomography, especially for in vivo imaging of fluorophore targets for diagnosis of different diseases and abnormalities.

Performance evaluation of principal component analysis for dynamic fluorescence tomographic imaging in measurement space

Challenges remain in resolving drug (fluorescent biomarkers) distributions within small animals by fluorescence diffuse optical tomography (FDOT). Principal component analysis (PCA) provides the capability of detecting organs (functional structures) from dynamic FDOT images. However, the resolving performance of PCA may be affected by various experimental factors, e.g., the noise levels in measurement data, the variance in optical properties, the number of acquired frames, and so on. To address the problem, based on a simulation model, we analyze and compare the performance of PCA when applied to three typical sets of experimental conditions (frames number, noise level, and optical properties). The results show that the noise is a critical factor affecting the performance of PCA. When input data containing a low noise (<5%), by a short (e.g., 6 frame) projection sequence, we can resolve the poly(DL-lactic-coglycolic acid)/indocynaine green (PLGA/ICG) distributions in heart and lungs, even though there are great variances in optical properties. In contrast, when 20% Gaussian noise is added to the input data, it hardly resolves the distributions of PLGA/ICG in heart and lungs even though accurate optical properties are used. However, with an increased number of frames, the resolving performance of PCA may gradually recover.

Hyperspectral time-resolved wide-field fluorescence molecular tomography based on structured light and single-pixel detection.

We present a time-resolved fluorescence diffuse optical tomography platform that is based on wide-field structured illumination, single-pixel detection, and hyperspectral acquisition. Two spatial light modulators (digital micro-mirror devices) are employed to generate independently wide-field illumination and detection patterns, coupled with a 16-channel spectrophotometer detection module to capture hyperspectral time-resolved tomographic data sets. The main system characteristics are reported, and we demonstrate the feasibility of acquiring dense 4D tomographic data sets (space, time, spectra) for time domain 3D quantitative multiplexed fluorophore concentration mapping in turbid media.

Optical Property of Indocyanine Green in a Tissue Model

The fluorescence temporal profile has been studied at the symmetrical point of the target in tissue phantom by a time-resolved fluorescence diffuse optical tomography method. Indocyanine green served as the fluorescence target. The results showed that the geometrical symmetry of the fluorescence peak intensity ratio was broken due to target position. By using this parameter, an unknown target location could be identified for fluorescence targeting and further reconstruction of the fluorescence image in a tissue model.

Light illumination and detection patterns for fluorescence diffuse optical tomography based on compressive sensing.

Fluorescence diffuse optical tomography (FDOT) is an emerging molecular imaging modality that uses near infrared light to excite the fluorophore injected into tissue; and to reconstruct the fluorophore concentration from boundary measurements. The FDOT image reconstruction is a highly ill-posed inverse problem due to a large number of unknowns and limited number of measurements. However, the fluorophore distribution is often very sparse in the imaging domain since fluorophores are typically designed to accumulate in relatively small regions. In this paper, we use compressive sensing (CS) framework to design light illumination and detection patterns to improve the reconstruction of sparse fluorophore concentration. Unlike the conventional FDOT imaging where spatially distributed light sources illuminate the imaging domain one at a time and the corresponding boundary measurements are used for image reconstruction, we assume that the light sources illuminate the imaging domain simultaneously several times and the corresponding boundary measurements are linearly filtered prior to image reconstruction. We design a set of optical intensities (illumination patterns) and a linear filter (detection pattern) applied to the boundary measurements to improve the reconstruction of sparse fluorophore concentration maps. We show that the FDOT sensing matrix can be expressed as a columnwise Kronecker product of two matrices determined by the excitation and emission light fields. We derive relationships between the incoherence of the FDOT forward matrix and these two matrices, and use these results to reduce the incoherence of the FDOT forward matrix. We present extensive numerical simulation and the results of a real phantom experiment to demonstrate the improvements in image reconstruction due to the CS-based light illumination and detection patterns in conjunction with relaxation and greedy-type reconstruction algorithms.

Iterated preconditioned LSQR method for inverse problems on unstructured grids

This article presents a method for solving large-scale linear inverse imaging problems regularized with a nonlinear, edge-preserving penalty term such as total variation or the Perona–Malik technique. Our method is aimed at problems defined on unstructured meshes, where such regularizers naturally arise in unfactorized form as a stiffness matrix of an anisotropic diffusion operator and factorization is prohibitively expensive. In the proposed scheme, the nonlinearity is handled with lagged diffusivity fixed point iteration, which involves solving a large-scale linear least squares problem in each iteration. Because the convergence of Krylov methods for problems with discontinuities is notoriously slow, we propose to accelerate it by means of priorconditioning (Bayesian preconditioning). priorconditioning is a technique that, through transformation to the standard form, embeds the information contained in the prior (Bayesian interpretation of a regularizer) directly into the forward operator and thence into the solution space. We derive a factorization-free preconditioned LSQR algorithm (MLSQR), allowing implicit application of the preconditioner through efficient schemes such as multigrid. The resulting method is also matrix-free i.e. the forward map can be defined through its action on a vector. We illustrate the performance of the method on two numerical examples. Simple 1D-deblurring problem serves to visualize the discussion throughout the paper. The effectiveness of the proposed numerical scheme is demonstrated on a three-dimensional problem in fluorescence diffuse optical tomography with total variation regularization derived algebraic multigrid preconditioner, which is the type of large scale, unstructured mesh problem, requiring matrix-free and factorization-free approaches that motivated the work here.

