Search

[ES] Este proyecto de tesis tiene como objetivo el estudio de flujos monofásicos y multifásicos en una tubería equipada con elementos estáticos, centrándose específicamente en los elementos tipo Kenics. Su diseño consta de un plano retorcido 180° ensamblado con diferentes configuraciones, cambiando el número y orientación de los elementos. El estudio pretende contribuir al conocimiento existente sobre los elementos estáticos, dadas sus ventajas y aplicaciones extendidas en la industria química y de procesos. Optar por elementos estáticos puede proporcionar una solución de alta eficiencia y bajo costo para mezclar o separar fases dentro de líneas de transporte. Se utilizó Velocimetría de Imagen de Partículas (PIV) para medir el campo de velocidad del flujo monofásico turbulento completamente desarrollado. El experimento comenzó con una tubería horizontal donde el escenario monofásico se analizó más detalladamente mediante Tomografía de Resistencia Eléctrica (ERT), observando la distribución de un trazador en varios planos de sección transversal a lo largo de la tubería. Además, se utilizó un transductor de presión diferencial para medir la caída de presión a través de los elementos estáticos. En los experimentos de flujo bifásico, la tubería se orientó verticalmente, donde se investigó el flujo ascendente de aire y agua. Con la ayuda de una cámara de alta velocidad y la técnica ERT mencionada anteriormente, se evaluó la distribución de la fracción volumétrica de gas aguas arriba y aguas abajo de los elementos estáticos, evaluando así la idoneidad de la configuración del elemento seleccionado para la separación gas-líquido. Finalmente, la combinación de técnicas ópticas y tomográficas ha permitido seleccionar el algoritmo de reconstrucción más fiable para aplicar en el tratamiento de datos de ERT. Además, el resultado del proyecto ayudará a conocer las principales características globales y locales del flujo y servirá como punto de referencia para la validación de sim, [EN] This thesis project is aimed at studying single-phase and multi-phase flows in a pipeline equipped with static elements, focused specifically on the Kenics-type elements. Its design consists of a flat plane twisted 180°, assembled with different configurations by changing the number and orientation of the elements. The study intends to contribute to the existing knowledge about static elements, given their advantages and extended applications in the chemical and process industry. Opting for these static elements can provide a high-efficiency and low-cost solution when trying to achieve mixing or separation inside transport pipelines. Particle Image Velocimetry (PIV) was used to measure the velocity field of the fully developed turbulent single-phase flow. The experiment started with a horizontal pipe where the single-phase scenario was further analysed using Electrical Resistance Tomography (ERT), observing the distribution of a tracer in various cross-sectional planes along the pipe. Additionally, a differential pressure transducer was used to measure the pressure drop across the static elements. In the two-phase flow experiments, the pipeline was orientated vertically, where the ascendant flow of air and water was investigated. With the aid of a high-speed camera and the previously mentioned ERT technique, the gas volume fraction distribution was evaluated upstream and downstream of the static elements, thus assessing the suitability of the selected element configuration for gas-liquid separation. Finally, the combination of optical and tomographic techniques has allowed the selection of the most reliable reconstruction algorithm to be applied for the ERT data treatment. Moreover, the outcome of the project will help gain insight into the main global and local characteristics of the flow and will serve as a benchmark for the validation of Computational Fluid Dynamics (CFD) simulations carried out in a parallel investigation.

Three-dimensional (3D) models of outdoor scenes can be widely used in a number of fields such as object recognition, navigation, scenic simulation, and mixed reality. Such models are often made manually with high cost, so that automatic 3D reconstruction has been widely investigated. In related works, a dense 3D model is generated by using a stereo method. However, such approaches cannot use several hundred images together or dense depth estimation of large constructs and urban environments because it is difficult to accurately calibrate a large number of cameras. This paper proposes a dense 3D reconstruction method that uses multiple image sequences. First, our method estimates extrinsic camera parameters of each image sequence, then reconstructs a dense 3D model of a scene using an extended multi-baseline stereo and voxel voting technique., MFI 2003 : International Conference on Multisensor Fusion and Integration for Intelligent Systems , Aug 1, 2003 , Tokyo, Japan

In recent years, deep learning has achieved remarkable empirical success for image reconstruction. This has catalyzed an ongoing quest for the precise characterization of the correctness and reliability of data-driven methods in critical use cases, for instance, in medical imaging. Notwithstanding the excellent performance and efficacy of deep learning-based methods, concerns have been raised regarding the approaches’ stability, or lack thereof, with serious practical implications. Significant advances have been made in recent years to unravel the inner workings of data-driven image recovery methods, challenging their widely perceived black-box nature. In this article, we specify relevant notions of convergence for data-driven image reconstruction, which forms the basis of a survey of learned methods with mathematically rigorous reconstruction guarantees. An example that is highlighted is the role of input-convex neural networks (ICNNs), offering the possibility to combine the power of deep learning with classical convex regularization theory for devising methods that are provably convergent. This survey article is aimed at both methodological researchers seeking to advance the frontiers of our understanding of data-driven image reconstruction methods as well as practitioners by providing an accessible description of useful convergence concepts and by placing some of the existing empirical practices on a solid mathematical foundation.

