<p>Standard MRI is used to image objects at rest. In addition to standard MRI images, which measure tissues at rest, Phase Contrast MRI can be used to quantify the motion of blood and tissue in the human body. The current method used in Phase Contrast MRI is time consuming. The development of new trajectories has minimized imaging time, but creates sub-sampling errors. The proposed method uses regularization of velocities and proton densities to eliminate errors arising from k-space under-sampling.</p> / Master of Applied Science (MASc)

Pour répondre à des questions biologiques émergentes, l’IRM 3D in vivo est une approche de choix, mais elle souffre principalement d’une faible résolution temporelle en raison d’une faible sensibilité. Par ailleurs, l’IRM gagnerait à une meilleure sensibilité aux agents de contrastes exogènes. Il est proposé ici des développements en IRM du petit animal permettant de réduire considérablement la durée d’acquisition des images à trois dimensions chez la souris, ou la détection de très faibles quantités d’agents de contraste. Ces développements reposent sur la manipulation de l’espace-k (espace des données acquises). La première partie de ce travail a reposé sur la mise en place d’une méthode d’acquisition rapide de l’imagerie 3D permettant de conserver la qualité de l’image. Le «keyhole» 3D a été la technique choisie pour accéder à une résolution temporelle très élevée. Ainsi, le temps d’acquisition en imagerie ciné 3D cardiaque, chez la souris, a été réduit par un facteur 4 tout en conservant la qualité de l’image (SSB) et les informations extraites. Le «keyhole» 3D est aussi une méthode favorable à la mesure de prise de contraste. La biodistribution d’agent de contraste, peut être suivie en imagerie 3D à contrastes T1 et T2* dans le corps entier de la souris en quasi temps réel. La manipulation de l’espace-k permet aussi d’améliorer la qualité de l’image en réalisant une imagerie très sensible au contraste T2*. Pour cela, la correction de mouvements intrinsèques, comme ceux issus de la respiration au niveau de l’abdomen, générant un effet de perte de résolution spatiale, est indispensable. L’utilisation d’un écho navigateur permettant de détecter et de supprimer tous les signaux indésirables apporte une amélioration nette de la qualité d’image. Le seuil de détection de l’agent de contraste testé est d’ailleurs inférieur à 100 picomole de fer par kilogramme dans l’abdomen de souris. L’utilisation des propriétés de l’espace-k montre à quel point la qualité de l’image peut être améliorée et adaptée à l’information souhaitée. C’est un moyen peu couteux et efficace pour rendre l’imagerie par résonance magnétique encore plus performante en terme de résolution spatiale et de résolution temporelle. / In vivo 3D MRI is a powerful method which can be used to answer emerging biological issues. However, low temporal resolution due to intrinsic low sensitivity is one of its main drawbacks. Similarly, breakthroughs are needed to detect by MRI low-concentrated contrast agents used for molecular imaging. In this work, several methodology developments in small animals are proposed to greatly reduce acquisition times of 3D MRI and to increase contrast sensitivity to T2* agents. Both achievements were performed through the manipulation of the k-space, i.e the acquired data space in a retrospective approach. To achieve very high temporal resolution a 3D keyhole technique was chosen. This allowed the acquisition time in cardiac 3D-cine imaging in mice to be reduced by a factor 4. Image quality (signal-to-noise ratio) and the extracted functional data were preserved. Interestingly, 3D keyhole imaging also allowed the evaluation of T1 and T2* contrast enhancement and biodistribution in real time in the whole mouse body. In the last part of the work, the goal was to generate highly T2*-sensitive 3D images in mouse abdomen to detect diluted iron-oxide-based contrast agents. The use of a navigator echo enabled efficient motion correction and detection threshold of less than 100 picomol iron per kilogram. The results are discussed in a general frame of future applications and development of fast and highly-resolved 3D imaging.

