Joyita Dutta

A three-dimensional multispectral fluorescence optical tomography imaging system for small animals based on a conical mirror design

We have developed a three dimensional (3D) multispectral fluorescence optical tomography small animal imaging system with an innovative geometry using a truncated conical mirror, allowing simultaneous viewing of the entire surface of the animal by an EMCCD camera. A conical mirror collects photons approximately three times more efficiently than a flat mirror. An x-y mirror scanning system makes it possible to scan a collimated excitation laser beam to any location on the mouse surface. A pattern of structured light incident on the small animal surface is used to extract the surface geometry for reconstruction. A finite element based algorithm is applied to model photon propagation in the turbid media and a preconditioned conjugate gradient (PCG) method is used to solve the large linear system matrix. The reconstruction algorithm and the system feasibility are evaluated by phantom experiments. These experiments show that multispectral measurements improve the spatial resolution of reconstructed images. Finally, an in vivo imaging study of a xenograft tumor in a mouse shows good correlation of the reconstructed image with the location of the fluorescence probe as determined by subsequent optical imaging of cryosections of the mouse.

Joint L1 and total variation regularization for fluorescence molecular tomography

Fluorescence molecular tomography (FMT) is an imaging modality that exploits the specificity of fluorescent biomarkers to enable 3D visualization of molecular targets and pathways in vivo in small animals. Owing to the high degree of absorption and scattering of light through tissue, the FMT inverse problem is inherently ill-conditioned making image reconstruction highly susceptible to the effects of noise and numerical errors. Appropriate priors or penalties are needed to facilitate reconstruction and to restrict the search space to a specific solution set. Typically, fluorescent probes are locally concentrated within specific areas of interest (e.g., inside tumors). The commonly used L(2) norm penalty generates the minimum energy solution, which tends to be spread out in space. Instead, we present here an approach involving a combination of the L(1) and total variation norm penalties, the former to suppress spurious background signals and enforce sparsity and the latter to preserve local smoothness and piecewise constancy in the reconstructed images. We have developed a surrogate-based optimization method for minimizing the joint penalties. The method was validated using both simulated and experimental data obtained from a mouse-shaped phantom mimicking tissue optical properties and containing two embedded fluorescent sources. Fluorescence data were collected using a 3D FMT setup that uses an EMCCD camera for image acquisition and a conical mirror for full-surface viewing. A range of performance metrics was utilized to evaluate our simulation results and to compare our method with the L(1), L(2) and total variation norm penalty-based approaches. The experimental results were assessed using the Dice similarity coefficients computed after co-registration with a CT image of the phantom.

Illumination pattern optimization for fluorescence tomography: theory and simulation studies.

Fluorescence molecular tomography is a powerful tool for 3D visualization of molecular targets and pathways in vivo in small animals. Owing to the high degrees of absorption and scattering of light through tissue, the fluorescence tomographic inverse problem is inherently ill-posed. In order to improve source localization and the conditioning of the light propagation model, multiple sets of data are acquired by illuminating the animal surface with different spatial patterns of near-infrared light. However, the choice of these patterns in most experimental setups is ad hoc and suboptimal. This paper presents a systematic approach for designing efficient illumination patterns for fluorescence tomography. Our objective here is to determine how to optimally illuminate the animal surface so as to maximize the information content in the acquired data. We achieve this by improving the conditioning of the Fisher information matrix. We parameterize the spatial illumination patterns and formulate our problem as a constrained optimization problem that, for a fixed number of illumination patterns, yields the optimal set of patterns. For geometric insight, we used our method to generate a set of three optimal patterns for an optically homogeneous, regular geometrical shape and observed expected symmetries in the result. We also generated a set of six optimal patterns for an optically homogeneous cuboidal phantom set up in the transillumination mode. Finally, we computed optimal illumination patterns for an optically inhomogeneous realistically shaped mouse atlas for different given numbers of patterns. The regularized pseudoinverse matrix, generated using the singular value decomposition, was employed to reconstruct the point spread function for each set of patterns in the presence of a sample fluorescent point source deep inside the mouse atlas. We have evaluated the performance of our method by examining the singular value spectra as well as plots of average spatial resolution versus estimator variance corresponding to different illumination schemes.

Non-local means denoising of dynamic PET images.

Objective Dynamic positron emission tomography (PET), which reveals information about both the spatial distribution and temporal kinetics of a radiotracer, enables quantitative interpretation of PET data. Model-based interpretation of dynamic PET images by means of parametric fitting, however, is often a challenging task due to high levels of noise, thus necessitating a denoising step. The objective of this paper is to develop and characterize a denoising framework for dynamic PET based on non-local means (NLM). Theory NLM denoising computes weighted averages of voxel intensities assigning larger weights to voxels that are similar to a given voxel in terms of their local neighborhoods or patches. We introduce three key modifications to tailor the original NLM framework to dynamic PET. Firstly, we derive similarities from less noisy later time points in a typical PET acquisition to denoise the entire time series. Secondly, we use spatiotemporal patches for robust similarity computation. Finally, we use a spatially varying smoothing parameter based on a local variance approximation over each spatiotemporal patch. Methods To assess the performance of our denoising technique, we performed a realistic simulation on a dynamic digital phantom based on the Digimouse atlas. For experimental validation, we denoised PET images from a mouse study and a hepatocellular carcinoma patient study. We compared the performance of NLM denoising with four other denoising approaches – Gaussian filtering, PCA, HYPR, and conventional NLM based on spatial patches. Results The simulation study revealed significant improvement in bias-variance performance achieved using our NLM technique relative to all the other methods. The experimental data analysis revealed that our technique leads to clear improvement in contrast-to-noise ratio in Patlak parametric images generated from denoised preclinical and clinical dynamic images, indicating its ability to preserve image contrast and high intensity details while lowering the background noise variance.

