Search

In magnetic resonance imaging, a birdcage coil is the most commonly used volumetric resonator creating a highly homogeneous radiofrequency magnetic field in a conductive sample. An artificial magnetic radiofrequency shield was recently shown to improve the magnetic field amplitude per unit power (transmit efficiency) of a preclinical birdcage coil by reducing the intrinsic losses in the coil and increasing power absorbed by the sample. In this paper, we propose a new application of an artificial shield in clinical MRI. Thanks to the proposed artificial shield a birdcage coil for human brain imaging operating at 300 MHz (Larmor frequency of protons at static fields of 7 T) can be expanded to increase free space. As a result, the coil becomes more comfortable for the patient and keeping comparable transmit efficiency. The same extended coil with a conventional copper shield would have at least 10% lower efficiency. The proposed artificial shield is implemented as an annular-ring cavity-backed slot in a copper cylinder that tightly surrounds the birdcage. To demonstrate the effect, radiofrequency magnetic field and specific absorption rate distributions were compared numerically and experimentally for the initial and extended coils with different shields.

The main goal of this methodological thesis was to present optimised magnetic resonance imaging (MRI) sequences for ex vivo and in vivo studies in ultra-high field scanners (UHF, from 7 T and above, beyond the conventional 1.5 or even 3 T available in the clinics). The researchers mainly worked with the STEAM sequence, which overcame several of its restrictions in these scanners. It resulted in ultra-high resolution whole-brain MRI data for ex vivo studies (achieving 400 μm , or 0.4 mm, isotropic resolution), and very highly accelerated acquisition protocols for in vivo studies (up to x 54 times the current total possible acceleration). With the current feasibility of acquiring ultra-high resolution data and highly sampled multi-contrast MRI images, data-expensive signal models can be employed; but above all, the biological microstructure features in the brain can be easily more revealed from the resulted analysis.

Das Ziel dieser Arbeit ist es, die klinische Relevanz von Hoch- und Ultrahochfeld Magnetresonanztomografie Tomographen in der Radiologietechnologie zu erforschen. Heutzutage sind Tomographen mit einer Magnetfeldstärke von 3 Tesla in vielen Institutionen längst Standard. In einigen Krankenhäusern kommen überdies bereits Tomographen mit einer Feldstärke von 7 Tesla zum klinischen Einsatz. Weiters existieren Tomographen mit mehr als 10 Tesla, welche derzeit allerdings nur experimentellen Zwecken dienen. Der Trend der Erhöhung der Magnetfeldstärke in der Magnetresonanztomografie ist deutlich zu erkennen. Dadurch ergeben sich neue Möglichkeiten und für manche Untersuchungen eine drastische Verbesserung der Qualität. Jedoch ist auch mit negativen Aspekten, erhöhte Risikofaktoren sowie Auswirkungen auf Patient*Innen und Personal zu rechnen. Um dies zu beantworten, wurden im Zuge dieser Bachelorarbeit diverse Studien und Literatur zu diesem Thema analysiert und ausgearbeitet. Dabei ergab sich, dass durch die Erhöhung der Magnetfeldstärke, Untersuchungen mit einer geringeren Sensitivität wie zum Beispiel die Spektroskopie, aufgrund des höheren SNR, verbessert wurden. Weiters ermöglichen höhere Magnetfeldstärken eine insgesamt bessere Neurodiagnostik. Allerdings sind die Nachteile, speziell die Implantatsicherheit und die erhebliche Erhöhung der SAR nicht zu vernachlässigen. Diese stellen zusammen mit der erhöhten Häufigkeit der Nebenwirkungen die größten Nachteile der Hoch- beziehungsweise Ultrahochfeld Magnetresonanztomografie dar. The aim of this thesis is to investigate the clinical relevance of high and ultra-high field magnetic resonance imaging tomographs in radiology technology. Tomographs with a magnetic field strength of 3 Tesla are already standard in many institutions. In some hospitals, tomographs with a field strength of 7 Tesla are already in clinical use. Furthermore, tomographs with more than 10 Tesla exist, which, however, currently only serve experimental purposes. The trend of increasing magnetic field strengths in magnetic resonance imaging is clearly visible. This opens new possibilities and for some examinations a drastic improvement in quality. However, negative aspects, increased risk factors as well as effects on the patients and personnel are also to be expected. To answer existing questions, various studies and literature on this topic were analyzed and elaborated during this bachelor thesis. It was found that by increasing the magnetic field strength, examinations with lower sensitivity, such as spectroscopy, were drastically improved due to the higher SNR. Furthermore, higher magnetic field strengths enable a better neurodiagnostic overall. However, there are drawbacks, specifically the implant safety and the significant increase of the SAR. These, together with the increased frequency of side effects, represent the greatest disadvantages of high or ultra-high field magnetic resonance imaging.

