Dissertations / Theses on the topic 'Echo planar imaging (EPI)'

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.

Functional Magnetic Resonance Imaging (fMRI) based studies are rapidly expanding in the field of preclinical research. The majority of these studies use Echo Planar Imaging (EPI) to measure Blood Oxygenation Level Dependent (BOLD) signal contrasts in the brain. In such studies the magnitude and statistical significances of these contrasts are then related to brain function and cognition. It is assumed that any observed signal contrast is ultimately due to differences in biological state and that scanner performance is stable and repeatable between subjects and studies. However, due to confounding issues introduced by in vivo subjects, little work has been undertaken to test this basic assumption. As the BOLD signal contrasts generated in such experiments are often very low, even small changes in scanner performance may dominate the BOLD contrast, distorting any biological conclusions drawn. A series of fMRI phantoms were produced to measure scanner performance independent of biological subjects. These phantoms produce specified signal contrast levels on demand during an fMRI scan by means of current-induced magnetic field gradients. These were used to generate data sets that emulated the BOLD signal contrast of in vivo imaging. Two studies examining scanner performance were then conducted on high-field preclinical MRI scanners. Firstly, in a longitudinal study on a single scanner, measurements were taken over a number of days across a week long period and then every two months over a year long period. Secondly, the behaviour of four preclinical scanners (three at 7T, one at 9.4T) was comparatively assessed. Measurements of several imaging parameters including contrast generated and functional contrast to noise ratio (fCNR) were obtained in both studies. If the scanners involved are truly comparable then they should generate similar measurement values. Across both studies parameter measurements showed significant differences for identical contrast settings on the phantom. Although signal contrast itself proved very comparable across the studies fCNR proved to be highly variable. As well as these measurements of longer tem behaviour proving variable, short and mid-term signal stability displayed a wide range of variability. Variations in the level and quality of both signal and noise were observed. Modelling of signal changes based on fundamental physical principles was also performed for comparison. The impact of these behaviours and variations on in vivo studies could result in skewed biological conclusions at any single site, with some sites exhibiting greater problems than others. The multisite results suggest potential difficulties when comparing biological conclusions between sites, even when using identical imaging parameters. In summary, these results suggest that a cautious approach should be taken with the conclusions of both fMRI and associated resting state connectivity studies that use EPI as their acquisition sequence. Improvements to both the experimental design of studies and regular quality monitoring of scanners should be undertaken to minimise these effects. Clinical MRI scanners should also be assessed for similar aberrations in behaviour.

Diffusion-weighted (DW) magnetic resonance imaging is an important neuroimaging technique that has successful applications in diagnosis of ischemic stroke and methods based on diffusion tensor imaging (DTI). Tensor measures have been used for detecting changes in tissue microstructure and for non-invasively tracing white matter connections in vivo. The most common image acquistion strategy is to use a DW single-shot echo-planar imaging (ss-EPI) pulse sequence, which is attractive due to its robustness to motion artefacts and high imaging speed. However, this sequence has limited achievable spatial resolution and suffers from geometric distortion and blurring artefacts. Readout-segmented echo-planar imaging (rs-EPI) is a DW sequence that is capable of acquiring high-resolution images by segmenting the acquisition of k- space into multiple shots. The fast, short readouts reduce distortion and blurring and the problem of artefacts due to motion-induced phase changes between shots can be overcome with navigator techniques. The rs-EPI sequence has two main shortcomings. (i) The method is slow to produce image volumes, which is limiting for clinical scans due to patient welfare and prevents us from acquiring very many directions in DTI. (ii) The sequence (like other diffusion techniques) is far from the optimum repetition time (TR) for acquiring data with the highest possible signal-to-noise ratio (SNR) in a given time. The work in this thesis seeks to address both of these important issues using a range of approaches. In Chapter 4 a partial Fourier extension is presented, which addresses point (i) by reducing the number of readout segments acquired and estimating the missing data. This allows reductions in scan time by approximately 40% and the reliability of the images is demonstrated in comparisons with the original images. The application of a simultaneous multi-slice scheme to rs-EPI, to address points (i) and (ii), is described in Chapter 5. Using the slice-accelerated rs-EPI sequence, tractography data were compared to ss-EPI data and high-resolution trace-weighted data were acquired in clinically relevant scan times. Finally, a 3D multi-slab extension that addresses point (i) is presented in Chapter 6. A 3D sequence could also allow higher resolution in the slice direction than 2D multi-slice methods, which are limited by the difficulties in exciting thin, accurate slices. A 3D version of rs-EPI was simulated and implemented and a k-space acquisition synchronised to the cardiac cycle showed substantial improvements in image artefacts compared to a conventional k-space acquisition.

Functional magnetic resonance imaging (fMRI) has provided neuroscientists with a powerful tool to non-invasively study brain function. Typically, fMRI data acquisition is performed using the well-established multi-slice two-dimensional echo planar imaging (2D EPI) technique. While 2D EPI has the considerable advantage of robustness, it is relatively SNR inefficient, particularly at high spatial resolution. Three dimensional (3D) sampling approaches, such as multi-shot 3D EPI provide a theoretical SNR gain compared to 2D EPI and can utilize parallel imaging acceleration along multiple dimensions, leading to the potential for higher spatial and temporal resolution. However, these multi-shot acquisitions span several seconds, making them susceptible to physiological fluctuations. In particular, subject motion is a major source of image degradation. This thesis aims to characterise and improve fMRI acquisition techniques based on 3D EPI approaches. We explored the temporal SNR characteristics of standard segmented 3D EPI for different spatial resolutions and acceleration factors. Specifically, we studied how physiological noise affects the optimal choice of imaging parameters, such as the amount of acceleration. To address some of the shortcomings of conventional 3D EPI, we implemented a hybrid radial-Cartesian 3D EPI trajectory, called TURBINE. This scheme collects EPI "blades" which are rotated about the phase-encoding axis using a golden angle rotation increment, allowing reconstruction at flexible temporal resolution. The self-navigating properties of the sequence are used to determine motion estimates from high temporal resolution navigator images and correct for subject motion as part of the image reconstruction process. We demonstrated that this scheme reduces the impact of motion on fMRI data in the presence of subtle and large subject motions. The techniques developed in this thesis aim to increase the flexibility and robustness of fMRI acquisitions. Ultimately, this research may help increase the utility of fMRI in difficult subjects or patient populations.