MRI is a very useful imaging modality in medical imaging for both diagnostic as well as functional studies. It provides excellent soft tissue contrast in several diagnostic studies. It is widely used to study the functional aspects of brain and to study the diffusion of water molecules across tissues. Image acquisition in MR is slow due to longer data acquisition time, gradient ramp-up and stabilization delays. Repetitive scans are also needed to overcome any artefacts due to patient motion, field inhomogeneity and to improve signal to noise ratio (SNR). Scanning becomes di cult in case of claustrophobic patients, and in younger/older patients who are unable to cooperate and prone to uncontrollable motions inside the scanner. New MR procedures, advanced research in neuro and functional imaging are demanding better resolutions and scan speeds which implies there is need to acquire more data in a shorter time frame. The hardware approach to faster k-space scanning methods involves efficient pulse sequence and gradient waveform design methods. Such methods have reached a physical and physiological limit. Alternately, methods have been proposed to reduce the scan time by under sampling the k-space data. Since the advent of Compressive Sensing (CS), there has been a tremendous interest in developing under sampling matrices for MRI. Mathematical assumptions on the probability distribution function (pdf) of k-space have led researchers to come up with efficient under sampling matrices for sampling MR k-space data. The recent approaches adaptively sample the k-space, based on the k-space of reference image as the probability distribution instead of a mathematical distribution, to come with an efficient under sampling scheme. In general, the methods use a deterministic central circular/square region and probabilistic sampling of the rest of the k-space. In these methods, the sampling distribution may not follow the selected pdf and viii Adaptive Sampling Pattern Design Methods for MR Images the selection of deterministic and probabilistic sampling distribution parameters are heuristic in nature. Two novel adaptive Variable Density Sampling (VDS) methods are proposed to address the heuristic nature of the sampling k-space such that the selected pdf matches the k-space energy distribution of a given fully sampled reference k-space or the MR image. The proposed methods use a novel approach of binning the pdf derived from the fully sampled k-space energy distribution of a reference image. The normalized k-space magnitude spectrum of the reference image is taken as a 2D probability distribution function which is divided in to number of exponentially weighted magnitude bins obtained from the corresponding histogram of the k-space magnitude spectrum. In the first method, the normalized k-space histogram is binned exponentially, and the resulting exponentially binned 2D pdf is used with a suitable control parameter to obtain a sampling pattern of desired under sampling ratio. The resulting sampling pattern is an adaptive VDS pattern mimicking the energy distribution of the original k-space. In the second method, the binning of the magnitude spectrum of k-space is followed by ranking of the bins by its spectral energy content. A cost function is de ned to evaluate the k-space energy being captured by the bin. The samples are selected from the energy rank ordered bins using a Knapsack constraint. The energy ranking and the Knapsack criterion result in the selection of sampling points from the highly relevant bins and gives a very robust sampling grid with well defined sparsity level. Finally, the feasibility of developing a single adaptive VDS sampling pattern for a organ specific or multi-slice MR imaging, using the concept of binning of magnitude spectrum of the k-space, is investigated. Based on the premise that k-space of different organs have a different energy distribution structure to one another, the MR images of organs can be classified based on their spectral content and develop a single adaptive VDS sampling pattern for imaging an organ or multiple slices of the same. The classification is done using the k-space bin histogram as feature vectors and k-means clustering. Based on the nearest distance to the centroid of the organ cluster, a template image is selected to generate the sampling grid for the organ under consideration. Using the state of the art MR reconstruction algorithms, the performance of the proposed novel adaptive Variable Density Sampling (VDS) methods using image quality measures is evaluated and compared with other VDS methods. The reconstructions show significant improvement in image quality parameters quantitatively and visual reduction in artefacts at 20% 15%, 10% and 5% under sampling