Background Ultra-high field magnetic resonance imaging (MR) may be used to improve intracranial blood flow measurements. However, standard cardiac synchronization methods tend to fail at ultra-high field MR. Therefore, this study aims to investigate an alternative synchronization technique using Doppler ultrasound. Methods Healthy subjects (n = 9) were examined with 7T MR. Flow was measured in the M1-branch of the middle cerebral artery (MCA) and in the cerebral aqueduct (CA) using through-plane phase contrast (2D flow). Flow in the circle of Willis was measured with three-dimensional, three-directional phase contrast (4D flow). Scans were gated with Doppler ultrasound (DUS) and electrocardiogram (ECG), and pulse oximetry data (POX) was collected simultaneously. False negative and false positive trigger events were counted for ECG, DUS and POX, and quantitative flow measures were compared. Results There were fewer false positive triggers for DUS compared to ECG (5.3 ± 11 vs. 25 ± 31, p = 0.031), while no other measured parameters differed significantly. Net blood flow in M1 was similar between DUS and ECG for 2D flow (1.5 ± 0.39 vs. 1.6 ± 0.41, bias ± 1.96SD: − 0.021 ± 0.36) and 4D flow (1.8 ± 0.48 vs. 9 ± 0.59, bias ± 1.96SD: − 0.086 ± 0.57 ml). Net CSF flow per heart beat in the CA was also similar for DUS and ECG (3.6 ± 2.1 vs. 3.0 ± 5.8, bias ± 1.96SD: 0.61 ± 13.6 μl). Conclusion Gating with DUS produced fewer false trigger events than using ECG, with similar quantitative flow values. DUS gating is a promising technique for cardiac synchronization at 7T.

Purpose Moving to ultra‐high fields (≥7 T), the inhomogeneity of both RF (B 1) and static (B 0) magnetic fields increases, which further motivates us to design a realistic head‐shaped phantom, especially for spectroscopic imaging. Such phantoms provide images similar to the human brain and serve as a reliable tool for developing and examining methods in MRI. This study aims to develop and characterize a realistic head‐shaped phantom filled with brain‐mimicking metabolites for MRS and magnetic resonance spectroscopic imaging in a 7 T MRI scanner. Methods A 3D head‐shaped container with three sections—mimicking brain, muscle and precranial lipid—was constructed. The phantom was designed to provide robustness to heating, mechanical damage and leakage, with easy refilling. The head’s shape and the agarose mixture were optimized to provide B 0 and B 1 distributions and T 1/T 2 relaxation values similar to those of human brain. Eight brain‐tissue‐mimicking metabolites were included for spectroscopy. The phantom was evaluated for localized spectroscopy, fast spectroscopic imaging and fat suppression. Results The B 0 and B 1 maps showed distribution similar to that of human brain, with increased B 0 inhomogeneity near the nasal and ear areas and reduced B 1 in the temporal lobe and brain stem regions, as expected in vivo. The metabolites’ concentrations were verified by single‐voxel spectroscopy, showing an average deviation of 11%. Fast spectroscopic imaging and imaging with fat suppression were demonstrated. Conclusion A 3D head‐shaped phantom for human brain imaging and spectroscopic imaging in 7 T MRI was demonstrated, making it a realistic phantom for methodology development at 7 T., A realistic head‐shaped phantom with brain‐mimicking metabolites was developed and characterized for magnetic resonance spectroscopy and spectroscopic imaging in a 7T MRI scanner. The phantom was designed to improve robustness to heating, mechanical damage and leakage, with handy refilling. The B0 and B1 maps showed similar to human brain distribution. The metabolites’ concentrations were verified by single voxel spectroscopy, showing an average deviation of 11% and fast spectroscopic imaging was demonstrated.

Background The diagnosis of Parkinson’s Disease (PD) remains a challenge and is currently based on the assessment of clinical symptoms. PD is also a heterogeneous disease with great variability in symptoms, disease course, and response to therapy. There is a general need for a better understanding of this heterogeneity and the interlinked long-term changes in brain function and structure in PD. Over the past years there is increasing interest in the value of new paradigms in Magnetic Resonance Imaging (MRI) and the potential of ultra-high field strength imaging in the diagnostic work-up of PD. With this multimodal 7 T MRI study, our objectives are: 1) To identify distinctive MRI characteristics in PD patients and to create a diagnostic tool based on these differences. 2) To correlate MRI characteristics to clinical phenotype, genetics and progression of symptoms. 3) To detect future imaging biomarkers for disease progression that could be valuable for the evaluation of new therapies. Methods The TRACK-PD study is a longitudinal observational study in a cohort of 130 recently diagnosed (≤ 3 years after diagnosis) PD patients and 60 age-matched healthy controls (HC). A 7 T MRI of the brain will be performed at baseline and repeated after 2 and 4 years. Complete assessment of motor, cognitive, neuropsychiatric and autonomic symptoms will be performed at baseline and follow-up visits with wearable sensors, validated questionnaires and rating scales. At baseline a blood DNA sample will also be collected. Discussion This is the first longitudinal, observational, 7 T MRI study in PD patients. With this study, an important contribution can be made to the improvement of the current diagnostic process in PD. Moreover, this study will be able to provide valuable information related to the different clinical phenotypes of PD and their correlating MRI characteristics. The long-term aim of this study is to better understand PD and develop new biomarkers for disease progression which may help new therapy development. Eventually, this may lead to predictive models for individual PD patients and towards personalized medicine in the future. Trial registration Dutch Trial Register, NL7558. Registered March 11, 2019.

Background To date, little is known about the presence and extent of cerebellar cortical pathology in early stages of MS. Objective The aims of this study were to (i) investigate microstructural changes in the normal-appearing cerebellar cortex of early MS patients by using 7 T MRI and (ii) evaluate the influence of those changes on clinical performance. Methods Eighteen RRMS patients and nine healthy controls underwent quantitative T1 and T2* measurement at 7 T MRI using high-resolution MP2RAGE and multi-echo gradient-echo imaging. After subtracting lesion masks, average T1 and T2* maps were computed for three layers in the cerebellar cortex and compared between groups using mixed effects models. Results The volume of the cerebellar cortex and its layers did not differ between patients and controls. In MS patients, significantly longer T1 values were observed in all vermis cortical layers and in the middle and external cortical layer of the cerebellar hemispheres. No between-group differences in T2* values were found. T1 values correlated with EDSS, SDMT and PASAT. Conclusions We found MRI evidence of damage in the normal-appearing cerebellar cortex at early MS stages and before volumetric changes. This microstructural alteration appears to be related to EDSS and cognitive performance.