In recent years, deep learning has achieved remarkable empirical success for image reconstruction. This has catalyzed an ongoing quest for the precise characterization of the correctness and reliability of data-driven methods in critical use cases, for instance, in medical imaging. Notwithstanding the excellent performance and efficacy of deep learning-based methods, concerns have been raised regarding the approaches' stability, or lack thereof, with serious practical implications. Significant advances have been made in recent years to unravel the inner workings of data-driven image recovery methods, challenging their widely perceived black-box nature. In this article, we specify relevant notions of convergence for data-driven image reconstruction, which forms the basis of a survey of learned methods with mathematically rigorous reconstruction guarantees. An example that is highlighted is the role of input-convex neural networks (ICNNs), offering the possibility to combine the power of deep learning with classical convex regularization theory for devising methods that are provably convergent. This survey article is aimed at both methodological researchers seeking to advance the frontiers of our understanding of data-driven image reconstruction methods as well as practitioners by providing an accessible description of useful convergence concepts and by placing some of the existing empirical practices on a solid mathematical foundation.

Learning-based surface reconstruction methods have received considerable attention in recent years due to their excellent expressiveness. However, existing learning-based methods lack scalability in processing large-scale point clouds. This paper proposes a novel scalable learning-based 3D surface reconstruction method based on octree, called SSRNet. SSRNet works in a scalable reconstruction pipeline, which divides oriented point clouds into different local parts and then processes them in parallel. Accommodating this scalable design pattern, SSRNet constructs local geometric features for octree vertices. Such features comprise the relation between the vertices and the implicit surface, ensuring geometric perception. Focusing on local geometric information also enables the network to avoid the overfitting problem and generalize well on different datasets. Finally, as a learning-based method, SSRNet can process large-scale point clouds in a short time. And to further solve the efficiency problem, we provide a lightweight and efficient version that is about five times faster while maintaining reconstruction performance. Experiments show that our methods achieve state-of-the-art performance with outstanding efficiency. IEEE

Non-line-of-sight (NLOS) imaging systems reconstruct hidden scenes using computational methods based on indirect light that diffusely reflected from relay walls. Due to the computation and memory requirements of reconstruction algorithms, real-time NLOS imaging for room-size scenes based on non-confocal data has long been challenging. This paper proposes a field programmable gate array (FPGA) accelerator for the recently proposed Rayleigh-Sommerfeld Diffraction (RSD)-based NLOS reconstruction method. In the proposed accelerator design, ring sampling and radius sampling techniques are proposed to reduce the memory requirements by reconstructing the RSD kernels with a set of kernel bases and ring sampling coefficients during the runtime. Based on that, a customized hardware architecture and the corresponding FPGA design for real-time RSD-based NLOS reconstruction is further proposed. Implementation results show that the proposed FPGA accelerator is capable of reconstructing NLOS scenes at 25 frames per second (FPS), running at a relatively slow clock frequency of 50 MHz. To the best knowledge of the authors, this is the first real-time enabled FPGA accelerator for room-size NLOS imaging with a resolution of 128×128 ., QC 20220615

The morphological analysis of dendritic spines is an important challenge for the neuroscientific community. Most state-of-the-art techniques rely on user-supervised algorithms to segment the spine surface, especially those designed for light microscopy images. Therefore, processing large dendritic branches is costly and time-consuming. Although deep learning (DL) models have become one of the most commonly used tools in image segmentation, they have not yet been successfully applied to this problem. In this article, we study the feasibility of using DL models to automatize spine segmentation from confocal microscopy images. Supervised learning is the most frequently used method for training DL models. This approach requires large data sets of high-quality segmented images (ground truth). As mentioned above, the segmentation of microscopy images is time-consuming and, therefore, in most cases, neuroanatomists only reconstruct relevant branches of the stack. Additionally, some parts of the dendritic shaft and spines are not segmented due to dyeing problems. In the context of this research, we tested the most successful architectures in the DL biomedical segmentation field. To build the ground truth, we used a large and high-quality data set, according to standards in the field. Nevertheless, this data set is not sufficient to train convolutional neural networks for accurate reconstructions. Therefore, we implemented an automatic preprocessing step and several training strategies to deal with the problems mentioned above. As shown by our results, our system produces a high-quality segmentation in most cases. Finally, we integrated several postprocessing user-supervised algorithms in a graphical user interface application to correct any possible artifacts.