Rapid MR imaging has facilitated the development of a variety of medical tools such as MR guided surgeries, drug delivery, stent placement, biopsies, and blood flow imaging. This rapid imaging is largely attributable to the development of parallel imaging techniques. In one such technique, single echo acquisition (SEA) imaging, scan time is reduced by substituting the lengthy phase encoding process with spatial information from an extensive receiver coil array. In order to easily construct and obtain images from this coil array, an ideal set of coil elements would be easily decoupled and tuned, possess high SNR and penetration depth, and would allow for operation in both transmit and receive mode. Several types of coils have been considered for use in massive coil arrays, including the planar pair coil, the loop coil and the stripline coil. This thesis investigates each of these coils for use in massive arrays for SEA imaging with enhanced penetration. This investigation includes: improving the currents on the planar pair coil, determining the feasibility of the loop coil and the stripline coil at the scale required for SEA, and comparing the salient properties of each coil type. This investigation revealed that the stripline coil appears to be the best coil element for SEA imaging with enhanced penetration.

This master’s thesis deals with functional magnetic resonance and monitoring of the effect of acquisition acceleration methods on the quality of functional images and observed BOLD signal. The basic principles of magnetic resonance imaging, the explanation of the specifics of functional magnetic resonance and the formation and scanning of BOLD signal are described here. Subsequently, there is the definition of fMRI experiment and description of sequences used for fMRI, focusing on aquisition acceleration techniques. The influence of sequence parameters on image quality and the data processing methods are explained aftewards. The practical part describes the parameters of used sequences, the acquisition procedure and the task for the subject during aquisition. Data from 26 healthy volunteers were obtained and analyzed afterwards. Based on this, the differencesbetween the different sequence variants were evaluated and the initial assumption that the multi-echo acquisition yields better results with faster measurements than single-echo was confirmed.

This thesis presents a motion compensation method based on orbital navigator echoes that measure the movement of a subject between subsequent acquisitions of a magnetic resonance image (MRI). The 3D orbital navigator method proposed addresses the speed requirements of a fast acquisition technique, namely echo planar imaging (EPI), used in functional MRI. Three orthogonal magnetic field gradient pulses were designed and implemented to produce circular sampling trajectories in the volume of the resonating spins of the subject under study. The variations in phase and magnitude of the subsequent navigator data samples serve as markers for motion occurring between repeated acquisitions. Algorithms to estimate the parameters of motion from the navigator signals were developed and their correctness and stability in the presence of noise were tested by means of numerical simulation. The phantom studies involving controlled rotation and displacement of an inhomogeneous test object during an MRI acquisition established the technique's accuracy. In-vivo experiments showed that the navigator echoes can be used in measuring the gross patient movement, as well as in monitoring the small displacements that may correlate with the respiratory cycle.

Biologists and physicians study complex biologic phenomena in which they use advanced imaging methods. They acquire images containing a lot of information which must be extracted in a correct way. This requires computer skills and knowledge in image processing methods which they seldom have. To overcome the problem, this master thesis aimed to develop a routine for artefact correction and information extraction from images acquired in a research project at the Karolinska Institutet in Stockholm. By developing the routine, the thesis showed how software developed for images of human can be applied to images of rats. The routine handles formatting issues and artefact corrections, calculates diffusion metrics, and performs statistical tests on spatially aligned magnetic resonance images of rats acquired with diffusion weighted echo planar imaging. The routine was verified by analysing the images that it had processed and was considered to create reliable images. Future studies within the field should focus on developing atlases of rats and continue the work with identifying how software developed for images of human can be applied to images of rats.
Biologer och läkare studerar komplexa biologiska processer för vilket de använder avancerade bildgivande metoder. De samlar bilder som innehåller mycket information vilken måste extraheras på ett korrekt sätt. Detta kräver god datorvana och kunskaper inom bildprocessning, vilket de sällan har. För att komma runt problemet, syftade den här masteruppsattsen till att utveckla en rutin för artefaktkorrigering och informationsextrahering från bilder tagna i ett forskningsprojekt vid Karolinska Institutet i Stockholm. Genom att utveckla rutinen, visar uppsattsen hur mjukvaror utvecklade för bilder av människa kan appliceras på bilder av råtta. Rutinen hanterar formatteringsproblem och artefaktkorrigering, beräknar diffusionsmått, och utför statistiska tester på spatiellt matchade magnetresonansavbildningar tagna med diffusionsviktade ekoplana metoder. Rutinen verifierades genom att analysera bilder som den processat och det konstaterades att den skapar korrekta bilder. Framtida studier inom området bör fokusera på att utveckla atlaser av råttor och fortsätta identifieringen av hur mjukvaror utvecklade för bilder av människa kan appliceras på bilder av råtta.