Background and purpose: In neuro-oncology, high spatial accuracy is needed for clinically acceptable high-precision radiation treatment planning (RTP). In this study, the clinical applicability of anatomically optimised 7-Tesla (7T) MR images for reliable RTP is assessed with respect to standard clinical imaging modalities. Materials and methods: System- and phantom-related geometrical distortion (GD) were quantified on clinically-relevant MR sequences at 7T and 3T, and on CT images using a dedicated anthropomorphic head phantom incorporating a 3D grid-structure, creating 436 points-of-interest. Global GD was assessed by mean absolute deviation (MADGlobal). Local GD relative to the magnetic isocentre was assessed by MADLocal. Using 3D displacement vectors of individual points-of-interest, GD maps were created. For clinically acceptable radiotherapy, 7T images need to meet the criteria for accurate dose delivery (GD

Background In postmortem analysis of late stage Parkinson's disease (PD) neuronal loss in the substantial nigra (SN) correlates with the antemortem severity of bradykinesia and rigidity, but not tremor. Objective To investigate the relationship between midbrain nuclei volume as an in vivo biomarker for surviving neurons in mild-to-moderate patients using 7.0 Tesla MRI. Methods We performed ultra-high resolution quantitative susceptibility mapping (QSM) on the midbrain in 32 PD participants with less than 10 years duration and 8 healthy controls. Following blinded manual segmentation, the individual volumes of the SN, subthalamic nucleus, and red nucleus were measured. We then determined the associations between the midbrain nuclei and clinical metrics (age, disease duration, MDS-UPDRS motor score, and subscores for bradykinesia/rigidity, tremor, and postural instability/gait difficulty). Results We found that smaller SN correlated with longer disease duration (r = -0.49, p = 0.004), more severe MDS-UPDRS motor score (r = -0.42, p = 0.016), and more severe bradykinesia-rigidity subscore (r = -0.47, p = 0.007), but not tremor or postural instability/gait difficulty subscores. In a hemi-body analysis, bradykinesia-rigidity severity only correlated with SN contralateral to the less-affected hemi-body, and not contralateral to the more-affected hemi-body, possibly reflecting the greatest change in dopamine neuron loss early in disease. Multivariate generalized estimating equation model confirmed that bradykinesia-rigidity severity, age, and disease duration, but not tremor severity, predicted SN volume. Conclusions In mild-to-moderate PD, SN volume relates to motor manifestations in a motor domain-specific and laterality-dependent manner. Non-invasive in vivo 7.0 Tesla QSM may serve as a biomarker in longitudinal studies of SN atrophy and in studies of people at risk for developing PD.

Background Although 7T functional MRI (fMRI) provides better signal‐to‐noise ratio and higher spatial resolution than 3T fMRI, geometric distortions become more challenging because fMRI is more susceptible to distortions than structural MRI. Accurate alignment of 7T fMRI to structural MRI data is critical for precise cortical surface‐based analysis. Purpose To quantify the effectiveness of distortion corrections of 7T fMRI data. Study Type Prospective. Subjects Fifteen healthy individuals aged 19–26 years (mean: 21.9 years). Field Strength/Sequence Multiband gradient‐echo echo‐planar imaging sequence at 7T; 3D T1/T2‐weighted sequences (magnetization prepared rapid acquisition with gradient echo [MPRAGE] and sampling perfection with application optimized contrast using different flip angle evolution [SPACE]) at 3T. Assessment fMRI data at 7T were registered to cortical surfaces reconstructed from 3T structural data acquired in the same subjects. Distortions induced by B0 inhomogeneity and gradient nonlinearity (B0 and gradient distortions) were evaluated as cortical fallout (misregistration of noncortical areas) and displacement (misregistration along gray matter). Statistical Tests Repeated measures analyses of variance with post‐hoc t‐tests with Bonferroni correction. Results The accuracy of fully corrected fMRI images based on the intensity distribution was 89.2%. Without any corrections, 9.7% of vertices in the whole surfaces were fallout and the average displacement was 0.96 mm for the rest of the vertices. B0 and gradient distortion corrections significantly reduced the fallout (to 2.1% and 8.7%) and displacement (to 0.29 mm and 0.86 mm). These corrections were effective even around regions with moderate distortions (the somatosensory and visual cortices for B0 distortion, and the anterior frontal, inferior temporal, and posterior occipital cortices for gradient distortion). Data Conclusion B0 distortion correction is crucial for surface‐based analysis of fine‐resolution fMRI at 7T. Gradient distortion correction should be considered when regions of interest include regions distant from the isocenter of scanners. Evidence Level 1 Technical Efficacy Stage 1