L'imagerie par résonance magnétique (IRM) est l'une des modalités d'imagerie les plus puissantes et les plus sures pour examiner le corps humain. L'IRM de haute résolution devrait aider à la compréhension et le diagnostic de nombreuses pathologies impliquant des lésions submillimétriques ou des maladies telles que la maladie d'Alzheimer et la sclérose en plaque. Bien que les systèmes à haut champ magnétique soient capables de fournir un rapport signal-sur-bruit permettant d'augmenter la résolution spatiale, le long temps d'acquisition et la sensibilité au mouvement continuent d'entraver l'utilisation de l'IRM de haute résolution. Malgré le développement de méthodes de correction du mouvement et du bruit physiologique, le long temps d'acquisition reste un obstacle majeur à l'IRM de haute résolution, en particulier dans les applications cliniques.Au cours de la dernière décennie, la nouvelle théorie du compressed sensing (CS) a proposé une solution prometteuse pour réduire le temps d'examen en IRM. Après avoir expliqué la théorie du compressed sensing, ce projet de thèse propose une étude empirique et quantitative du facteur de sous-échantillonnage maximum réalisable grâce au CS pour l'imagerie pondérée en T ₂ *.En outre, l'application de CS en IRM repose généralement sur l'utilisation de courbes d'échantillonnage simples telles que les lignes droites, spirales ou des légères variations de ces formes élémentaires qui ne tirent pas pleinement parti des degrés de liberté offerts par le hardware et ne peuvent être facilement adaptées à une distribution d'échantillonnage arbitraire. Dans cette thèse, j'ai introduit une méthode appelée SPARKLING, qui permet de surmonter ces limitations en adoptant une approche radicalement nouvelle de la conception de l'échantillonnage de l'espace-k. L'acronyme SPARKLING signifie Spreading Projection Algorithm for Rapid K-space sampLING. C'est une méthode flexible inspirée des techniques de stippling qui génère automatiquement, grâce à un algorithme d'optimisation, des courbes d'échantillonnage non-cartésiennes optimisées et compatibles avec les contraintes hardware de l'IRM en termes d'amplitude de gradient maximale et d'accélération maximale. Ces courbes d'échantillonnage sont conçues pour répondre à des critères clés pour un échantillonnage optimal : une distribution contrôlée des échantillons et une couverture de l'espace-k localement uniforme. Avant de s'engager dans des acquisitions, nous avons vérifié que notre système de gradient était bien capable d'exécuter ces trajectoires complexes. Nous avons implémenté une méthode de mesure de phase et avons observé une très bonne adéquation entre trajectoires prescrites et mesurées.Enfin, en alliant une efficacité d'échantillonnage avec le compressed sensing et l'imagerie parallèle, les trajectoires SPARKLING ont permis de réduire jusqu'à 20 fois le temps d'acquisition d'un examen IRM T ₂ * par rapport aux acquisitions cartésiennes de référence, sans détérioration de la qualité d'image. Ces résultats expérimentaux ont été obtenus à 7 Tesla pour de l'imagerie cérébrale in vivo. Par rapport aux stratégies d'échantillonnage non-cartésiennes usuelles (spirale et radiale), la technique proposée a également permis d'obtenir une qualité d'image supérieure. Enfin, l'approche proposée a été étendue à l'imagerie 3D et appliquée à 3 Tesla pour laquelle des résultats préliminaires ex vivo à une résolution isotrope de 0.6 mm suggèrent la possibilité d'atteindre des facteurs d'accélération très élevés jusqu'à 60 pour la pondération T ₂ * et l'imagerie pondérée en susceptibilité. / Magnetic resonance imaging (MRI) is one of the most powerful and safest imaging modalities for examining the human body. High-resolution MRI is expected to aid in the understanding and diagnosis of many neurodegenerative pathologies involving submillimetric lesions or morphological alterations, such as Alzheimer’s disease and multiple sclerosis. Although high-magnetic-field systems can deliver a sufficient signal-to-noise ratio (SNR) to increase spatial resolution, long scan times and motion sensitivity continue hindering the utilization of high resolution MRI. Despite the development of corrections for bulk and physiological motion, lengthy acquisition times remain a major obstacle to high-resolution acquisition, especially in clinical applications.In the last decade, the newly developed theory of compressed sensing (CS) offered a promising solution for reducing the MRI scan time. After having explained the theory of compressed sensing, this PhD project proposes an empirical and quantitative analysis of the maximum undersampling factor achievable with CS for T ₂ *-weighted imaging.Furthermore, the application of CS to MRI commonly relies on simple sampling patterns such as straight lines, spirals or slight variations of these elementary shapes, which do not take full advantage of the degrees of freedom offered by the hardware and cannot be easily adapted to fit an arbitrary sampling distribution. In this PhD thesis, I have introduced a method called SPARKLING, that may overcome these limitations by taking a radically new approach to the design of k-space sampling. The acronym SPARKLING