We study the decomposition of multivariate polynomials as sums of powers of linear forms. We give a randomized algorithm for the following problem: If a homogeneous polynomial f ∈ K[x_1 , . . . , x_n] (where K ⊆ ℂ) of degree d is given as a blackbox, decide whether it can be written as a linear combination of d-th powers of linearly independent complex linear forms. The main novel features of the algorithm are: - For d = 3, we improve by a factor of n on the running time from the algorithm in [Pascal Koiran and Mateusz Skomra, 2021]. The price to be paid for this improvement is that the algorithm now has two-sided error. - For d > 3, we provide the first randomized blackbox algorithm for this problem that runs in time poly(n,d) (in an algebraic model where only arithmetic operations and equality tests are allowed). Previous algorithms for this problem [Kayal, 2011] as well as most of the existing reconstruction algorithms for other classes appeal to a polynomial factorization subroutine. This requires extraction of complex polynomial roots at unit cost and in standard models such as the unit-cost RAM or the Turing machine this approach does not yield polynomial time algorithms. - For d > 3, when f has rational coefficients (i.e. K = ℚ), the running time of the blackbox algorithm is polynomial in n,d and the maximal bit size of any coefficient of f. This yields the first algorithm for this problem over ℂ with polynomial running time in the bit model of computation. These results are true even when we replace ℂ by ℝ. We view the problem as a tensor decomposition problem and use linear algebraic methods such as checking the simultaneous diagonalisability of the slices of a tensor. The number of such slices is exponential in d. But surprisingly, we show that after a random change of variables, computing just 3 special slices is enough. We also show that our approach can be extended to the computation of the actual decomposition. In forthcoming work we plan to extend the

A well-characterized wavefront is important for many x-ray free-electron laser (XFEL) experiments, especially for single-particle imaging (SPI), where individual biomolecules randomly sample a nanometer region of highly focused femtosecond pulses. We demonstrate high-resolution multiple-plane wavefront imaging of an ensemble of XFEL pulses, focused by Kirkpatrick–Baez mirrors, based on mixed-state ptychography, an approach letting us infer and reduce experimental sources of instability. From the recovered wavefront profiles, we show that while local photon fluence correction is crucial and possible for SPI, a small diversity of phase tilts likely has no impact. Our detailed characterization will aid interpretation of data from past and future SPI experiments and provides a basis for further improvements to experimental design and reconstruction algorithms.

The use of multiple wavelength illumination in digital in-line holographic microscopy is presented. A biological sample is illuminated sequentially with three different optical sources and the separate holograms reconstructed with the appropriate wavelength. For controlling the scale of the reconstructed holograms, a reconstruction algorithm that allows for choosing the pixel size at the reconstruction plane independently of the wavelength and the reconstruction distance is used. The method is illustrated with experimental results. © OSA 2012.

© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., We present the phylogenetic quartet reconstruction method SAQ (Semi-Algebraic Quartet reconstruction). SAQ is consistent with the most general Markov model of nucleotide substitution and, in particular, it allows for rate heterogeneity across lineages. Based on the algebraic and semi-algebraic description of distributions that arise from the general Markov model on a quartet, the method outputs normalized weights for the three trivalent quartets (which can be used as input of quartet-based methods). We show that SAQ is a highly competitive method that outperforms most of the well known reconstruction methods on data simulated under the general Markov model on 4-taxon trees. Moreover, it also achieves a high performance on data that violates the underlying assumptions., The authors were partially supported by Spanish government Secretar´ıa de Estado de Investigaci´on, Desarrollo e Innovaci´on [MTM2015-69135-P (MINECO/FEDER)] and [PID2019- 103849GB-I00 (MINECO)]; Generalitat de Catalunya [2014 SGR-634]. M. Garrote-L´opez was also funded by Spanish government, Ministerio de Econom´ıa y Competitividad research project Maria de Maeztu [MDM-2014-0445]., Peer Reviewed, Postprint (author's final draft)

We present an all-in-one camera model that encompasses the architectures of most existing compressive-sensing light-field cameras, equipped with a single lens and multiple amplitude coded masks that can be placed at different positions between the lens and the sensor. The proposed model, named the equivalent multi-mask camera (EMMC) model, enables the comparison between different camera designs, e.g using monochrome or CFA-based sensors, single or multiple acquisitions, or varying pixel sizes, via a simple adaptation of the sampling operator. In particular, in the case of a camera equipped with a CFA-based sensor and a coded mask, this model allows us to jointly perform color demosaicing and light field spatio-angular reconstruction. In the case of variable pixel size, it allows to perform spatial super-resolution in addition to angular reconstruction. While the EMMC model is generic and can be used with any reconstruction algorithm, we validate the proposed model with a dictionary-based reconstruction algorithm and a regularization-based reconstruction algorithm using a 4D Total-Variation-based regularizer for light field data. Experimental results with different reconstruction algorithms show that the proposed model can flexibly adapt to various sensing schemes. They also show the advantage of using an in-built CFA sensor with respect to monochrome sensors classically used in the literature. © 2015 IEEE.