Diffusion MRI (dMRI) in ex vivo human brain specimens is an important research tool for neuroanatomical investigations and the validation of dMRI techniques. Many ex vivo dMRI applications have benefited from very high dMRI resolutions achievable on small-bore preclinical or animal MRI scanners for small tissue samples. However, the investigation of entire human brains post mortem provides the important context of entire white matter (WM) network systems and entire gray matter (GM) areas connected through these systems. The investigation of intact ex vivo human brains in large bore systems creates challenges due to the limited gradient performance and transmit radio-frequency (B1+) inhomogeneities, specially at ultra-high field (UHF, 7T and higher). To overcome these issues, it is necessary to tailor ex vivo diffusion-weighted sequences specifically for high resolution and high diffusion-weighting. Here, we present kT-dSTEAM, which achieves B1+ homogenization across whole human brain specimens using parallel transmit (pTx) on a 9.4T MR system. We use kT-dSTEAM to obtain multi-shell high b-value and high resolution diffusion-weighted data in ex vivo whole human brains. Isotropic whole brain data can be acquired at high b-value (6000-8000 s/mm2) at high resolution (1000 μm) and at moderate b-value (3000 s/mm2) at ultra-high isotropic resolution (400 μm). As an illustration of the advantages of the ultra-high resolution, tractography across the WM/GM border shows less of the unwanted gyral crown bias, and more high-curvature paths connecting the sulcal wall than at lower resolution. The kT-dSTEAM also allows for acquisition of T1 and T2 weighted images suitable for estimating quantitative T1 and T2 maps. Finally, multi-shell analysis of kT-dSTEAM data at variable mixing time (TM) is shown as an approach for ex vivo data analysis which is adapted to the strengths of STEAM diffusion-weighting. Here, we use this gain for multi-orientation modelling and crossing-fiber tractography. We show that multi-shell data allows superior multiple orientation tractography of known crossing fiber structures in the brain stem.

Modern high field and ultra high field magnetic resonance imaging (MRI) experiments routinely collect multi-dimensional data with high spatial resolution, whether multi-parametric structural, diffusion or functional MRI. While diffusion and functional imaging have benefited from recent advances in multi-dimensional signal analysis and denoising, structural MRI has remained untouched. In this work, we propose a denoising technique for multi-parametric quantitative MRI, combining a highly popular denoising method from diffusion imaging, over-complete local PCA, with a reconstruction of the complex-valued MR signal in order to define stable estimates of the noise in the decomposition. With this approach, we show signal to noise ratio (SNR) improvements in high resolution MRI without compromising the spatial accuracy or generating spurious perceptual boundaries.

During natural behavior, much of the motion signal falling into our eyes is due to our own movements. Therefore, in order to correctly perceive motion in our environment, it is important to parse visual motion signals into those caused by self-motion such as eye- or head-movements and those caused by external motion. Neural mechanisms underlying this task, which are also required to allow for a stable perception of the world during pursuit eye movements, are not fully understood. Both, perceptual stability as well as perception of real-world (i.e. objective) motion are the product of integration between motion signals on the retina and efference copies of eye movements. The central aim of this thesis is to examine whether different levels of cortical depth or distinct columnar structures of visual motion regions are differentially involved in disentangling signals related to self-motion, objective, or object motion. Based on previous studies reporting segregated populations of voxels in high level visual areas such as V3A, V6, and MST responding predominantly to either retinal or extra- retinal (‘real’) motion, we speculated such voxels to reside within laminar or columnar functional units. We used ultra-high field (9.4T) fMRI along with an experimental paradigm that independently manipulated retinal and extra-retinal motion signals (smooth pursuit) while controlling for effects of eye-movements, to investigate whether processing of real world motion in human V5/MT, putative MST (pMST), and V1 is associated to differential laminar signal intensities. We also examined motion integration across cortical depths in human motion areas V3A and V6 that have strong objective motion responses. We found a unique, condition specific laminar profile in human area V6, showing reduced mid-layer responses for retinal motion only, suggestive of an inhibitory retinal contribution to motion integration in mid layers or alternatively an excitatory contribution in deep and superficial layers. We also found evidence indicating that in V5/MT and pMST, processing related to retinal, objective, and pursuit motion are either integrated or colocalized at the scale of our resolution. In contrast, in V1, independent functional processes seem to be driving the response to retinal and objective motion on the one hand, and to pursuit signals on the other. The lack of differential signals across depth in these regions suggests either that a columnar rather than laminar segregation governs these functions in these areas, or that the methods used were unable to detect differential neural laminar processing. Furthermore, the thesis provides a thorough analysis of the relevant technical modalities used for data acquisition and data analysis at ultra-high field in the context of laminar fMRI. Relying on our technical implementations we were able to conduct two high-resolution fMRI experiments that helped us to further investigate the laminar organization of self-induced and externally induced motion cues in human high-level visual areas and to form speculations about the site and the mechanisms of their integration.