stands for Spreading Projection Algorithm for Rapid K-space sampLING. It is a versatile method inspired from stippling techniques that automatically generates optimized non-Cartesian sampling patterns compatible with MR hardware constraints on maximum gradient amplitude and slew rate. These sampling curves are designed to comply with key criteria for optimal sampling: a controlled distribution of samples and a locally uniform k-space coverage. Before engaging into experiments, we verified that our gradient system was capable of executing the complex gradient waveforms. We implemented a local phase measurement method and we observed a very good adequacy between prescribed and measured k-space trajectories.Finally, combining sampling efficiency with compressed sensing and parallel imaging, the SPARKLING sampling patterns allowed up to 20-fold reductions in MR scan time, compared to fully-sampled Cartesian acquisitions, for T ₂ *-weighted imaging without deterioration of image quality, as demonstrated by our experimental results at 7 Tesla on in vivo human brains. In comparison to existing non-Cartesian sampling strategies (spiral and radial), the proposed technique also yielded superior image quality. Finally, the proposed approach was also extended to 3D imaging and applied at 3 Tesla for which preliminary results on ex vivo phantoms at 0.8 mm isotropic resolution suggest the possibility to reach very high acceleration factors up to 60 for T ₂ *-weighting and susceptibility-weighted imaging.

A novel non-invasive imaging method of unique k-space trajectory named “3D center-out EPI with cylindrical encoding” was developed and implemented for fast imaging of the human brain. The method based on a variant of 3D hybrid EPI combines advantages of the Cartesian and the radial encoding to achieve ultra-short echo time independent of spatial resolution and reasonably short echo train length yielding a quality image of high signal-to-noise ratio. Unlike rectilinear sampling, the method offers not only less motion and flow artifacts but enables also the undersampling capability. As a result, the method improves temporal resolution by shortening the measurement time. Nonetheless, artifacts induced from long-term drifts of the magnetic field as well as geometrical distortions caused by B0 inhomogeneity were removed with the average phase of the k-space center lines and an additional field map scan. Compared to other cylindrical k-space trajectories based on echo-planar imaging, which lead to progressively increasing echo time upon increasing the spatial resolution, the proposed method offers more benefits. As a significant application, imaging readout of the novel technique was applied to true 3D cine imaging which was later used in the combination of pseudo-continuous arterial spin labeling module in order to track a short arterial spin labeling (ASL) bolus of well-defined length along the fast passage through the large vessel compartment of the brain. Parametric maps of ASL signal change, estimated time-to-peak and ASL bolus width were extracted in order to characterize the macrovascular compartments of the brain-feeding arteries. Consequently, bolus dispersion within a single arterial branch was also assessed.

A novel non-invasive imaging method of unique k-space trajectory named “3D center-out EPI with cylindrical encoding” was developed and implemented for fast imaging of the human brain. The method based on a variant of 3D hybrid EPI combines advantages of the Cartesian and the radial encoding to achieve ultra-short echo time independent of spatial resolution and reasonably short echo train length yielding a quality image of high signal-to-noise ratio. Unlike rectilinear sampling, the method offers not only less motion and flow artifacts but enables also the undersampling capability. As a result, the method improves temporal resolution by shortening the measurement time. Nonetheless, artifacts induced from long-term drifts of the magnetic field as well as geometrical distortions caused by B0 inhomogeneity were removed with the average phase of the k-space center lines and an additional field map scan. Compared to other cylindrical k-space trajectories based on echo-planar imaging, which lead to progressively increasing echo time upon increasing the spatial resolution, the proposed method offers more benefits. As a significant application, imaging readout of the novel technique was applied to true 3D cine imaging which was later used in the combination of pseudo-continuous arterial spin labeling module in order to track a short arterial spin labeling (ASL) bolus of well-defined length along the fast passage through the large vessel compartment of the brain. Parametric maps of ASL signal change, estimated time-to-peak and ASL bolus width were extracted in order to characterize the macrovascular compartments of the brain-feeding arteries. Consequently, bolus dispersion within a single arterial branch was also assessed.