This paper considers the compressive sensing framework as a way of overcoming the spatio-angular trade-off inherent to light field acquisition devices. We present a novel method to reconstruct a full 4D light field from a sparse set of data samples or measurements. The approach relies on the assumption that sparse models in the 4D Fourier domain can efficiently represent light fields. The proposed algorithm reconstructs light fields by selecting the frequencies of the Fourier basis functions that best approximate the available samples in 4D hyper-blocks. The performance of the reconstruction algorithm is further improved by enforcing orthogonality of the approximation residue at each iteration, i.e. for each selected basis function. Since sparsity is better preserved in the continuous Fourier domain, we propose to refine the selected frequencies by searching for neighboring non-integer frequency values. Experiments show that the proposed algorithm yields performance improvements of more than 1 dB compared to state-of-the-art compressive light field reconstruction methods. The frequency refinement step also significantly enhances the visual quality of reconstruction results of our method by a 1.8 dB average. © 2020

Computer Tomography (CT) scanners for research applications are often designed to facilitate flexible acquisition geometries. Making full use of such CT scanners requires advanced reconstruction software that can (i) deal with a broad range of geometrical scanning settings, (ii) allows for customization of processing algorithms, and (iii) has the capability to process large amounts of data. FleXbox is a Python-based tomographic reconstruction toolbox focused on these three functionalities. It is built to bridge the gap between low-level tomographic reconstruction packages (e.g. ASTRA toolbox) and high-level distributed systems (e.g. Livermore Tomography Tools). FleXbox allows to model arbitrary source, detector and object trajectories. The modular architecture of FleXbox allows to design an optimal reconstruction approach for a single CT dataset. When multiple datasets of an object are acquired (either different spatial regions or different snapshots in time), they can be combined into a larger high resolution volume or a time series of volumes. The software allows to then create a computational reconstruction pipeline that can run without user interaction and enables efficient computation on large-scale 3D volumes on a single workstation.

Recovering a high-quality image from noisy indirect measurements is an important problem with many applications. For such inverse problems, supervised deep convolutional neural network (CNN)-based denoising methods have shown strong results, but the success of these supervised methods critically depends on the availability of a high-quality training dataset of similar measurements. For image denoising, methods are available that enable training without a separate training dataset by assuming that the noise in two different pixels is uncorrelated. However, this assumption does not hold for inverse problems, resulting in artifacts in the denoised images produced by existing methods. Here, we propose Noise2Inverse, a deep CNN-based denoising method for linear image reconstruction algorithms that does not require any additional clean or noisy data. Training a CNN-based denoiser is enabled by exploiting the noise model to compute multiple statistically independent reconstructions. We develop a theoretical framework which shows that such training indeed obtains a denoising CNN, assuming the measured noise is element-wise independent, and zero-mean. On simulated CT datasets, Noise2Inverse demonstrates an improvement in peak signal-To-noise ratio and structural similarity index compared to state-of-The-Art image denoising methods, and conventional reconstruction methods, such as Total-Variation Minimization. We also demonstrate that the method is able to significantly reduce noise in challenging real-world experimental datasets.

Background: To evaluate whether a model based on radiomic and clinical features may be associated with lymph node (LN) status and overall survival (OS) in lung cancer (LC) patients; to evaluate whether CT reconstruction algorithms may influence the model performance. Methods: patients operated on for LC with a pathological stage up to T3N1 were retrospectively selected and divided into training and validation sets. For the prediction of positive LNs and OS, the Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression model was used; univariable and multivariable logistic regression analysis assessed the association of clinicalradiomic variables and endpoints. All tests were repeated after dividing the groups according to the CT reconstruction algorithm. p-values < 0.05 were considered significant. Results: 270 patients were included and divided into training (n = 180) and validation sets (n = 90). Transfissural extension was significantly associated with positive LNs. For OS prediction, high-and low-risk groups were different according to the radiomics score, also after dividing the two groups according to reconstruction algorithms. Conclusions: a combined clinical–radiomics model was not superior to a single clinical or single radiomics model to predict positive LNs. A radiomics model was able to separate high-risk and low-risk patients for OS; CTs reconstructed with Iterative Reconstructions (IR) algorithm showed the best model performance.

© 2018 Optical Society of America. The Spanish Ministerio de Economía y Competitividad is acknowledged for funding the project TEC2014-57394-P., Quantitative label-free imaging is an important tool for the study of living microorganisms that, during the last decade, has attracted wide attention from the optical community. Optical diffraction tomography (ODT) is probably the most relevant technique for quantitative label-free 3D imaging applied in wide-field microscopy in the visible range. The ODT is usually performed using spatially coherent light illumination and specially designed holographic microscopes. Nevertheless, the ODT is also compatible with partially coherent illumination and can be realized in conventional wide-field microscopes by applying refocusing techniques, as it has been recently demonstrated. Here, we compare these two ODT modalities, underlining their pros and cons and discussing the optical setups for their implementation. In particular, we pay special attention to a system that is compatible with a conventional wide-field microscope that can be used for both ODT modalities. It consists of two easily attachable modules: the first for sample illumination engineering based on digital light processing technology; the other for focus scanning by using an electrically driven tunable lens. This hardware allows for a programmable selection of the wavelength and the illumination design, and provides fast data acquisition as well. Its performance is experimentally demonstrated in the case of ODT with partially coherent illumination providing speckle-free 3D quantitative imaging., Ministerio de Economía y Competitividad (MINECO), Depto. de Óptica, Fac. de Ciencias Físicas, TRUE, pub

We release two tomographic scans with two levels of radiation dosage of two measured objects for noise-level comparative studies in data analysis, reconstruction or segmentation methods. The objects are referred to as apple and pebbles (more specific, hydrograins), respectively. The dataset collected with higher dosage is referred to as the "good" dataset; and the other as the "noisy" dataset, as a way to distinguish between the two dosage levels.