Background Seven T ultra-high field MRI systems have recently been approved for clinical use by the U.S. and European regulatory agencies. These systems are now being used clinically and will likely be more widely available in the near future. One of the applications of 7 T systems is musculoskeletal disease and particularly peripheral arthritis imaging. Since the introduction of potent anti-rheumatic therapies over the last two decades MRI has gained increasing importance particularly for assessment of disease activity in early stages of several rheumatic disorders. Commonly gadolinium-based contrast agents are used for assessment of synovitis. Due to potential side-effects of gadolinium non-enhanced techniques are desirable that enable visualization of inflammatory disease manifestations. The feasibility of 7 T MRI for evaluation of peripheral arthritis has not been shown up to now. Aim of our study was to evaluate the feasibility of contrast-enhanced (CE) and non-enhanced MRI at 7 T for the assessment of knee joint synovitis. Method Seven T MRI was acquired for 10 patients with an established diagnosis of psoriatic or rheumatoid arthritis. The study pulse sequence protocol was comprised of a sagittal intermediate-weighted fat-suppressed (FS), axial fluid-attenuated inversion recovery (FLAIR) FS, sagittal 3D T1-weighted dynamic contrast enhanced (DCE) and an axial static 2D T1-weighted FS contrast-enhanced sequence (T1-FS CE). Ordinal scoring on non-enhanced (Hoffa- and effusion-synovitis) and enhanced MRI (11-point synovitis score), and comparison of FLAIR-FS with static T1-FS CE MRI using semiquantitative (SQ) grading and volume assessment was performed. For inter- and intra-reader reliability assessment weighted kappa statistics for ordinal scores and intraclass correlation coefficients (ICC) for continuous variables were used. Results The total length of study protocol was 15 min 38 s. Different amounts of synovitis were observed in all patients (mild: n = 3; moderate: n = 5; severe: n = 2). Consistently, SQ assessment yielded significantly lower peripatellar summed synovitis scores for the FLAIR-FS sequence compared to the CE T1-FS sequence (p

Item does not contain fulltext OBJECTIVES: The use of 7 Tesla (T) magnetic resonance imaging (MRI) has recently shown great potential for high-resolution soft-tissue neuroimaging and visualization of microvascularization in glioblastoma (GBM). We have designed a clinical trial to explore the value of 7 T MRI in radiation treatment of GBM. For this aim we performed a preparatory study to investigate the technical feasibility of incorporating 7 T MR images into the neurosurgical navigation and radiotherapy treatment planning (RTP) systems via qualitative and quantitative assessment of the image quality. MATERIALS AND METHODS: The MR images were acquired with a Siemens Magnetom 7 T whole-body scanner and a Nova Medical 32-channel head coil. The 7 T MRI pulse sequences included magnetization-prepared two rapid acquisition gradient echoes (MP2RAGE), T2-SPACE, SPACE-FLAIR and gradient echo sequences (GRE). A pilot study with three healthy volunteers and an anthropomorphic 3D phantom was used to assess image quality and geometrical image accuracy. RESULTS: The MRI scans were well tolerated by the volunteers. Susceptibility artefacts were observed in both the cortex and subcortical white matter at close proximity to air-tissue interfaces. Regional loss of signal and contrast could be minimized by the use of dielectric pads. Image transfer and processing did not degrade image quality. The system-related spatial uncertainty of geometrical distortion-corrected MP2RAGE pulse sequences was

Researchers in the field of MRI (Magnetic resonance imaging) are pushing towards the use of stronger and stronger magnets in MRI scanners. These extremely strong magnets are used because the signal-to-noise ratio in MRI scans increases with increasing magnetic field strength. A higher signal-to-noise ratio implies that MRI scans can be acquired at higher resolution. Standard MRI machines operating at a field strength of 3 “Tesla” are able to resolve structures with details up to 1mm, while less common 7 Tesla MRI machines can go to 0.5mm. A few research centers around the world are building MRI scanners with magnets stronger than 7 Tesla, with the aim of imaging the finely detailed structures of the brain and the body. In addition to a very strong static magnetic field, all MRI scanners use radio-frequency (RF) antennas to transmit and receive the signals that generate MRI scans. For MRI scanner operating at ultra-high field strength, the design of these radio-frequency antennas becomes vital to the image quality. If the antennas are not designed or operated properly, the image quality deteriorates. In this thesis, several improvements have been made to RF antennas for body imaging at 7 Tesla and the world’s first 10.5 Tesla. In addition, critical safety aspects of these RF antennas have been studied. The new RF antenna designs are applied to image prostate cancer and the heart at unprecedented resolutions. Finally, the potential gains of MRI scanners operating at even stronger fields (14 Tesla) are demonstrated through a simulation study.

Researchers in the field of MRI (Magnetic resonance imaging) are pushing towards the use of stronger and stronger magnets in MRI scanners. These extremely strong magnets are used because the signal-to-noise ratio in MRI scans increases with increasing magnetic field strength. A higher signal-to-noise ratio implies that MRI scans can be acquired at higher resolution. Standard MRI machines operating at a field strength of 3 “Tesla” are able to resolve structures with details up to 1mm, while less common 7 Tesla MRI machines can go to 0.5mm. A few research centers around the world are building MRI scanners with magnets stronger than 7 Tesla, with the aim of imaging the finely detailed structures of the brain and the body. In addition to a very strong static magnetic field, all MRI scanners use radio-frequency (RF) antennas to transmit and receive the signals that generate MRI scans. For MRI scanner operating at ultra-high field strength, the design of these radio-frequency antennas becomes vital to the image quality. If the antennas are not designed or operated properly, the image quality deteriorates. In this thesis, several improvements have been made to RF antennas for body imaging at 7 Tesla and the world’s first 10.5 Tesla. In addition, critical safety aspects of these RF antennas have been studied. The new RF antenna designs are applied to image prostate cancer and the heart at unprecedented resolutions. Finally, the potential gains of MRI scanners operating at even stronger fields (14 Tesla) are demonstrated through a simulation study.