Quantitative Statistical Methods for Image Quality Assessment

Quantitative measures of image quality and reliability are critical for both qualitative interpretation and quantitative analysis of medical images. While, in theory, it is possible to analyze reconstructed images by means of Monte Carlo simulations using a large number of noise realizations, the associated computational burden makes this approach impractical. Additionally, this approach is less meaningful in clinical scenarios, where multiple noise realizations are generally unavailable. The practical alternative is to compute closed-form analytical expressions for image quality measures. The objective of this paper is to review statistical analysis techniques that enable us to compute two key metrics: resolution (determined from the local impulse response) and covariance. The underlying methods include fixed-point approaches, which compute these metrics at a fixed point (the unique and stable solution) independent of the iterative algorithm employed, and iteration-based approaches, which yield results that are dependent on the algorithm, initialization, and number of iterations. We also explore extensions of some of these methods to a range of special contexts, including dynamic and motion-compensated image reconstruction. While most of the discussed techniques were developed for emission tomography, the general methods are extensible to other imaging modalities as well. In addition to enabling image characterization, these analysis techniques allow us to control and enhance imaging system performance. We review practical applications where performance improvement is achieved by applying these ideas to the contexts of both hardware (optimizing scanner design) and image reconstruction (designing regularization functions that produce uniform resolution or maximize task-specific figures of merit).

DigiWarp: a method for deformable mouse atlas warping to surface topographic data

For pre-clinical bioluminescence or fluorescence optical tomography, the animals surface topography and internal anatomy need to be estimated for improving the quantitative accuracy of reconstructed images. The animals surface profile can be measured by all-optical systems, but estimation of the internal anatomy using optical techniques is non-trivial. A 3D anatomical mouse atlas may be warped to the estimated surface. However, fitting an atlas to surface topography data is challenging because of variations in the posture and morphology of imaged mice. In addition, acquisition of partial data (for example, from limited views or with limited sampling) can make the warping problem ill-conditioned. Here, we present a method for fitting a deformable mouse atlas to surface topographic range data acquired by an optical system. As an initialization procedure, we match the posture of the atlas to the posture of the mouse being imaged using landmark constraints. The asymmetric L(2) pseudo-distance between the atlas surface and the mouse surface is then minimized in order to register two data sets. A Laplacian prior is used to ensure smoothness of the surface warping field. Once the atlas surface is normalized to match the range data, the internal anatomy is transformed using elastic energy minimization. We present results from performance evaluation studies of our method where we have measured the volumetric overlap between the internal organs delineated directly from MRI or CT and those estimated by our proposed warping scheme. Computed Dice coefficients indicate excellent overlap in the brain and the heart, with fair agreement in the kidneys and the bladder.

Posture matching and elastic registration of a mouse atlas to surface topography range data

Estimation of internal mouse anatomy is required for quantitative bioluminescence or fluorescence tomography. However, only surface range data can be recovered from all-optical systems. These data are at times sparse or incomplete. We present a method for fitting an elastically deformable mouse atlas to surface topographic range data acquired by an optical system. In this method, we first match the postures of a deformable atlas and the range data of the mouse being imaged. This is achieved by aligning manually identified landmarks. We then minimize the asymmetric L2 pseudo-distance between the surface of the deformable atlas and the surface topography range data. Once this registration is accomplished, the internal anatomy of the atlas is transformed to the coordinate system of the range data using elastic energy minimization. We evaluated our method by using it to register a digital mouse atlas to a surface model produced from a manually labeled CT mouse data set. Dice coefficents indicated excellent agreement in the brain and heart, with fair agreement in the kidneys and bladder. We also present example results produced using our method to align the digital mouse atlas to surface range data.

Respiratory motion compensation in simultaneous PET/MR using a maximum a posteriori approach

The quantitative reliability of pulmonary PET scans is compromised by respiratory motion. The emergence of simultaneous, whole-body PET/MR imaging enables us to correct for motion artifacts in PET using motion information derived from anatomical MR images. We present here a framework for respiratory motion-compensated PET image reconstruction using simultaneous PET/MR. We have developed a radial FLASH pulse sequence for generating gated volumetric MR images at a reasonable speed without significantly sacrificing image quality. A navigator encapsulated within the pulse sequence enables us to retrospectively compute time bins corresponding to each gate. The deformation fields for each gate with respect to a reference gate are computed from the gated MR images by means of non-rigid registration. The gated MR images are also used to generate individual attenuation maps for each gate. Finally motion-compensated PET reconstruction is performed using a maximum a posteriori (MAP) approach. The complete framework was applied to a clinical study conducted on the Biograph mMR scanner (Siemens Medical Solutions), which allows simultaneous acquisition of whole-body PET/MR data. This study demonstrates the utility of our framework in generating meaningful estimates of deformation fields and correcting for motion artifacts in PET.

Hybrid Pixel-Waveform (HPWF) Enabled CdTe Detectors for Small Animal Gamma-Ray Imaging Applications

This paper presents the design and preliminary evaluation of small-pixel CdTe gamma ray detectors equipped with a hybrid pixel-waveform (HPWF) readout system for gamma ray imaging applications with additional discussion on CZT due to its similarity. The HPWF readout system utilizes a pixelated anode readout circuitry which is designed to only provide the pixel address. This readout circuitry works in coincidence with a high-speed digitizer to sample the cathode waveform which provides the energy, timing, and depth-of-interaction (DOI) information. This work focuses on the developed and experimentally evaluated prototype HPWF-CdTe detectors with a custom CMOS pixel-ASIC to readout small anode pixels of

Spatially varying regularization for motion compensated PET reconstruction

350~\mu \text {m}