A major drawback of the optical microscope is its limitation to resolve finer details. Many microscopes have been developed to overcome the limitations set by the diffraction of visible light. The scanning electron microscope (SEM) is one such alternative: it uses electrons for imaging, which have much smaller wavelength than photons. As a result high magnification with superior image resolution can be achieved. However, SEM generates 2D images which provide limited data for surface measurements and analysis. Often many research areas require the knowledge of 3D structures as they contribute to a comprehensive understanding of microstructure by allowing effective measurements and qualitative visualization of the samples under study. For this reason, stereo photogrammetry technique is employed to convert SEM images into 3D measurable data. This paper aims to utilize a stereoscopic reconstruction technique as a reliable method for characterization of surface topography. Reconstructed results from SEM images are compared with coherence scanning interferometer (CSI) results obtained by measuring a roughness reference standard sample. This paper presents a method to select the most robust/consistent surface texture parameters that are insensitive to the uncertainties involved in the reconstruction technique itself. Results from the two-stereoscopic reconstruction algorithms are also documented in this paper. © 2018 IOP Publishing Ltd.

[EN] Even though QR-factorization of the system matrix for tomographic devices has been already used for medical imaging, to date, no satisfactory solution has been found for solving large linear systems, such as those used in computed tomography (CT) (in the order of 106 equations). In CT, the Feldkamp, Davis, and Kress back projection algorithm (FDK) and iterative methods like conjugate gradient (CG) are the standard methods used for image reconstruction. As the image reconstruction problem can be modeled by a large linear system of equations, QR-factorization of the system matrix could be used to solve this system. Current advances in computer science enable the use of direct methods for solving such a large linear system. The QR-factorization is a numerically stable direct method for solving linear systems of equations, which is beginning to emerge as an alternative to traditional methods, bringing together the best from traditional methods. QR-factorization was chosen because the core of the algorithm, from the computational cost point of view, is precalculated and stored only once for a given CT system, and from then on, each image reconstruction only involves a backward substitution process and the product of a vector by a matrix. Image quality assessment was performed comparing contrast to noise ratio and noise power spectrum; performances regarding sharpness were evaluated by the reconstruction of small structures using data measured from a small animal 3-D CT. Comparisons of QR-factorization with FDK and CG methods show that QR-factorization is able to reconstruct more detailed images for a fixed voxel size.

This work reports three reconstruction algorithms developed to address the practical issues encountered in 3D tomography diagnostics, such as the limited view angles available in many practical applications, the large scale and nonlinearity of the problems when they are in 3D, and the measurement uncertainty. These algorithms are: an algebraic reconstruction technique (ART) screening algorithm, a nonlinear iterative reconstruction technique (NIRT), and an iterative reconstruction technique integrating view registration optimization (IRT-VRO) algorithm. The ART screening algorithm was developed to enhance the performance of the traditional ART algorithm to solve linear tomography problems, the NIRT was to solve nonlinear tomography problems, and the IRT-VRO was to address the issue of view registration uncertainty in both linear and nonlinear problems. This dissertation describes the mathematical formulations, and the experimental and numerical validations for these algorithms. It is expected that the results obtained in this dissertation to lay the groundwork for their further development and expanded adaption in the deployment of tomography diagnostics in various practical applications.

We report on the design, fabrication, and first tests of a tomographic scanner developed for proton computed tomography (pCT) of head-sized objects. After extensive preclinical testing, pCT is intended to be employed in support of proton therapy treatment planning and pre-treatment verification in patients undergoing particle-beam therapy. The scanner consists of two silicon-strip telescopes that track individual protons before and after the phantom, and a novel multistage scintillation detector that measures a combination of the residual energy and range of the proton, from which we derive the water equivalent path length (WEPL) of the protons in the scanned object. The set of WEPL values and the associated paths of protons passing through the object over a 360° angular scan are processed by an iterative, parallelizable reconstruction algorithm that runs on modern GP-GPU hardware. In order to assess the performance of the scanner, we have performed tests with 200 MeV protons from the synchrotron of the Loma Linda University Medical Center and the IBA cyclotron of the Northwestern Medicine Chicago Proton Center. Our first objective was calibration of the instrument, including tracker channel maps and alignment as well as the WEPL calibration. Then we performed the first CT scans on a series of phantoms. The very high sustained rate of data acquisition, exceeding one million protons per second, allowed a full 360° scan to be completed in less than 10 minutes, and reconstruction of a CATPHAN 404 phantom verified accurate reconstruction of the proton relative stopping power in a variety of materials.