High spatial resolution functional MRI (fMRI) and advanced multivariate analysis techniques are promising tools for studying the cortical basis of human cognitive processes at the level of columns and layers. However the true spatial specificity of high-resolution fMRI has not been quantified, and the basis for decoding from fine scale structures using large voxels and relatively low magnetic field strength is unknown. It is also not yet known what method and voxel size is optimal for decoding and what voxel size is optimal for high-resolution imaging. In this thesis we present four studies that answer part of these questions using a model-based approach of imaging cortical columns. We started our investigation of model-based analysis of high-resolution fMRI of cortical columns by addressing the specific problem of how it is possible to decode information thought to be mediated by cortical columns using large voxels at low field strength. Multivariate machine learning algorithms applied to human functional MRI (fMRI) data can decode information conveyed by cortical columns, despite the voxel-size being large relative to the width of columns. Several mechanisms have been proposed to underlie decoding of stimulus orientation or the stimulated eye. These include: (I) aliasing of high spatial-frequency components, including the main frequency component of the columnar organization, (II) contributions from local irregularities in the columnar organization, (III) contributions from large-scale non-columnar organizations, (IV) functionally selective veins with biased draining regions, and (V) complex spatio-temporal filtering of neuronal activity by fMRI voxels. Here we sought to assess the plausibility of two of the suggested mechanisms: (I) aliasing and (II) local irregularities, using a naive model of BOLD as blurring and MRI voxel sampling. To this end, we formulated a mathematical model that encompasses both the processes of imaging ocular dominance (OD) columns and the subsequent linear classification analysis. Through numerical simulations of the model, we evaluated the distribution of functional differential contrasts that can be expected when considering the pattern of cortical columns, the hemodynamic point spread function, the voxel size, and the noise. We found that with data acquisition parameters used at 3 Tesla, sub-voxel supra-Nyquist frequencies, including frequencies near the main frequency of the OD organization (0.5 cycles per mm), cannot contribute to the differential contrast. The differential functional contrast of local origin is dominated by low-amplitude contributions from low frequencies, associated with irregularities of the cortical pattern. Realizations of the model with parameters that reflected a best-case scenario and the reported BOLD point-spread at 3 Tesla (3.5 mm) predicted decoding performances lower than those that have been previously obtained at this magnetic field strength. We conclude that low frequency components that underlie local irregularities in the columnar organization are likely to play a role in decoding. We further expect that fMRI-based decoding relies, in part, on signal contributions from large-scale, non-columnar functional organizations, and from complex spatio-temporal filtering of neuronal activity by fMRI voxels, involving biased venous responses. Our model can potentially be used for evaluating and optimizing data-acquisition parameters for decoding information conveyed by cortical columns. Having developed a model of imaging ODCs we then used this model to estimate the spatial specificity of BOLD fMRI, specifically at high field (7 T). Previous attempts at characterizing the spatial specificity of the blood oxygenation level dependent functional MRI (BOLD fMRI) response by estimating its point-spread function (PSF) have conventionally relied on spatial representations of visual stimuli in area V1. Consequently, their estimates were confounded by the width and scatter of receptive fields of V1 neurons. Here, we circumvent these limits by instead using the inherent cortical spatial organization of ocular dominance columns (ODCs) to determine the PSF for both Gradient Echo (GE) and Spin Echo (SE) BOLD imaging at 7 Tesla. By applying Markov Chain Monte Carlo sampling on a probabilistic generative model of imaging ODCs, we quantified the PSFs that best predict the spatial structure and magnitude of differential ODCs’ responses. Prior distributions for the ODC model parameters were determined by analyzing published data of cytochrome oxidase patterns from post-mortem histology of human V1 and of neurophysiological ocular dominance indices. The most probable PSF full-widths at half-maximum were 0.82 mm (SE) and 1.02 mm (GE). Our results provide a quantitative basis for the spatial specificity of BOLD fMRI at ultra-high fields, which can be used for planning and interpretation of high-resolution differential fMRI of fine-scale cortical organizations. Our BOLD fMRI PSF findings show that the PSF is considerably smaller than what was reported previously. This in turn raised the question of the role of the imaging PSF, which now has become relevant. Next, we show that the commonly used magnitude point-spread function fails to accurately represent the true effects of k-space sampling and signal decay, and propose an alternative model that accounts more accurately for these effects. The effects of k-space sampling and signal decay on the effective spatial resolution of MRI and functional MRI (fMRI) are commonly assessed by means of the magnitude point-spread function (PSF), defined as the absolute values (magnitudes) of