L’imagerie proton est étudiée comme alternative à la tomodensitométrie X pour la planification de traitement en hadronthérapie. En obtenant directement les pouvoirs d’arrêt relatifs des tissus, l’incertitude sur le parcours des particules pourrait être réduite. Un scanner à protons est constitué d’un calorimètre ou d’un détecteur de parcours afin d’obtenir l’information sur l’énergie déposée par chaque proton dans l’objet imagé et de deux ensembles de trajectographes enregistrant la position et direction de chaque particule en amont et en aval de l’objet. Ce travail concerne l’étude des données d’un scanner à protons et l’utilisation possible de toutes les informations enregistrées. Une étude de reconstruction d’image a permis de montrer que les informations sur le taux de transmission et sur la déviation de chaque particule peuvent être utilisées pour produire des images aux propriétés visuelles intéressantes pour le diagnostic. La preuve de concept de la possibilité d’une imagerie quantitative utilisant ces informations est présentée. Ces résultats sont une première étape vers l’imagerie proton utilisant toutes les données enregistrées. / Proton computed tomography is being studied as an alternative to X-ray CT imaging for charged particle therapy treatment planning. By directly mapping the relative stopping power of the tissues, the uncertainty on the range of the particles could be reduced. A proton scanner consists in a calorimeter or range-meter to obtain the information on the energy lost by each proton in the object, as well as two sets of tracking planes to record the position and direction of each particle upstream and downstream from the object. This work concerns the study of the outputs of a proton scanner and the possible use of all the recorded information. A reconstruction study made it possible to show that the information on the transmission rate and on the scattering of each particle can be used to produce images with visual properties that could be of interest for diagnostics. The proof of concept of the possibility of quantitative imaging using this information is also put forward. These results are the first step towards a clinical use of proton imaging with all the recorded data.

No description available.

Segmentation, recognition and 3D reconstruction of objects have been cutting-edge research topics, which have many applications ranging from environmental and medical to geographical applications as well as intelligent transportation. In this dissertation, I focus on the study of segmentation, recognition and 3D reconstruction of objects using LiDAR data/MRI. Three main works are that (I). Feature extraction algorithm based on sparse LiDAR data. A novel method has been proposed for feature extraction from sparse LiDAR data. The algorithm and the related principles have been described. Also, I have tested and discussed the choices and roles of parameters. By using correlation of neighboring points directly, statistic distribution of normal vectors at each point has been effectively used to determine the category of the selected point. (II). Segmentation and 3D reconstruction of objects based on LiDAR/MRI. The proposed method includes that the 3D LiDAR data are layered, that different categories are segmented, and that 3D canopy surfaces of individual tree crowns and clusters of trees are reconstructed from LiDAR point data based on a region active contour model. The proposed method allows for delineations of 3D forest canopy naturally from the contours of raw LiDAR point clouds. The proposed model is suitable not only for a series of ideal cone shapes, but also for other kinds of 3D shapes as well as other kinds dataset such as MRI. (III). Novel algorithms for recognition of objects based on LiDAR/MRI. Aimed to the sparse LiDAR data, the feature extraction algorithm has been proposed and applied to classify the building and trees. More importantly, the novel algorithms based on level set methods have been provided and employed to recognize not only the buildings and trees, the different trees (e.g. Oak trees and Douglas firs), but also the subthalamus nuclei (STNs). By using the novel algorithms based on level set method, a 3D model of the subthalamus nuclei (STNs) in the brain has been successfully reconstructed based on the statistical data of previous investigations of an anatomy atlas as reference. The 3D rendering of the subthalamic nuclei and the skull directly from MR imaging is also utilized to determine the 3D coordinates of the STNs in the brain. In summary, the novel methods and algorithms of segmentation, recognition and 3D reconstruction of objects have been proposed. The related experiments have been done to test and confirm the validation of the proposed methods. The experimental results also demonstrate the accuracy, efficiency and effectiveness of the proposed methods. A framework for segmentation, recognition and 3D reconstruction of objects has been established, which has been applied to many research areas.