Since the 1990s, the number of clinical proton therapy facilities around the world has been growing exponentially. Because of this, and the lack of imaging support for proton therapy in the treatment room, a renewed interest in proton radiography and computed tomography (CT) has emerged. This imaging modality was largely abandoned in the 1970s and '80s in favor of the already successful x-ray CT, for reasons including long acquisition times and inadequate spatial resolution. Protons are particularly useful for radiotherapy because of their well-defined range in matter and their favorable energy profile which facilitates greater conformality than other radiotherapies; however, in order to realize the full potential of proton radiotherapy, the range of protons in the patient must be precisely known. Presently, proton therapy treatment planning is accomplished by taking x-ray CTs of the patient and converting each voxel into proton relative stopping power with respect to water (RSP) via a stoichiometrically-acquired calibration curve. However, since there is no unique relationship between Hounsfield values and RSP, this procedure has inherent uncertainties of a few percent in the proton range, requiring additional distal uncertainty margins in proton treatment plans. In contrast to x-ray CT, proton CT measures the RSP of an object directly,eliminating the need for Hounsfield-value-to-RSP conversion.In the prototype proton CT scanner that we have developed, a low-intensity beam of 200 MeV protons traverses a patient, entirely, and stops in a downstream energy/range detector. The entry and exit vectors of each proton are measured in order to determine a most-likely path of the proton through the object, and the response of the energy/range detector is converted to the water-equivalent path length of each proton in the object. These measurements are made at many angles between 0 and 360 degrees in order to reconstruct a three-dimensional map of proton RSP in the object.In

In a standard compressive sampling (CS) setup, we develop a universal algorithm where multiple CS reconstruction algorithms participate and their outputs are fused to achieve a better reconstruction performance. The new method is called universal algorithm for CS (UACS) that is iterative in nature and has a restricted isometry property (RIP) based theoretical convergence guarantee. It is shown that if one participating algorithm in the design has a converging recurrence inequality relation then the UACS also holds a converging recurrence inequality relation over iterations. An example of the UACS is presented and studied through simulations for demonstrating its flexibility and performance improvement., QC 20160617

The European Solar Telescope is a 4-m planned facility designed to have high spatial resolution capabilities tounderstand the mechanisms of magnetic coupling in the chromosphere and the photosphere. It will feature both aconventional and a multi-conjugate adaptive optics (AO) of similar complexity than the systems for night-timeExtremely Large Telescopes. A particularity of solar AO is that it uses the solar granulation as a reference; therefore thewavefront sensing is performed using correlations on images with a field of view of ~10”. A sensor collecting such awide field of view averages wavefront information from different sky directions, affecting the sensing of high altitudeturbulence, the sampling of which does not depend anymore on the size of the subapertures only, but rather on the size ofthe projection of the extended field of view. Understanding this effect is crucial for the design of future solar facilities,i.e. to choose the adequate height of the DMs on MCAO systems, and also to predict the quality of the reconstructionthat such system would be able to achieve. For that reason, we have studied wide field sensing and found the analyticalequations that describe the process, in order to use this information as an input to improved designs of solar AO systems.

We present a computational approach for fast approximation of nonlinear tomographic reconstruction methods by filtered backprojection (FBP) methods. Algebraic reconstruction algorithms are the methods of choice in a wide range of tomographic applications, yet they require significant computation time, restricting their usefulness. We build upon recent work on the approximation of linear algebraic reconstruction methods and extend the approach to the approximation of nonlinear reconstruction methods which are common in practice. We demonstrate that if a blueprint image is available that is sufficiently similar to the scanned object, our approach can compute reconstructions that approximate iterative nonlinear methods, yet have the same speed as FBP.

Considering the defects of the traditional orthogonal matching pursuit (OMP) algorithm and the sparsity adaptive matching pursuit (SAMP) algorithm, the multi-matching orthogonal pursuit (MMOP) algorithm was proposed under the condition of blind sparse with noise speech. The proposed algorithm used method of the match more than one atom to solve the over matching and less matching problems of SAMP algorithm as well as method of growing and shrinking atomic set. In addition, a new method of speech de-noising and the initial step, and the phased step was also proposed to reconstruct the original speech signal. The results show that the proposed algorithm can not only solve the SAMP algorithm problem of over matching and less matching, but also improve the matching speed and lessen the iterations number, and at the same time improve the accuracy of reconstruction. What’s more, the proposed algorithm can also be applied to the common noisy speech, with a better speech quality evaluation than the traditional OMP algorithm and SAMP algorithm.

The track reconstruction algorithms of the ATLAS experiment have demonstrated excellent performance in all of the data delivered so far by the LHC. The expected large increase in the number of interactions per bunch crossing introduces new challenges both in the computational aspects and physics performance of the algorithms. With the aim of taking advantage of modern CPU design and optimizing memory and CPU usage in the reconstruction algorithms a number of projects are being pursued. These include rationalization of the event data model, vectorization of the core components of the algorithms, and removing algorithm bottlenecks by using modern code analysis tools. Recent results of the advances made in these ongoing projects indicate up to three-fold speedup in the optimized modules while in some modules the code size could be reduced by up to 97% leading to higher readability, maintainability and decreased interface complexity., QC 20150608

The PI and collaborators developed novel algorithms for positioning, building on earlier developments in matrix completion and high-dimensional statistics. In particular, a distributed version of matrix completion-based positioning, and a gossip version of low-rank approximation were developed. A convex relaxation for positioning in the presence of noise was shown to be constant-optimal. Additional contributions were made in several other areas: Finding dense substructures of large networks in nearly linear time; Approximate message passing algorithms and in particular their application to spatially-coupled compressed sensing; measures of statistical significance in high dimension., The original document contains color images.