the complex MR imaging PSF. It is commonly assumed that this magnitude PSF signifies blurring, which can be quantified by its full-width at half-maximum (FWHM). Here we show that the magnitude PSF fails to accurately represent the true effects of k-space sampling and signal decay. Firstly, a substantial part of the width of the magnitude PSF is due to MRI sampling per se. This part is independent of any signal decay and its effect depends on the spatial frequency composition of the imaged object. Therefore, it cannot always be expected to introduce blurring. Secondly, MRI reconstruction is typically followed by taking the absolute values (magnitude image) of the reconstructed complex image. This introduces a non-linear stage into the process of image formation. The complex imaging PSF does not fully describe this process, since it does not reflect the stage of taking the magnitude image. Its corresponding magnitude PSF fails to correctly describe this process, since convolving the original pattern with the magnitude PSF is different from the true process of taking the absolute following a convolution with the complex imaging PSF. Lastly, signal decay can have not only a blurring, but also a high-pass filtering effect. This cannot be reflected by the strictly positive width of the magnitude PSF. As an alternative, we propose to model the imaging process by decomposing it into a signal decay-independent MR sampling part and an approximation of the signal decay effect. We approximate the latter as a convolution with a Gaussian PSF or, if the effect is that of high-pass filtering, as reversing the effect of a convolution with a Gaussian PSF. We show that for typical high-resolution fMRI at 7 Tesla, signal decay in Spin-Echo has a moderate blurring effect (FWHM = 0.89 voxels, corresponds to 0.44 mm for 0.5 mm wide voxels). In contrast, Gradient-Echo acts as a moderate high-pass filter that can be interpreted as reversing a Gaussian blurring with FWHM = 0.59 voxels (0.30 mm for 0.5 mm wide voxels). Our improved approximations and findings hold not only for Gradient-Echo and Spin-Echo fMRI but also for GRASE and VASO fMRI. Our findings support the correct planning, interpretation, and modeling of high-resolution fMRI. In our first study we used our model to analyze imaging of cortical columns under a very specific scenario. We studied a best case scenario for decoding the stimulated eye from ODCs imaged at 3T using large voxels. In order to do so, we formalized available knowledge about fMRI of cortical columns. In particular, the ability of fMRI to resolve cortical columnar organization depends on several interdependent factors, e.g. the spatial scale of the columnar pattern, the point-spread of the BOLD response, voxel size and the signal-to-noise ratio. In our fourth study we aim to analyze how these factors contribute and combine in imaging of arbitrary cortical columnar patterns at varying field strengths and voxel sizes. In addition, we compared different pattern imaging approaches. We show how detection, decoding and reconstruction of a fine scale organization depend on the parameters of the model, and we predict optimal voxel sizes for each approach under various scenario. The capacity of fMRI to resolve cortical columnar organizations depends on several factors, e.g. the spatial scale of the columnar pattern, the point-spread of the fMRI response, the voxel size, and the SNR considering thermal and physiological noise. How these factors combine, and what is the voxel size that optimizes fMRI of cortical columns remain unknown. Here we combine current knowledge into a quantitative model of fMRI of patterns of cortical columns. We compare different approaches for imaging patterns of cortical columns, including univariate and multivariate based detection, multi-voxel pattern analysis (MVPA) based decoding, and reconstruction of the pattern of cortical columns. We present the dependence of their performance on the parameters of the imaged pattern and the data acquisition, and predict voxel sizes that optimize fMRI under various scenarios. To this end, we modeled differential imaging of realistic patterns of cortical columns with different spatial scales and degrees of irregularity. We quantified the capacity to detect and decode stimulus-specific responses by analyzing the distribution of voxel-wise differential responses relative to noise. We quantified the accuracy with which the spatial pattern of cortical columns can be reconstructed as the correlation between the underlying columnar pattern and the imaged pattern. For regular patterns, optimal voxel widths for detection, decoding and reconstruction were close to half the main cycle length of the organization. Optimal voxel widths for irregular patterns were less dependent on the main cycle length, and differed between univariate detection, multivariate detection and decoding, and reconstruction. We compared the effects of different factors of Gradient Echo fMRI at 3 Tesla (T), Gradient Echo fMRI at 7T and Spin-Echo fMRI at 7T, and found that for all measures (detection, decoding, and reconstruction), the width of the fMRI point-spread has the most significant effect. In contrast, different response amplitudes and noise characteristics played a comparatively minor role. We recommend specific voxel widths for optimal univariate detection, for multivariate detection and decoding, and for reconstruction under these three data-acquisition scenarios. Our study supports the planning, optimization, and interpretation of fMRI of cortical columns and the decoding of information conveyed by these columns.