Monitoring of public areas and their automatic real-time processing became increasingly significant due to the changing security situation in the world. However, the problem is an analysis of low-quality records, where even the state-of-the-art methods fail in some cases. This work investigates an important area of image similarity – biometric identification based on face image. The work deals primarily with the face super-resolution from a sequence of low-resolution images and it compares this approach to the single-frame methods, that are still considered as the most accurate. A new dataset was created for this purpose, which is directly designed for the multi-frame face super-resolution methods from the low-resolution input sequence, and it is of comparable size with the leading world datasets. The results were evaluated by both a survey of human perception and defined objective metrics. A hypothesis that multi-frame methods achieve better results than single-frame methods was proved by a comparison of both methods. Architectures, source code and the dataset were released. That caused a creation of the basis for future research in this field.

The main goal of this thesis was to increase reconstruction quality of sparse sampled microscopic images by using neural networks. The thesis will cover various approaches for image reconstruction and will also include descriptions of implementations, which were used. Implementations will be evaluated based on quality of reconstruction, but also based on segmentation, which could be their main possible application.

In der nuklearmedizinischen Bildgebung mit Anger-Kameras wird ein höheres Auflösungsvermögen durch Limitierung der Nachweiseffizienz erreicht. Compton Cameras können die Nachweiseffizienz mittels elektronischer Kollimation, die zur Ortsbestimmung die Compton-Kinematik anwendet, erhöhen. Ein alternativer Ansatz für die Konstruktion einer Compton Camera, die Streu- und Absorptionsebenen in einer Ebene kombiniert, wurde in der Vergangenheit untersucht. Die sogenannte Single Plane Compton Camera ist in der Lage, Punktquellen im Vakuum getrennt aufzulösen. Zur Darstellung komplexerer Bildinhalte wird die Optimierung der Bildrekonstruktion angestrebt. Diese umfasst ein umfängliches Verständnis des Messprinzips, das in dieser Arbeit dargelegt wird. Jeweils ein 3D-Rekonstruktionsalgorithmus wurde für konventionell gebinnte und List-Mode-Daten implementiert. Anhand eines vorliegenden Simulationsdatensatzes einer einfachen Detektorkonfiguration wurden Messdaten generiert und rekonstruiert. Es konnte gezeigt werden, dass aufgrund der hohen Zählstatistik ein robustes Signal-to-Noise-Ratio erhalten wird. List-Mode-Verfahren eignen sich aufgrund eines höheren Rechenaufwandes nicht. Die mittlere Ortsinformation der Ereignisse ist systembedingt gering und beeinträchtigt die Ortsauflösung, welche für E = 662 keV etwa 15 mm in einem Abstand von 50cmm beträgt. Eine Verbesserung der Auflösung ist durch die Algorithmen nicht möglich, sondern umfasst technische Maßnahmen, welche anhand dieser Arbeit in weiteren Studien umgesetzt werden können.:1 Einleitung 2 Single Plane Compton Imaging 3 Bildrekonstruktion beim Single Plane Compton Imaging 4 Materialien und Methoden 5 Ergebnisse 6 Diskussion 7 Zusammenfassung und Ausblick Literaturverzeichnis Abkürzungsverzeichnis Abbildungsverzeichnis Tabellenverzeichnis Selbstständigkeitserklärung A Herleitung des ML-EM Algorithmus B Tiefergehende Informationen zu OE C C++-Code für die Bildrekonstruktion beim SPCI / In nuclear medicine imaging, the Anger camera imposes a limit on the detection efficiency in order to improve the spatial resolution. The detection efficiency can be increased with electronically collimated systems known as Compton Cameras, which use the kinematics of Compton scattering to locate the detected events. An alternative approach to the design of a Compton Camera combining scatter and absorption planes was investigated in the past. It was shown that the so-called Single Plane Compton Camera is able to separately reconstruct two point sources in empty space. Further optimization is required to reconstruct more complex images. Thus, an extensive understanding of the measurement principle is provided. Two 3D-algorithms were implemented for binned data and list mode data. Measurement data were generated by means of an existing simulated data set of a simple detector design and reconstructed. It is shown that a robust signal-to-noise ratio can be achieved due to high numbers of detected counts. List mode algorithms produce high computational costs and binned algorithms may be used instead. The average position information is low and imposes a negative impact on the spatial resolution, which is about 15mm at a distance of 50mm for E = 662 keV. The implemented algorithms cannot increase the spatial resolution due to lack of precise position information. Therefore, future studies should focus on technical measures, which are given in this thesis.:1 Einleitung 2 Single Plane Compton Imaging 3 Bildrekonstruktion beim Single Plane Compton Imaging 4 Materialien und Methoden 5 Ergebnisse 6 Diskussion 7 Zusammenfassung und Ausblick Literaturverzeichnis Abkürzungsverzeichnis Abbildungsverzeichnis Tabellenverzeichnis Selbstständigkeitserklärung A Herleitung des ML-EM Algorithmus B Tiefergehende Informationen zu OE C C++-Code für die Bildrekonstruktion beim SPCI