Echo planar imaging (EPI) is widely used for applications where speed and/or sensitivity are essential, such as functional MRI (fMRI) and perfusion MRI. Gradient echo EPI suffers from signal drop-out due to through-slice magnetic field (Bo) inhomogeneity, which prevents certain brain areas from being imaged, e.g. around the frontal sinus and auditory canals. Signal loss can be corrected using the z-shim method, which approximates local 80 variations as linear and recovers signal by applying an opposite refocusing gradient. However, the required gradient is spatially varying and unknown, and. each refocused image requires a separate acquisition. Multiple z-shims are required for full correction making the method inefficient. Signal loss increases with increasing magnetic field strength. The need for an effective correction strategy is driven by the current trend towards high field strengths of 3T and above. This work explores signal loss and recovery in the brain at 3T, and investigates signal loss correction methods culminating in a method which achieves successful signal recovery from two optimally chosen z-shims. The linear approximation made by z-shim methods was found to be valid for the majority of pixels in the brain ,at 3T. Parallel imaging in the through-slice direction to reconstruct points of the k-space slice profile not originally measured was investigated but found to be limited by current hardware. An efficient signal loss correction method was developed which requires only two optimally spaced zshims, positioned by a rapid calibration method for maximum signal recovery. Full signal restoration (to within 2% of the correct value) was achieved in 96% of all brain pixels for 3mm slices, and partial correction in pixels outside this range. This method was applied to a language fMRI study which suffers from signal loss, and recovered activation in regions of 80 inhomogeneity revealing language activation not detectable by conventional fMRI.

Deux méthodes d'imagerie par résonance magnétique sont proposées pour analyser in vivo le tissu cérébral de la matière blanche. La première méthode permet l'acquisition ultra-rapide de cartes des métabolites cérébraux par une lecture de l'espace réciproque répétée à des intervalles de quelques millisecondes à l'aide d'une nouvelle trajectoire excentrée, combinée à un gradient de retour. Une procédure de correction de phase, pour prévenir la formation d'artéfacts de repliement dans l'image et le spectre, est introduite sur la base de paramètres déterminés à partir du signal des protons de l'eau. Une acquisition des cartes métaboliques tridimensionnelles de la créatine, de la choline, du N-acétylaspartate, du glutamate et du myo-inositol ont été déterminées de manière fiable dans la substance blanche humaine à 3 Tesla avec une matrice de taille 32 × 32 × 16 et une résolution isotropique de 7 mm. La deuxième méthode permet l'acquisition d'un train de 32 images échantillonnées géométriquement le long d'une courbe de recroissance, en employant une série d'échos de gradient excités par un angle de bascule de 5° pour éviter des effets de saturation. Après transformée inverse de Laplace utilisant une régularisation spatiale, on obtient une distribution continue des temps de relaxation spin-réseau, T1. Dans la région de T1 entre 100 ms et 230 ms, on distingue un pic attribué à l'eau hydratant les membranes de la myéline. La fraction apparente de cette composante de l'eau de myéline augmente en fonction de l'intensité du champ magnétique, de 8,3 % à 3 Tesla, à 11,3 % à 4 Tesla, pour atteindre 15,0 % à 7 Tesla
Two magnetic resonance imaging methods are proposed for the in vivo investigation of human brain white matter tissue. The first method allows the ultra-fast acquisition of maps of brain metabolites by repeating the sampling of k-space at intervals of a few milliseconds, with a center-out trajectory combined with flyback gradients. A phase-correction procedure is introduced to prevent the formation of aliasing artifacts in the image and in the spectrum, on the basis of parameters determined from the signal of the ubiquitous water protons. An acquisition of threedimensional metabolite maps of creatine, choline, N-acetylaspartate, glutamate, and myo-inositol were determined reliably in human brain white matter at 3 Tesla with a 32 × 32 × 16 matrix and a 7-mm isotropic resolution. The second method enables the acquisition of a train of 32 images geometrically sampled along an inversion-recovery curve, using a series of gradient echoes excited by a low 5° flip angle to avoid saturation effects. After inverse Laplace transform, using a spatial regularization, a continuous distribution of the spin-lattice relaxation times, T1, is obtained. In the region of T1 between 100 ms and 230 ms, a small component is attributed to water hydrating myelin membranes. The apparent fraction of this myelin water component increases with the strength of the magnetic field, from 8.3% at 3 Tesla, to 11.3% at 4 Tesla, and 15.0% at 7 Tesla

博士
國立臺灣大學
電機工程學研究所
93
A technique suitable for diffusion tensor imaging at high field strengths is presented in this thesis. The method is based on PROPELLER k-space trajectory using EPI as the signal readout module, hence dubbed PROPELLER EPI. The implementation of PROPELLER EPI included a series of correction schemes to reduce possible errors associated with the intrinsically higher sensitivity of EPI to off-resonance effects. Experimental results on a 3.0 Tesla MR system showed that the PROPELLER EPI images exhibit substantially reduced geometric distortions compared with single-shot EPI, at a much lower RF specific absorption rate than the original version of PROPELLER fast spin-echo technique. For diffusion tensor imaging, the self-navigated phase correction capability of the PROPELLER EPI sequence was shown to be effective for in vivo imaging. The reduced data acquisition window also allows more signal averages to be performed to achieve higher signal-to-noise ratio when compared with single-shot EPI at identical total scan time, an advantage beneficial for routine diffusion tensor imaging applications in clinical practice.

博士
國立臺灣大學
電機工程學研究所
95
Magnetic Resonance Spectroscopy (MRS) has been used for almost three decades to investigate the living tissue for prognostic or diagnostic markers. Fast magnetic resonance spectroscopy imaging technique, PEPSI, can reduce the long scan time of conventional SI to a single minute, which significantly raise the clinical potential of SI. In this thesis, a PEPSI sequence suitable for general clinical application has been implemented on the clinical machine at various field strengths. We investigate the possibility to combine PEPSI with parallel imaging techniques to further improve the temporal resolution of SI. The experiments showed that the parallel acceleration on PEPSI is feasible with the enrollment of multiple array coils to achieve sub-minute measurement. With the current developed large-N array coils, the temporal resolution of SI can be further reduce. Beneficial from the fast property of PEPSI technique, spatial distributed metabolite T2 relaxation time can be derived in clinical acceptable time. Consistent and reliable results were observed using the proposed protocol based on PEPSI sequence. The acquisition of less than 30 minutes is potentially applicable in clinical practice.