The aim of this thesis is to design and evaluate k-space weighting filters for simultaneously improving the spatial and temporal resolution of cardiovascular MRI, with the ultimate goal of improving the accuracy of quantitative flow measurements, which are important for diagnosis and follow-up of heart dysfunction. Two different k-space filters were implemented and evaluated retrospectively to already acquired data. In addition, evaluation was performed with respect to tapering of the filters in the radial k-space direction, as well as accelerated imaging using undersampling. To better utilize the properties of the golden-angle acquisition, a k-space filter was also implemented where the temporal footprint increased in discrete steps, referred to as rings. The temporal footprint of each ring was calculated according to the Fibonacci sequence, and the starting position for each ring was computed to satisfy the Nyquist criterion. The k-space filters were evaluated in comparison to non-filtered reconstructions of cine and phase-contrast images. Motion-mode images were created from the cine images and used to evaluate the edge sharpness of the septal wall indicating the spatial resolution of the image. Phase-contrast images were used to measure peak flow velocity over the mitral valve, and the myocardial velocity in the early and late filling phases. The resolution of the peak is highly dependent on the temporal resolution. Measuring the peak velocity gave an indication of the temporal resolution, which could be compared to non-filter reconstructions. This study showed that k-space filters adapted to the Nyquist criterion improve the temporal resolution of peak velocity measures. Further investigation is justified to conclude if the performance exceeded the best performing method without k-space filters. However, the k-space filter showed substantial agreement with the best performing temporal footprint without k-space filter. / Syftet med arbetet är att designa och utvärdera k-space viktade filter för att förbättra den spatiala och temporala upplösningen av kardiovaskulär MRI, med målet att förbättra noggrannheten i kvantitativa flödesmätningar, som är viktiga för diagnos och uppföljning av hjärtdysfunktion. Två typer av k-space filter skapades och utvärderades retrospektivt på redan inhämtade data. Dessutom utfördes utvärdering med avseende på avsmalning av filtren i den radiella k-rymdsriktningen, såväl som accelererad avbildning med undersampling. För att bättre utnyttja egenskaperna hos den gyllene vinkeln skapades det ena k-rumsfilter så att det temporala fotavtrycket ökade i diskreta steg, så kallade ringar. Det temporala fotavtrycket för varje ring beräknades enligt Fibonacci talen, och startpositionen för varje ring beräknades så att den uppfyllde Nyquistkriteriet. k-Spacefiltren utvärderades i jämförelse med icke-filtrerade rekonstruktioner av tidsupplösta, anatomiska bilder (cine) och tidsupplösta faskontrastbilder. Bilder i motion-mode skapades från cine-bilderna och användes för att utvärdera kantskärpan av hjärtats skiljevägg (septum), vilket användes som en indikator för bildens spatiala upplösning. Faskontrastbilder användes för att mäta den maximala flödeshastigheten över mitralisklaffen och myokardiets hastighet i den tidiga och sena fyllnadsfasen. Maximal flödeshastighet är starkt beroende av den temporala upplösningen och gav därav en indikation på den temporala upplösningen. Denna studie visade att k-rumsfilter anpassade till Nyquist-kriteriet förbättrar den temporala upplösningen av topphastigheten. Ytterligare undersökning behövs dock för att säkerställa att prestandan översteg den bäst presterande metoden utan k-rumsfilter. Bilder rekonstruerade med filtret visade dock god överensstämmelse med det minsta temporala fotavtrycket, utan filter.

One way of reducing the examination time in magnetic resonance imaging (MRI) is to reduce the amount of raw data acquired, by performing so-called undersampling. Conventionally, MRI data is acquired line-by-line on a Cartesian grid. In the field of Cardiovascular Magnetic Resonance (CMR), however, radial k-space sampling is seen as a promising emerging technique for rapid image acquisitions, mainly due to its robustness against motion disturbances occurring from the beating heart. Whereas Cartesian undersampling will result in image aliasing, radial undersampling will introduce streak artifacts. The objective of this work was to train the deep learning architecture, CycleGAN, to reduce streak artifacts in radially undersampled CMR images, and to evaluate the model performance. A benefit of using CycleGAN over other deep learning techniques for this application is that it can be trained on unpaired data. In this work, CycleGAN network was trained on 3060 radial and 2775 Cartesian unpaired CMR images acquired in human subjects to learn a mapping between the two image domains. The model was evaluated in comparison to images reconstructed using another emerging technique called GRASP. Whereas more investigation is warranted, the results are promising, suggesting that CycleGAN could be a viable method for effective streak-reduction in clinical applications.