Reconstruction of localized fluorescent target from multi-view continuous-wave surface images of small animal with lp sparsity regularization.

Fluorescence diffuse optical tomography using a multi-view continuous-wave and non-contact measurement system and an algorithm incorporating the lp (0 < p ≤ 1) sparsity regularization reconstructs a localized fluorescent target in a small animal. The measurement system provides a total of 25 fluorescence surface 2D-images of an object, which are acquired by a CCD camera from five different angles of view with excitation from five different angles. Fluorescence surface emissions from five different angles of view are simultaneously imaged on the CCD sensor, thus leading to fast acquisition of the 25 images within three minutes. The distributions of the fluorophore are reconstructed by solving the inverse problem based on the photon diffusion equations. In the reconstruction process incorporating the lp sparsity regularization, the regularization term is reformulated as a differentiable function for gradient-based non-linear optimization. Numerical simulations and phantom experiments show that the use of the lp sparsity regularization improves the localization of the target and quantitativeness of the fluorophore concentration. A mouse experiment demonstrates that a localized fluorescent target in a mouse is successfully reconstructed.

Integrated Multichannel Electronics for High Performance Time Correlated Single Photon Counting

The analysis of extremely weak, ultrafast and periodical light signals is increasingly widespread in a large number of fields, from chemistry and biology to telemetry and communication, and in a great variety of applications, from quantum cryptography to fluorescence imaging.Speaking of fluorescent lifetime analysis, for example, the detection of very faint and fast optical signals is fundamental to investigate the complex interactions between molecules that play a key role in many crucial issues such as the origin and growth mechanisms of tumors. In this domain the Time-Correlated Single Photon Counting (TCSPC) technique is very effective to reconstruct the light waveform when both extremely high time resolution and linearity are required.TCSPC is based on the detection of single photons and on the record of their arrival times: if the probability of detecting a photon during one excitation cycle is far less than one, the histogram of photon arrivals per time bin exactly corresponds to the temporal trend of the light signal. In order to properly reconstruct the light waveform a time measurement block featuring very high performance is necessary: this circuit has to be able to measure time intervals of few hundreds of nanoseconds with picosecond time resolution and a differential non-linearity much lower than few percents of the time bin width.Moreover a growing number of techniques do now require to collect data from multiple channels at the same time. Förster Resonance Energy Transfer (FRET), Fluorescence Diffuse Optical Tomography (FDOT) and single-molecule spectroscopy, for example, can benefit a lot of a system capable of providing simultaneous analysis in temporal or spectral range. Indeed, concurrency not only reduces the measurement time, revealing dynamic phenomena even on very short timescales but it also opens the way to spectrally resolved measurements that analyze different events occurring simultaneously, like Fluorescence Cross-Correlation Spectroscopy (FCCS).Nevertheless the development of a truly parallel acquisition system featuring both high performance and a high number of independent channels is still an open challenge: to achieve this goal the design of a time measurement circuit capable of acquiring data from several channels in parallel while providing high performance on each channel is of the utmost importance. In this work a fully integrated array of 16 time to amplitude converters (TAC) designed in 0.35µm SiGe technology is presented.The first simulations show very high performance on each channel with a FWHM time resolution lower than 19ps and a differential non linearity lower than 0.4% peak-to-peak of the LSB. In order to reduce the overall area of the chip, pairs of converter stages share all the conversion chain through a 2 input multiplexer. The simulated dead time on the entire full scale range is 191ns, corresponding to a maximum conversion frequency of 5.24MHz per channel. The obtained results on the designed 16 channel TAC array are extremely good and, if the performance are verified by experimental measurements, this array will be one of the fundamental blocks of a fully parallel high-performance TCSPC acquisition system. Based on high efficiency fully custom single photon detectors (SPAD) and on a front-end electronics specifically designed to properly work with these sensors, the final system will be the upgrade of a few channels high performance acquisition chain already developed in the same technology.

Instrumentation in Diffuse Optical Imaging.

Diffuse optical imaging is highly versatile and has a very broad range of applications in biology and medicine. It covers diffuse optical tomography, fluorescence diffuse optical tomography, bioluminescence, and a number of other new imaging methods. These methods of diffuse optical imaging have diversified instrument configurations but share the same core physical principle – light propagation in highly diffusive media, i.e., the biological tissue. In this review, the author summarizes the latest development in instrumentation and methodology available to diffuse optical imaging in terms of system architecture, light source, photo-detection, spectral separation, signal modulation, and lastly imaging contrast.