碩士
國立清華大學
原子科學系
93
Recent study had suggested diffusion-weighted (DW) gradient-echo (GE) pulse sequence null the majority of fMRI signal arise from intravascular(IV) at 1.5 T and retains water spins with lower ADC close to tissue water. Moreover, some groups indicate that extravascular (EV) component of large vessels contribute only to GE fMRI, not to spin echo (SE) fMRI, because the 180°RF pulse in SE fMRI can refocus the dephasing effect of the static field inhomogeneity around large vessels. Since blood flow occur earlier in capillaries and small venules near the activation site and gradually extend to larger draining vein, to improve temporal specify and reduce both the IV and EV signal contribution from large vessels distant to neuronal activity, this study apply bipolar diffusion gradients into SE pulse sequence only in slice selection direction at 1.5 T in response to brief visual stimulus and compare onset time variance at the same CNR to that detected in conventional GE and non-diffusion-weighted SE. The results show onset variance is smaller in DW SE than that of GE at the same CNR but no difference between DW SE b value of 50 s mm-2 and DW SE b value of 200 s mm-2. This illustrate DW SE signal is purified and b value 50 s mm-2 is sufficient to suppress contaminated signal from large vessels. Further, onset time of DW SE with b values of 200 s mm-2 precedes that of GE (2.23±0.11 vs. 2.92±0.08, p<0.005) and time to peak (TTP) of DW SE with b values of 200 s mm-2 is also earlier than that of GE (4.62 ±0.07 vs. 4.97 ±0.19, p<0.05). Meanwhile, to take the direction of applied diffusion gradients into consideration, this study also simultaneously incorporated 3 axes bipolar diffusion gradients into SE sequence and make the same comparison as previous did. The results show earlier onset time, time to peak, and smaller onset time variance in DW SE, as previous. This is consistent with what we expected because only in voxels with strong anisotropy of flow does the gradient direction become important.

碩士
國立臺灣科技大學
電子工程系
101
Non-alcoholic fatty liver disease (NAFLD) is a common liver disease in developed country, for preventing it develop to further complications such as fibrosis and cirrhosis, early detection of NAFLD becomes very important. In the past, we use proton echo planar spectroscopic imaging (PEPSI) to acquire the hepatic fat content (HFC) and PEPSI is proved that it is able to acquire the HFC accurately. For matrix size 16x32 the scan time is 18 seconds, and the scan time is short enough to ask the subject to hold his breath. Now, we want to further the matrix size to 24x32 and the scan time can be as short as before. So, we combine PEPSI and parallel imaging technology to shorten scan time. We want to investigate the performance of PEPSI with parallel imaging to see if it has the same accuracy as PEPSI. We set full-sampled PEPSI as the standard, compared with PEPSI with parallel imaging. One of the parallel imaging technology, generalized autocalibrating partially parallel acquisitions (GRAPPA) is enrolled. The results showed that the reconstruction of water signal is more stable than lipid signal. Simulated GRAPPA is more close to full-sampled PEPSI than real GRAPPA on quantifying HFC. In conclusion, 2-fold accelerated PEPSI is implemented successfully but the performance need further improvement.