O objetivo deste trabalho é o desenvolvimento e implementação de metodologias de imagens por Ressonância Magnética Nuclear, para diminuição do tempo de aquisição, já que nos exames clínicos convencionais esse tempo é muito superior ao utilizado nessas seqüências, que é da ordem de T_ 2 , essas seqüências são baseadas na varredura única do espaço-k, convencionalmente denominada Echo Planar Imaging. Os propósitos de utilização dessa metodologia compreendem desde exames clínicos convencionais, em que se pretende analisar, em projetos futuros, eventos não periódicos de curta duração e a dinâmica dos sistemas biológicos estudados, até imagens de cavidades utilizando gases hiperpolarizados. As técnicas implementadas em comparação com as inicialmente propostas por Masfield apresentam uma diferença que é a inexistência do pulso de RF de inversão e, com isso, o tempo de duração das seqüências implementadas é ainda menor. Apenas não se deve esperar muito da qualidade das imagens sem o pós-processamento, uma vez que esse trabalho já está em andamento. / The objective of this work is the development and implementation of methodologies of images for Nuclear Magnetic Resonance, for reduction of the time of acquisition, since in the conventional clinical examinations this time is very superior to the used one in these sequences, that are of the order of T_ 2 , these sequences is based on the only sweepings of the space-k, conventionally called Echo Planar Imaging. The intentions of use of this methodology understand since conventional clinical examinations, where if it intends to analyze, in future projects, not periodic events of short duration and the dynamics of the biological systems studied, until socket images using hiperpolarizados gases. The techniques implemented in comparison with initially the proposals for Masfield present a difference that is the inexistence of the pulse of RF of inversion and, with this, the time of duration of the implemented sequences are still lesser. But if it does not have to wait very of the quality of the images without the after-processing, a time that this work already is in progress.

O objetivo deste trabalho é o desenvolvimento e implementação de metodologias de imagens por Ressonância Magnética Nuclear, para diminuição do tempo de aquisição, já que nos exames clínicos convencionais esse tempo é muito superior ao utilizado nessas seqüências, que é da ordem de T_ 2 , essas seqüências são baseadas na varredura única do espaço-k, convencionalmente denominada Echo Planar Imaging. Os propósitos de utilização dessa metodologia compreendem desde exames clínicos convencionais, em que se pretende analisar, em projetos futuros, eventos não periódicos de curta duração e a dinâmica dos sistemas biológicos estudados, até imagens de cavidades utilizando gases hiperpolarizados. As técnicas implementadas em comparação com as inicialmente propostas por Masfield apresentam uma diferença que é a inexistência do pulso de RF de inversão e, com isso, o tempo de duração das seqüências implementadas é ainda menor. Apenas não se deve esperar muito da qualidade das imagens sem o pós-processamento, uma vez que esse trabalho já está em andamento. / The objective of this work is the development and implementation of methodologies of images for Nuclear Magnetic Resonance, for reduction of the time of acquisition, since in the conventional clinical examinations this time is very superior to the used one in these sequences, that are of the order of T_ 2 , these sequences is based on the only sweepings of the space-k, conventionally called Echo Planar Imaging. The intentions of use of this methodology understand since conventional clinical examinations, where if it intends to analyze, in future projects, not periodic events of short duration and the dynamics of the biological systems studied, until socket images using hiperpolarizados gases. The techniques implemented in comparison with initially the proposals for Masfield present a difference that is the inexistence of the pulse of RF of inversion and, with this, the time of duration of the implemented sequences are still lesser. But if it does not have to wait very of the quality of the images without the after-processing, a time that this work already is in progress.