The aim of this thesis is to investigate applicability of regulariza- tion by Krylov subspace methods to discrete inverse problems arising in single particle analysis (SPA). We start with a smooth model formulation and describe its discretization, yielding an ill-posed inverse problem Ax ≈ b, where A is a lin- ear operator and b represents the measured noisy data. We provide theoretical background and overview of selected methods for the solution of general linear inverse problems. Then we focus on specific properties of inverse problems from SPA, and provide experimental analysis based on synthetically generated SPA datasets (experiments are performed in the Matlab enviroment). Turning to the solution of our inverse problem, we investigate in particular an approach based on iterative Hybrid LSQR with inner Tikhonov regularization. A reliable stopping criterion for the iterative part as well as parameter-choice method for the inner regularization are discussed. Providing a complete implementation of the proposed solver (in Matlab and in C++), its performance is evaluated on various SPA model datasets, considering high levels of noise and realistic distri- bution of orientations of scanning angles. Comparison to other regularization methods, including the ART method traditionally used in SPA,...

Contrast agents are often crucial in medical imaging for disease diagnosis. Novel contrast agents, such as gold nanoparticles (AuNPs) and lanthanides, are being ex- plored for a variety of clinical applications. Preclinical testing of these contrast agents is necessary before being approved for use in humans, which requires the use of small animal imaging techniques. Small animal imaging demands the detection of these contrast agents in trace amounts at acceptable imaging time and radiation dose. Two such imaging techniques include x-ray fluorescence computed tomography (XFCT) and photon-counting CT (PCCT). XFCT combines the principles of CT with x-ray fluorescence by detecting fluorescent x-rays from contrast agents at various projections to reconstruct contrast agent maps. XFCT can image trace amounts of AuNPs but is limited to small animal imaging due to fluorescent x-ray attenuation and scatter. PCCT uses photon-counting detectors that separate the CT data into energy bins. This enables contrast agent detection by recognizing the energy dependence of x-ray attenuation in different materials, independent of AuNP depth, and can provide anatomical information that XFCT cannot. To achieve the best of both worlds, we modeled and built a table-top x-ray imaging system capable of simultaneous XFCT and PCCT imaging. We used Monte Carlo simulation software for the following work in XFCT imaging of AuNPs. We simulated XFCT induced by x-ray, electron, and proton beams scanning a small animal-sized object (phantom) containing AuNPs with Monte Carlo techniques. XFCT induced by x-rays resulted in the best image quality of AuNPs, however high-energy electron and medium-energy proton XFCT may be feasible for on-board x-ray fluorescence techniques during radiation therapy. We then simulated a scan of a phantom containing AuNPs on a table-top system to optimize the detector arrangement, size, and data acquisition strategy based on the resulting XFCT image quality and available detector equipment. To enable faster XFCT data acquisition, we separately simulated another AuNP phantom and determined the best collimator geometry for Au fluorescent x-ray detection. We also performed experiments on our table-top x-ray imaging system in the lab. Phantoms containing multiples of three lanthanide contrast agents were scanned on our tabletop x-ray imaging system using a photon-counting detector capable of sustaining high x-ray fluxes that enabled PCCT. We used a novel subtraction algorithm for reconstructing separate contrast agent maps; all lanthanides were distinct at low concentrations including gadolinium and holmium that are close in atomic number. Finally, we performed the first simultaneous XFCT and PCCT scan of a phantom and mice containing both gadolinium and gold based on the optimized parameters from our simulations. This dissertation outlines the development of our tabletop x-ray imaging system and the optimization of the complex parameters necessary to obtain XFCT and PCCT images of multiple contrast agents at biologically-relevant concentrations. / Graduate