Tese de mestrado integrado em Engenharia Biomédica e Biofísica, apresentada à Universidade de Lisboa, através da Faculdade de Ciências, 2015
O Developing Human Connectome Project pretende realizar um progresso científico único através da criação do primeiro connectome 4D no início da vida do bebé. De forma a criar um mapa dinâmico da conectividade do cérebro do bebé, é fundamental obter imagens funcionais e com ponderação em difusão. A imagem eco-planar (EPI) é a principal sequência de ressonância magnética aplicada na aquisição dessa informação. Esta sequência permite uma aquisição rápida e repetida de imagens cerebrais permitindo mapear as flutuações da atividade cerebral (imagiologia funcional) bem como ter uma boa resolução nas imagens de difusão (movimento de moléculas de água no volume cerebral). No entanto, esta técnica está associada a artefactos de suscetibilidade. Estes artefactos surgem quando existem interfaces entre duas amostras com suscetibilidades magnéticas diferentes como sejam o tecido biológico e o ar. De forma a minimizar esses artefactos é necessário reduzir as heterogeneidades do campo magnético principal B0 através de shimming. O presente trabalho focou-se em shimming ativo, no qual o campo magnético é mapeado com base num modelo composto por funções harmónicas esféricas e são calculadas as correntes a aplicar às bobinas de shimming. Essas bobinas geram um campo magnético que compensa as heterogeneidades presentes anteriormente. Convencionalmente, as tentativas para superar este problema envolvem a utilização do método disponibilizado no sistema de ressonância magnética, nas quais o campo é mapeado com base em projecções (ex: FASTMAP). Este método é de execução rápida mas apresenta duas desvantagens principais: em primeiro lugar, o utilizador tem um controlo reduzido sobre o processo; em segundo, as regiões nas quais o campo é mapeado não são definidas com base na anatomia de interesse. No contexto deste trabalho, o controlo sobre o processo é importante no sentido de ser possível aplicar exatamente a mesma metodologia a um grupo elevado de sujeitos. Por seu lado, o mapeamento com base na anatomia permite obter uma optimização mais precisa. Com o surgimento de novas tecnologias passou a ser possível fazer um mapeamento mais detalhado do campo magnético com técnicas baseadas em imagem ao invés de projecções. Estas técnicas envolvem a definição de um volume relacionado com a anatomia, e que é incluído na totalidade na optimização do campo. O principal objetivo deste trabalho foi desenvolver uma ferramenta de shimming baseado em imagem a fim de otimizar o campo magnético no contexto de imagens de EPI do cérebro neonatal e fetal. O cérebro do bebé sofre alterações na sua dimensão e forma durante o seu desenvolvimento desde a idade fetal até neonatal. Em cada uma dessas fases o bebé encontra-se cercado por um ambiente diferente que requere uma abordagem específica ao mesmo. Neste sentido, o trabalho desenvolvido foi dividido em três partes principais: descrição da estrutura necessária para a correta aplicação do shimming, shimming neonatal e shimming fetal. Em primeiro lugar, as limitações do shimming baseado em imagem foram estudadas e o algoritmo básico para aplicar o método foi testado. Nesta parte do trabalho foi demonstrado que os campos gerados pelas bobinas de shim presentes no equipamento de ressonância magnética são consistentes com as funções harmónicas esféricas que compõem o modelo aplicado. O efeito do movimento da cama do equipamento sobre a eficiência do shimming foi também estudada. Foi possível corrigir a informação do sistema de coordenadas que descrevem o mapa de campo B0 de forma a incluir o movimento da cama necessário para a obtenção das imagens em sujeitos fetais. A segunda parte do trabalho focou-se no desenvolvimento do shimming para o caso neonatal. Foi desenvolvida uma ferramenta para definição de uma região de interesse, unwrapping da fase e cálculo das correntes de shim. Esta foi desenvolvida em ambiente MATLAB. Nos recém-nascidos o shimming deve ser aplicado numa região de interesse restrita ao cérebro devido à presença da interface ar/tecido no escalpe do bebé. Assim, a definição da região de interesse consistiu principalmente na aplicação de um limiar a fim de binarizar a imagem de magnitude, ajustada pelo utilizador. Em simultâneo foi implementada uma técnica de exclusão dos olhos com base na anatomia dos diferentes planos. No seu conjunto o método apresentou-se flexível de forma a ser ajustado ao sujeito em estudo. Quando aplicado com a mesma metodologia (limiar e exclusão de olhos) o volume incluído foi semelhante entre bebés. O método de shimming foi avaliado com base em três medidas de dispersão do mapa de campo residual: largura a meia altura, desvio padrão dos pixéis no interior da região de interesse e o intervalo de frequências no interior do qual 95 % dos pixéis se encontravam. A utilização de diferentes medidas permitiu a avaliação do m´etodo em relação a diferentes aspetos. Este método foi aplicado a 52 participantes recém-nascidos com idade gestacional média de 39.8 ± 1.7 semanas. O cálculo das correntes de shim permitiu gerar um campo magnético que melhorou a homogeneidade do campo B0 no volume adquirido, sendo consistente com o previsto. Uma imagem média do campo residual foi calculada mostrando que existem duas regiões (occipital e pequenas regiões laterais) nas quais o campo magnético B0 apresenta ainda heterogeneidades. Por fim, os resultados indicam que este método melhorou o campo perto da periferia do cérebro quando comparado com o método convencional disponibilizado no equipamento. O shimming neonatal (shimming ótimo ou OIBS) foi utilizado como alicerce para a implementação de um método ajustado às características das aquisições fetais. Existem duas características principais que devem ser tidas em conta. Em primeiro lugar, os fetos encontram-se envoltos em líquido amniótico e tecido materno pelo que o ambiente no qual estão inseridos permite que a região de interesse seja definida de forma menos restrita. Em segundo lugar, o facto de a cabeça do feto ser pequena pode levar à existência de valores de corrente das bobinas de shim elevados. Essas correntes, principalmente associadas às bobinas de segunda ordem geram campos de magnitude elevada na região do tecido adiposo, o que altera a sua frequência de ressonância. Desta forma, as técnicas de supressão de gordura específicas em frequência são menos efetivas e a imagem de EPI apresenta artefactos. Assim, a ferramenta para shimming fetal incluiu a definição de uma região de interesse cilíndrica e um método de shimming baseado em imagem com limites lineares (shimming limitado ou CIBS) impostos com base na frequência de ressonância do tecido adiposo. O CIBS consistiu na aplicação de limites superiores e inferiores ([-300 100] Hz) para a frequência dos pixéis pertencentes à gordura após a aplicação do shimming. Adicionalmente, o impulso de radiofrequência utilizado para a supressão de gordura foi estudado a fim o otimizar para a sua utilização no contexto do shimming fetal. Para o estudo dos parâmetros do impulso de radiofrequência, os rins de dois voluntários adultos saudáveis foram utilizados como simulação do ambiente fetal, devido as semelhanças entre a localização e interface entre tecidos. Os métodos OIBS e CIBS foram aplicados em 6 grávidas saudáveis com idades gestacional média de 28±6 semanas. Os mapas de campo residuais mostraram que as técnicas eram semelhantes em termos de homogeneidade no interior da região de interesse definida como cérebro, mas a segunda (CIBS) apresentou melhores resultados na supressão de gordura. Como estudo do impulso de radiofrequência foi demonstrado que o desvio do impulso em cerca de 100 Hz no sentido da frequência de ressonância da água melhoraria a supressão de gordura sem detrimento do sinal da água. A utilização do novo método CIBS em simultâneo com um impulso de radiofrequência otimizado mostrou ser a melhor solução para homogeneizar o campo e suprimir a gordura. Em conclusão, as ferramentas apresentadas permitiram melhorar a qualidade das imagens de EPI da cabeça do feto e do recém-nascido no contexto do Developing Human Connectome Project. O shimming neonatal mostrou ser um método consistente que pode facilmente ser utilizado por parte da equipa clínica. A nível fetal foi apresentado um método que demonstra a capacidade de superar as limitações demonstradas pelas técnicas convencionais.
The Developing Human Connectome Project (dHCP) aims to make major scientific progress by creating the first 4-dimensional connectome of early life. Echo planar imaging (EPI) is the main acquisition technique applied in functional and diffusion imaging, which are central to map the human brain. This technique allows fast acquisition of spatial information enabling volumetric coverage of the whole brain, but it is associated with susceptibility artefacts. In order to minimize those artefacts it is necessary to reduce main magnetic field B0 in homogeneities through shimming. Conventionally, the attempts to overcome this problem use the manufacturer’s default method. Unfortunately, with those techniques the user has little control over the process, and the regions within which the field is corrected are not anatomically based. The main objective of this project was to develop an image-based shimming tool to optimize the magnetic field in the context of EPI images adjusted to the neonatal and foetal brains. The babies’ brain suffers changes in dimension and shape during its development from foetal to neonatal age. In each one of those stages the baby is surrounded by a different environment which requires a distinct shimming approach. As a result, the work was divided into three main parts: framework description, neonatal shimming and foetal shimming. First, the limitations of image-based shimming were investigated, and the framework to apply the method was described. It was demonstrated that fields generated by shim coils were consistent with the spherical harmonic model applied. In addition, the coordinate information of the B0 field map was corrected in order to include the table displacement needed for foetal imaging. Second, a tool was developed for neonatal shimming. The tool included region-of-interest (ROI) definition, phase unwrapping and shim calculation. The ROI definition implemented was flexible in order to adjust to each subject under study. When applying this approach while keeping the same threshold/eye exclusion methodology the volume included was similar between babies. The shim calculation allowed to generate shim values that improved homogeneity of the magnetic field within the volume imaged. This method slightly improved the field near the brain’s margins when compared with the manufacturer’s default techniques. Finally, for foetal shimming the groundwork of the neonatal tool was adjusted to this cohort characteristics. The tool for foetal shimming included additional cylindrical ROI definition and constrained image-based shimming. The constrained shimming allowed to account for the mother’s adipose tissue which in the presence of high shim values can lead to imperfect fat suppression. Along with the implementation of shimming tools, the radio frequency pulse used for fat suppression was studied. The new constrained image-based shimming showed similar results in terms of field homogeneity within the fetus’ brain when compared with the optimal image based shimming, with improvement of fat suppression that is enhanced when simultaneously used with the optimized fat suppression radiofrequency pulse.