Nonlinear microwave imaging heavily relies on an accurate numerical electromagnetic model of the antenna system. The model is used to simulate scattering data that is compared to its measured counterpart in order to reconstruct the image. In this paper an antenna system immersed in water is used to image different canonical objects in order to investigate the implication of modeling errors on the final reconstruction using a time domain-based iterative inverse reconstruction algorithm and three-dimensional FDTD modeling. With the test objects immersed in a background of air and tap water, respectively, we have studied the impact of antenna modeling errors, errors in the modeling of the background media, and made a comparison with a two-dimensional version of the algorithm. In conclusion even small modeling errors in the antennas can significantly alter the reconstructed image. Since the image reconstruction procedure is highly nonlinear general conclusions are very difficult to make. In our case it means that with the antenna system immersed in water and using our present FDTD-based electromagnetic model the imaging results are improved if refraining from modeling the water-wall-air interface and instead just use a homogeneous background of water in the model.

In the harsh environment of the Large Hadron Collider at CERN (design luminosity of 1034 cm"2 s-1) efficient reconstruction of vertices is crucial for many physics analyses. Described in this paper is the expected performance of the vertex reconstruction used in the ATLAS experiment. The algorithms for the reconstruction of primary and secondary vertices as well as for finding photon conversions and vertex reconstruction in jets are described. The implementation of vertex algorithms which follows a very modular design based on object-oriented C++ is presented. A user-friendly concept allows event reconstruction and physics analyses to compare and optimize their choice among different vertex reconstruction strategies. The performance of implemented algorithms has been studied on a variety of Monte Carlo samples and results are presented. © 2010 IEEE.

This program was intended to develop theory a broad range of algorithms and proof of concept in detection, estimation, and reconstruction of objects embedded in turbid media, which hamper visibility (such as fog, clouds, smoke, and other types of aerosol particles).

A growing interest in the development of dual modality PET/CT scanners prompts the comparative study of numerous scintillators to select the best one, which could be used simultaneously in PET detectors working in the pulsing mode and in the CT detectors working in the current mode. In the comparative measurements, done in the same experimental conditions, various samples of BGO, GSO, GSO:Ce,Zr, LGSO, LSO, LYSO, MLS, LaCl3, LaBr3 and CWO scintillators were tested. The measurements covered a determination of the light output, energy resolution, non-proportionality of the light yield, decay times of the light pulses and for the selected crystals their time resolution for 511 keV annihilation quanta. Moreover, a comparative study of afterglow, induced by 60 keV gamma-rays from a strong 241-Am source (13.9 GBq), was done in the second range of time. The LSO-like crystals are best in the PET scanners application. However, they do not fit to the CT requirements, due to a high afterglow. The studies conclude that besides of the well known BGO, only GSO:Ce and most likely LaBr3 might be considered for the simultaneous PET/CT detectors.

BandPass Sampling (BPS) is an undersampling technique by intentional aliasing. BPS enables one to have an interface between the IF stage and the ADC in a radio receiver. Conventional uniform BPS at Nyquist rate normally results in a low Signal-to-Noise Ratio (SNR) due to noise spectrum aliasing. The noise (e.g. kT/C noise introduced in a voltage-mode sampler) is combined in each of the Nyquist bands within the bandwidth of the sampling device. Also timing jitter causes a performance degradation in BPS. In this paper, signal spectrum aliasing, noise aliasing and jitter effects in BPS is analyzed. It is verified by simulation that NonUniform Sampling (NUS) has the potential to suppress signal spectrum aliasing and relax the requirement on the anti-aliasing (AA) filter. Jitter effects in BPS are compared to LowPass Sampling (LPS) case. However, a signal cannot be reconstructed from its nonuniform samples by using only ideal lowpass filtering (classic Shannon's reconstruction). Finally, signal reconstruction in the presence of noise and jitter are investigated for three Reconstruction Algorithms (RAs) aimed at NUS., QC 20100910