Electrical impedance imaging is an imaging technique which has the capability of revealing the spatial distribution of the electrical impedance inside biological tissues. Classical electrical impedance imaging including Electrical Impedance Tomography (EIT) typically has low resolution. Advances in electrical impedance imaging typically involve methods that either increase image resolution or image contrast. This study investigates the possibility of the resolution improvement for electrical impedance imaging using motion, and presents a novel high-resolution and calibrated impedance imaging method called Scanning electrical Impedance Imaging (SII). SII uses an electrical probe held at a known voltage and scanned over a thin sample immersed in a conductive medium on a grounded conducting plane to obtain high-resolution calibrated impedance images of samples. For system improvement and image reconstruction, a numerical model is developed to describe the SII system. This model simulates the measurement process by solving a 3-D electrostatic field at each scanning position using a modified approach of the finite difference method (FDM). The simulation consists of a quasi-statics problem involving inhomogeneous media with a complicated boundary condition. This 3-D model is used to optimize both the probe height and the shield-spacing for probe fabrication and also to evaluate system parameters including the frequency and the resistor in the peripheral circuit. Based on this model, an approach is also developed to quantifying conductivity values using the SII system. However, a large computational cost due to the motion involved in SII leads to challenges for a fast and accurate image reconstruction based on this 3-D model. Alternative fast models are derived as a replacement of the 3-D model for quick image reconstruction. In particular, the Modified Linear Approximation (MLA) involving two conductivity-weighted convolutions based on the reciprocity principle, explains the function of the special shield design introduced in the SII system reasonably well. Based on the MLA a nonlinear inverse method using total variation regularization and the Polak-Ribi'{e}re variant of the nonlinear conjugate-gradient method is developed for fast image reconstruction of the SII system. The inverse method is accelerated using convolution which eliminates the requirement of a numerical solver for the 3-D electrostatic field. 2-D images of small biological tissues and cells are measured using the SII system. The corresponding conductivity images are reconstructed using the MLA method. The successful improvement of resolution shown in both simulation and experimental results demonstrates that the idea of this approach can potentially be expanded to other imaging modalities for resolution improvement using motion.

This dissertation approaches high resolution scatterometry from a new perspective. Three related general topics are addressed: high resolution σ^0 imaging, wind estimation from high resolution σ^0 images over the ocean, and high resolution wind estimation directly from the scatterometer measurements. Theories of each topic are developed, and previous approaches are generalized and formalized. Improved processing algorithms for these theories are developed, implemented for particular scatterometers, and analyzed. Specific results and contributions are noted below. The σ^0 imaging problem is approached as the inversion of a noisy aperture-filtered sampling operation-extending the current theory to deal explicitly with noise. A maximum aposteriori (MAP) reconstruction estimator is developed to regularize the problem and deal appropriately with noise. The method is applied to the SeaWinds scatterometer and the Advanced Scatterometer (ASCAT). The MAP approach produces high resolution σ^0 images without introducing the ad-hoc processing steps employed in previous methods. An ultra high resolution (UHR) wind product has been previously developed and shown to produce valuable high resolution information, but the theory has not been formalized. This dissertation develops the UHR sampling model and noise model, and explicitly states the implicit assumptions involved. Improved UHR wind retrieval methods are also developed. The developments in the σ^0 imaging problem are extended to deal with the nonlinearities involved in wind field estimation. A MAP wind field reconstruction estimator is developed and implemented for the SeaWinds scatterometer. MAP wind reconstruction produces a wind field estimate that is consistent with the conventional product, but with higher resolution. The MAP reconstruction estimates have a resolution similar to the UHR estimates, but with less noise. A hurricane wind model is applied to obtain an informative prior used in MAP estimation, which reduces noise and ameliorates ambiguity selection and rain contamination.