Das Signal-zu-Rausch-Verhältnis (SNR) stellt bei modernen Bildgebungstechniken in der Magnetresonanz-Tomographie heutzutage oftmals die entscheidende Limitation dar. Eine Verbesserung durch Modifikation der Hardware ist kostspielig und führt meistens zu einer Verstärkung anderer Probleme, wie zum Beispiel erhöhte Energiedeposition ins Gewebe. Im Gegensatz dazu ist Dichtegewichtung eine Methode, die eine SNR-Erhöhung durch Modifikation der Aufnahmetechnik ermöglicht. In der MR-Bildgebung erfolgt oftmals eine retrospektive Filterung des aufgenommenen Signalverlaufs, beispielsweise zur Artefaktreduktion. Damit einhergehend findet eine Veränderung der Modulationstransferfunktion (MTF) bzw. ihrer Fouriertransformierten, der räumlichen Antwortfunktion (SRF), statt. Optimales SNR wird nach dem Matched Filter-Theorem erzielt, wenn die nachträgliche Filterung dem aufgenommenen Signalverlauf proportional ist. Dies steht dem Ziel der Artefaktreduktion entgegen. Bei Dichtegewichtung steht durch nicht-kartesische Abtastung des k-Raums mit der k-Raum-Dichte ein zusätzlicher Freiheitsgrad zur Verfügung. Dieser ermöglicht es, im Falle eines konstanten Signalverlaufs eine gewünschte MTF ohne Filterung zu erreichen. Bei veränderlichem Signalverlauf kann ein SNR Matched Filter angewendet werden, dessen negative Einflüsse auf die MTF durch Dichtegewichtung kompensiert werden. Somit ermöglicht Dichtegewichtung eine vorgegebene MTF und gleichzeitig ein optimales SNR. In der vorliegenden Arbeit wurde Dichtegewichtung erstmals bei den schnellen Multi-Echo-Sequenzen Turbo-Spin-Echo und Echoplanar-Bildgebung (EPI) angewendet. Im Gegensatz zu bisherigen Implementierungen muss hier der Signalabfall durch T2- bzw. T2*-Relaxation berücksichtigt werden. Dies führt dazu, dass eine prospektiv berechnete dichtegewichtete Verteilung nur bei einer Relaxationszeit optimal ist. Bei Geweben mit abweichenden Relaxationszeiten können sich wie auch bei den kartesischen Varianten dieser Sequenzen Änderungen an SRF und SNR ergeben. Bei dichtegewichteter Turbo-Spin-Echo-Bildgebung des Gehirns konnte mit den gewählten Sequenzparametern ein SNR-Vorteil von 43 % gegenüber der kartesischen Variante erzielt werden. Die Akquisition wurde dabei auf die T2-Relaxationszeit von weißer Substanz optimiert. Da die meisten Gewebe im Gehirn eine ähnliche Relaxationszeit aufweisen, blieb der visuelle Gesamteindruck identisch zur kartesischen Bildgebung. Der SNR-Gewinn konnte in der dichtegewichteten Implementierung zur Messzeithalbierung genutzt werden. Dichtegewichtete EPI weist eine hohe Anfälligkeit für geometrische Verzerrungen, welche durch Inhomogenitäten des Hauptmagnetfeldes verursacht werden, auf. Die Verzerrungen konnten erfolgreich mit einer Conjugate Phase-Methode korrigiert werden. Dazu muss die räumliche Verteilung der Feldinhomogenitäten bekannt sein. Dazu ist zusätzlich zur eigentlichen EPI-Aufnahme die zeitaufwendige Aufnahme einer sogenannten Fieldmap erforderlich. Im Rahmen dieser Arbeit konnte eine Methode entwickelt werden, welche die zur Erlangung einer Fieldmap notwendige Aufnahmedauer auf wenige Sekunden reduziert. Bei dieser Art der Fieldmap-Aufnahme müssen jedoch durch Atmung hervorgerufene Effekte auf die Bildphase berücksichtigt werden. Die Fieldmap-Genauigkeit kann durch Aufnahme unter Atempause, Mittelung oder retrospektiver Phasenkorrektur erhöht werden. Für die gewählten EPI-Sequenzparameter wurde mit Dichtegewichtung gegenüber der kartesischen Variante ein SNR-Gewinn von 14 % erzielt. Anhand einer funktionellen MRT (fMRI)-Fingertapping-Studie konnte demonstriert werden, dass die SNR-Steigerung auch zu einer signifikant erhöhten Aktivierungsdetektion in Teilen der Hirnareale führt, die bei der Fingerbewegung involviert sind. Die Verwendung von zusätzlicher EPI-Phasenkorrektur und iterativer Optimierung der dichtegewichteten k-Raum-Abtastung führt zu weiteren Verbesserungen der dichtegewichteten Bildgebung mit Multi-Echo-Sequenzen
Magnetic resonance imaging (MRI) is often limited by the signal to noise ratio (SNR). In standard Cartesian acquisition methods, the SNR can be improved by applying a so-called matched filter to the acquired raw data, which correlates with the anticipated signal profile. Unfortunately, this filter changes the spatial response function (SRF), which characterizes the imaging properties of the imaging method, in an undesired way. For example, a matched filter often amplifies undesired image artifacts and is thus normally omitted. In contrast, filters which change the SRF are typically applied, e.g., for artifact reduction. These however do not provide an optimal SNR. Density weighting is a method which allows a desired SRF and an optimal SNR at the same time. This is achieved by introducing a new degree of freedom to the SRF; the density of the acquisition steps in k-space. In this work, density weighting was adapted to turbo spin echo (TSE) and echo planar imaging (EPI). In contrast to earlier implementations of density weighting, signal relaxation has to be taken into consideration with these multi-echo sequences. As a result, the desired SRF and SNR are only obtained for one prospectively determined relaxation time. For deviating relaxation times, changes in SRF and SNR may occur. In density weighted TSE brain imaging, an average SNR gain of 43 % over Cartesian imaging could be achieved for the chosen sequence parameters. The density weighted acquisition was optimized for the T2 relaxation time of white matter. Since the relaxation times of most other tissues in the brain did not significantly differ, the overall visual impression of density weighted and Cartesian images was identical. The achieved SNR gain could be used to halve the acquisition time of the density weighted implementation. Density weighted EPI is especially prone to geometric distortions caused by inhomogeneities of the main magnetic field. The distortions could be successfully corrected with a conjugate phase method. For these methods, a time-consuming acquisition of a so-called field map is typically required. A method could be developed which greatly reduces the field map acquisition time to a few seconds. It was found that phase changes caused by respiration influence the field map accuracy of this and similar methods. A significantly higher accuracy could be achieved by an acquisition under breath-hold or by retrospective phase correction or averaging. It was demonstrated in an fMRI group study that an average SNR gain of 14 % for density weighted EPI resulted in an increased detection power in the activated brain areas. First results involving additional EPI phase correction and iterative k-space sampling optimization demonstrate further improvements of density weighted imaging with multi-echo sequences