Purpose: Improve tumor localization in In-111 ProstaScint SPECT scans through improved reconstruction and identification/removal of non-specific blood pool volumes using simultaneously acquired Tc-99m tagged red blood cell (RBC) SPECT scans. Methods: We chose 30 patients with a history of prostate cancer who had undergone CT/MR and simultaneous Tc-99m RBC/In-111 ProstaScint SPECT scans due to rising PSA. To estimate the impact of reconstruction methods on anatomic definition and artifacts, SPECT volume data sets were reconstructed using ordered set-expectation maximization (OS-EM) with varying numbers of iterations and subsets, and these were compared against each other and against standard filtered back projection (FBP) reconstruction. Non-blood pool bladder activity in the Tc-99m scans was suppressed prior to subtraction from the In-111 scans by using an averaging algorithm within an ellipsoid volume encompassing the bladder. Outside the ellipsoid volume, Tc-99m voxel values were subtracted from the corresponding In-111 voxels after normalization of the data sets based on peak activity within the descending aorta. Results: OS-EM reconstruction using 3 iterations and 45 subsets showed improved representation of anatomy compared with FBP. Bladder suppression reduced artifacts in the prostate bed. The subtraction method reduced the blood pool signal, confirmed visually by superimposition with matched CT/MR scans. Conclusions: OS-EM reconstruction together with bladder suppression and subtraction of the blood pool may help improve the specificity of ProstaScint SPECT interpretation and increase its utility in radiation therapy treatment planning., QC 20111205

application/pdf, Three-dimensional (3D) models of outdoor scenes can be widely used in a number of fields such as object recognition, navigation, scenic simulation, and mixed reality. Such models are often made manually with high cost, so that automatic 3D reconstruction has been widely investigated. In related works, a dense 3D model is generated by using a stereo method. However, such approaches cannot use several hundred images together or dense depth estimation of large constructs and urban environments because it is difficult to accurately calibrate a large number of cameras. This paper proposes a dense 3D reconstruction method that uses multiple image sequences. First, our method estimates extrinsic camera parameters of each image sequence, then reconstructs a dense 3D model of a scene using an extended multi-baseline stereo and voxel voting technique., MFI 2003 : International Conference on Multisensor Fusion and Integration for Intelligent Systems , Aug 1, 2003 , Tokyo, Japan, conference paper

application/pdf, Three-dimensional (3D) models of outdoor scenes can be widely used in a number of fields such as object recognition, navigation, scenic simulation, and mixed reality. Such models are often made manually with high cost, so that automatic 3D reconstruction has been widely investigated. In related works, a dense 3D model is generated by using a stereo method. However, such approaches cannot use several hundred images together or dense depth estimation of large constructs and urban environments because it is difficult to accurately calibrate a large number of cameras. This paper proposes a dense 3D reconstruction method that uses multiple image sequences. First, our method estimates extrinsic camera parameters of each image sequence, then reconstructs a dense 3D model of a scene using an extended multi-baseline stereo and voxel voting technique.

Two methods of chromotomographic reconstruction applicable to the MWIR region of the electromagnetic spectrum were investigated. The first method involved the decomposition of the datacube into principal component space using the center order, broadband image to form and estimate the spatial components. Then this estimate was used to deconvolve an estimate of the spectral components. The second method used an extension of the traditional maximum-likelihood approach to suppress signal-independent noise in reconstructions of simulated low background scenes.

This report represents the work of eight (8) researchers providing support for six (6) ongoing programs at AFRL/SNH. The research topics include: (1) Optical properties of turbulent media. (2) InP based HEMPTs, MISFETs, and MSMs. (3) Bulk crystal growth of ZnO and GaN substrates. (4) Bulk crystal growth of InP substrates. (5) Mathematics of chromotomography. (6) Thermionic emission based thermal detectors. In order to establish the context of the work, this report is best read alongside the published papers and presentations of the primary AFRL scientists conducting the research. Among the new concepts and achievements that emerged from this work are: a new class of uncooled thermal infrared imagers based on Schottky diodes, two orders of magnitude improvement in the convergence rate of chromotomography algorithms, a new amenothermal growth technique for GaN substrates, and a 25 GHz InP wafer lined optical receiver.

Peer Reviewed, Postprint (published version)

This thesis examines the application of internal imaging technologies to the detection of defects within felled logs. This application requires images of moderate resolution to be generated at high-speeds. Transmission tomography is suggested as the most appropriate imaging technology, with particular reference to X-ray transmission tomography. High-speed X-ray scanners suitable to the application exist. An international literature search was performed to find reconstruction systems capable of high-speed reconstruction of the X-ray data. Those systems claiming to be high-speed are discussed in regard to their means of achieving high-speed reconstruction. The discussions show that most effort has been directed toward sophisticated hardware implementation of the reconstruction process, rather then the mathematics of the reconstruction process itself. To increase the achievable rate of reconstruction, the mathematics of each process in the reconstruction algorithm are examined in this thesis, with the aim of reducing computational complexity. Convolution backprojection is the most commonly used reconstruction method in transmission tomography when the X-ray data are complete, and can be neatly separated into two separate processes, convolution (filtering) and backprojection, as the name suggests. This reconstruction method was chosen as suitable for the log processing application because of its mathematical simplicity and quality of image reconstruction. Truncation of the convolution kernel is examined, and simulated results show adequate reconstruction quality with significant truncation. An inexpensive hardware design capable of performing the convolution operation in real-time is developed. The backprojection process normally employed is computationally expensive, and is the major encumbrance to the realisation of simple high speed image reconstruction from projections. A new backprojection algorithm for use in high-speed parallel-ray tomographic reconstruction syst