碩士
國立臺灣科技大學
醫學工程研究所
99
The study aims to investigate the feasibility and sensitivity of using MR spectroscopic imaging (MRSI) to observe the hepatic steatosis in normal subjects and in subjects with non-alcoholic fatty liver diseases (NAFLD). To our understanding, measuring hepatic steatosis using MRSI techniques has never been reported before. Measuring hepatic steatosis using MRSI techniques will combine all the advantages of MRI and MRS techniques. The measurement will be noninvasive, without radiation hazard. Most important of all, it can provide precise quantification with regional distribution information. The long scan time of MRSI will be shortened by our fast acquisition sequence, proton echo-planar spectroscopic imaging (PEPSI). Since it can be applied to normal subject studies, hepatic MRSI will become a powerful tool for diagnosis of liver diseases. To achieve this goal, we need to investigate post-processing procedure, and quantification method. From the result of investigation, we develop a reasonable time to breath-hold and scan for clinical applications, and can quickly and successfully detect liver lipid contents.

博士
國立臺灣大學
電機工程學研究所
93
Diffusion-weighted magnetic resonance imaging (DWI) sensitizes the magnetic resonance images to the diffusive mobility of water and maps water diffusion in tissue. It can highlight the microstructural characteristics of biological tissues and serve as a useful imaging tool for both clinical diagnosis and basic medical research. A large number of images with different magnitudes and directions of the diffusion sensitizing gradients need be acquired in order to estimate the diffusion properties. For efficiency, these images are usually acquired using diffusion-weighted echo-planar imaging sequence (DW-EPI). The rapid switching of the gradient pulses of DW-EPI can generate eddy currents in conducting surfaces surrounding the gradient coils. Although generation of eddy currents is greatly decreased in an active shielded gradient system, this can still occur especially when using large and rapidly rising and falling diffusion sensitization gradient pulses. This study describes the application of the feedback linearization neural networks, known from neural network computing, to the problem of gradient preemphasis. This approach of preemphasis adjustment doesn’t require an iterative procedure between measurement and adjustment, therefore is essentially instantaneous in its execution. Based on our study, gradient compensation determined by our procedure effectively suppressed eddy current induced geometric distortion and spatial shift of diffusion-weighted EPI images. Comparing the manual preemphasis adjustments, this approach not only is reliable and accurate but also can reduce the spent time from several hours to several minutes. We have successfully applied this technique to the pig heart fiber tracking with diffusion tensor echo planar imaging (DT-EPI). In the future, the human brain white matter connectivity will be also studied.

碩士
國立臺灣科技大學
電子工程系
97
Nyquist ghost is a well known problem in echo planar images (EPI), which manifests itself as an aliasing ghost displaced at FOV/2 apart from the object along the phase encoding direction and a consequent decrease in the object signal intensity. One common method to correct for the problem is to perform a reference scan without blip phase gradient, with odd and even phases corrected using the reference data. A disadvantage of this method is the temporal deviation between reference scan and image scan, causing phase estimation error in the presence of motion. In this work, we proposed a simple alternative method which corrected the Nyquist ghost without reference scan. The method based on the estimation of phase profile from image itself, and correct the phase in image domain to remove the ghost and recover the signal of object. Combine the proposed method with Homodyne method in half Fourier image to correct Nyquist and blurring artifact, results demonstrate the combination is workable and may provide benefit in further application.

碩士
國立臺灣大學
電機工程學研究所
94
A technique suitable for echo planar imaging (EPI) geometrical distortion correction is presented in this thesis. The method is based on reversed gradient method (RG) using phase-encoding gradient with reversed polarity that produces distortion in opposite direction, from which the information on the difference in spatial displacements is then used to correct for distortion. However, this method is problematic in regions with low signals because of a lack of information on distortions which leads an error correction in regions with low signals. The implementation of reversed gradient method induced a displacement map concept to reduce the error correction in regions with low signals. The displacement map concept consists of eliminating error displacement map, surface fitting and filtering. Experimental results on a 3.0 Tesla MR system showed that the modified reversed gradient method corrected images exhibit substantially reduced geometric distortions and correction error in regions with low signals compared with original reversed gradient method. Applied reversed gradient method in high-resolution DTI using PROPELLER EPI to extra correct each blade, shows benefits from the modified reversed gradient method. It can increase the image quality for routine diffusion tensor imaging applications in clinical practice.

碩士
國立臺灣科技大學
電子工程系
100
Cingulate cortex (CC) is involved in many pathological conditions including psychiatric disorders and chronic pain. As the major excitatory neurotransmitter of the brain, glutamate (Glu) plays an important role not only in the pathology of these conditions but also as a mechanism for drug intervention. Glutamine (Gln) is the precursor of both Glu and Gamma-Amino Butyric acid (GABA), the major inhibitory neurotransmitter of the brain. Quantification of Glu and Gln in CC may provide important information about pathological mechanisms and drug dynamics. We used Proton Echo Planar Spectroscopic Imaging (PEPSI), a fast magnetic resonance spectroscopic imaging (MRSI) technique, sequence with short-TE (TE30) and averaged TE (TEavg) protocols to detect Glu and Gln in CC. Although a better fitting was obtained for Glu with TEavg, there was no obvious difference in coefficient of variance (COV) of Glu between two protocols. In conclusion, PEPSI is suitable for assessment of short-term and long-term changes in brain metabolites. Compared to TEavg, TE30 protocol had similar performance on Glu but provides more accurate quantification of Gln and other metabolites.

碩士
國立中山大學
資訊工程學系研究所
100
Proton echo planar spectroscopic imaging(PEPSI) is a novel and rapid technique of magnetic resonance spectroscopic imaging(MRSI). To analyze the metabolite in PEPSI by using LCModel, an automatic reconstruction system is necessary. Recently, many researches use graphic processing unit(GPU) to accelerate imaging reconstruction, and Compute Unified Device Architecture(CUDA) is developed by C language, so the programmers can write the program in parallel computing easily. PEPSI data acquisition includes non water suppression and water suppression scans, each scan contains odd and even echoes, these two data are reconstructed separately. The image reconstruction contains k-space filter, time-domain filter, three-dimension fast Fourier transform(FFT), phase correction and combine odd and even data. We use MATLAB, C++ and parallel computing to implement PEPSI reconstruction, and parallel computing applied CUDA which proposed by NVIDIA. In our study, the averaged non water suppression spectroscopic imaging executed by three different programming language are almost the same. In our data scale, the execution time of parallel computing is faster than MATLAB and C++, especially in the FFT step. Therefore, we simulated and compared the performance of one- to three-dimension FFT. Our result shows that accelerating performance of GPU depends on the number of data points according to the performance of FFT and the execution time of single coil PEPSI reconstruction. While the amount of data points is larger than 65536, as demonstrated in our study, parallel computing contribute in terms of computational acceleration.

碩士
長庚大學
電機工程學系
100
Magnetic Resonance Spectroscopy (MRS) and Magnetic Resonance Spectroscopic Imaging (MRSI) are good tools to measure the metabolites of our brain, and both of them are non-invasive. T2 relaxation times of cerebral metabolites are important for the estimation of quantification of metabolite concentrations. Here we propose to acquire 3-dimensional distribution of three cerebral metabolites N-acetyl aspartate (NAA), creatine (Cre), and choline (Cho), at 3T using Proton Echo Planar Spectroscopy imaging (PEPSI) with the total acquisition time less than 30 minutes. After regular post processing including baseline correction and phase correction for the PEPSI data and calculated the T2 values by using linear regression, our results show consistent T2 values between subject and regional difference of NAA between WM and GM which in agreement with previous studies. The estimation of T2 values is stable according to the high Pearson’s correlation coefficients between logarithmic MR signals and TE. In summary, PEPSI technique is a robust method to obtain fast maps of metabolite T2 values.