The purpose of this work was to develop and investigate a radiofrequency (RF) coil to perform image studies on small animals using the 7T magnetic resonance imaging (MRI) system, installed in the imaging platform in the autopsy room (Portuguese acronym PISA), at the University of Sao Paulo, Brazil, which is the unique 7T MRI scanner installed in South America. Due to a high demand to create new specific coils for this 7T system, it is necessary to carefully assess the distribution of electromagnetic (EM) fields generated by the coils and evaluate the patient/object safety during MRI procedures. To achieve this goal 3D numerical methods were used to design and analyse a 8-rungs transmit/receive linearly driven birdcage coil for small animals. Calculated magnetic field (B1) distributions generated by the coil were crosschecked with measured results, indicating good confidence in the simulated results. Electric field results were post-processed and predictions of local specific absorption rate (SAR) values were achieved for a spherical phantom filled with muscle-like tissue, indicating that the sample would not suffer any unsafe deposition of energy. Post mortem abdomen images obtained from a rat presented good image quality and no artifacts related to field non-homogeneity were observed.

The aim of ultra-high field strength (≥7T) and ultra-strong gradient systems (≥300mT/m) is to go beyond the millimeter resolution imposed at lower field and to reach the mesoscopic scale in neuroimaging. This scale is essential to understand the link between brain structure and function. However, despite recent technological improvements of clinical UHF-MRI, gradient systems remain too limited to reach this resolution. This thesis aims at answering the need for mapping the human brain at a mesoscopic scale by the study of post mortem samples. An alternative approach has been developed, based on the use of preclinical systems equipped with ultra-high fields (7T/11.7T) and strong gradients (780mT). After its extraction and fixation at Bretonneau University Hospital (Tours), an entire human brain specimen was scanned on a 3T clinical system, before separating its two hemispheres and cutting each hemisphere into seven blocks that could fit into the small bore of an 11.7T preclinical system. An MRI acquisition protocol targeting a mesoscopic resolution was then set up at 11.7T. This protocol, including anatomical, quantitative, and diffusion-weighted sequences, was validated through the study of two key structures: the hippocampus and the brainstem. From the high resolution anatomical and diffusion dataset of the human hippocampus, it was possible to segment the hippocampal subfields, to extract the polysynaptic pathway, and to observe a positive gradient of connectivity and neuritic density in the posterior-anterior direction of the hippocampal formation. The use of advanced microstructural models (NODDI) also highlighted the potential of these techniques to reveal the laminar structure of the Ammon’s horn. A high resolution anatomical and diffusion MRI dataset was obtained from the human brainstem with an enhanced resolution of a hundred micrometers. The segmentation of 53 of its 71 nuclei was performed at the Bretonneau University Hospital, making it the most complete MR-based segmentation of the human brainstem to date. Major white matter bundles were reconstructed, as well as projections of the locus coeruleus, a structure known to be impaired in Parkinson’s disease. Buoyed by these results, a dedicated acquisition campaign targeting the entire left hemisphere was launched for total scan duration of 10 months. The acquisition protocol was performed at 11.7T and included high resolution anatomical sequences (100/150μm) as well as 3D diffusion-weighted sequences (b=1500/4500/8000 s/mm², 25/60/90 directions, 200μm). In addition, T1-weighted inversion recovery turbo spin echo scans were performed at 7T to further investigate the myeloarchitecture of the cortical ribbon at 300µm, revealing its laminar structure. A new method to automatically segment the cortical layers was developed relying on a Gaussian mixture model integrating both T1-based myeloarchitectural information and diffusion-based cytoarchitectural information. The results gave evidence that the combination of these two contrasts highlighted the layers of the visual cortex, the myeloarchitectural information favoring the extraction of the outer layers and the neuritic density favoring the extraction of the deeper layers. Finally, the analysis of the MRI dataset acquired at 11.7T on the seven blocks required the development of a preprocessing pipeline to correct artifacts and to reconstruct the entire hemisphere using advanced registration methods. The aim was to obtain an ultra-high spatio-angular resolution MRI dataset of the left hemisphere, in order to establish a new mesoscopic post mortem MRI atlas of the human brain, including key information about its structure, connectivity and microstructure., L’ambition des très hauts champs magnétiques (≥ 7T) à forts gradients (≥ 300mT/m) est de dépasser la résolution millimétrique imposée à plus bas champ pour atteindre l’échelle mésoscopique en neuroimagerie. Etudier le cerveau à cette échelle est essentiel pour comprendre le lien entre fonction et substrat anatomique. Malgré les progrès réalisés sur les aimants cliniques à 7T, il n’en est pas de même des gradients. Cette thèse vise à cartographier le cerveau humain à l’échelle mésoscopique via l’étude de pièces anatomiques post mortem. Une approche alternative a été choisie, reposant sur l'utilisation d'imageurs précliniques à très hauts champs (7T et 11.7T) et forts gradients (780mT/m). Après une première étape de préparation (extraction et fixation) opérée au CHU de Tours, une pièce anatomique complète a été scannée à 3T, avant découpe de l’hémisphère gauche en sept blocs. Un protocole d’acquisition IRM ciblant une résolution mésoscopique a ensuite été mis en place à 11.7T. Ce protocole, incluant des séquences anatomiques, relaxométriques, et de diffusion, a été validé à l’aide de deux structures clé: un hippocampe et un tronc cérébral. Les données anatomiques et de diffusion acquises à une résolution mésoscopique sur l’hippocampe ont permis de segmenter ses sous-champs, d’extraire le circuit polysynaptique et d’observer l’existence d’un gradient de connectivité et de densité neuritique positif dans la direction postéro-antérieure de l’hippocampe. L’utilisation de modèles avancés d’étude de la microstructure a également révélé l’apport de ces techniques pour la segmentation de l’hippocampe, les cartes de densité neuritique révélant les trois couches des champs ammoniens. Un tronc cérébral a ensuite été scanné, avec une résolution atteignant la centaine de micromètres. Une segmentation de 53 de ses 71 noyaux a été réalisée au sein du CHU de Tours, permettant d’établir la cartographie IRM du tronc cérébral humain la plus complète à ce jour. Les principaux faisceaux de la substance blanche ont été reconstruits, ainsi que les projections du locus coeruleus, structure connue pour être atteinte dans le maladie de Parkinson. Forts de ces résultats, la campagne d'acquisition de l'hémisphère gauche, d’une durée de 10 mois, a été initiée. Le protocole d’acquisition à 11.7T intègre des séquences anatomiques (100/150µm) ainsi que des séquences d'imagerie 3D pondérées en diffusion (b=1500/4500/8000 s/mm², 25/60/90 directions) à 200µm. Des acquisitions complémentaires réalisées à 7T comprenant des séquence d’écho de spin rapide avec inversion-récupération ont par ailleurs permis d’étudier la myéloarchitecture du cortex cérébral et d’identifier automatiquement sa structure laminaire. Un nouveau modèle de mélange de Gaussiennes a été développé, intégrant les informations myéloarchitecturales issues de la cartographie T1 et les informations cytoarchitecturales issues de l’imagerie de diffusion. Il a ainsi pu être démontré que l’utilisation conjointe de ces deux informations permettait de mettre en évidence des couches du cortex visuel, l’information myéloarchitecturale favorisant l’extraction des couches externes et la densité neuritique celle des couches plus profondes. Enfin, l’exploitation des données IRM acquises à 11.7T sur les différents blocs a nécessité la mise en place d’une chaîne de prétraitements pour corriger les artéfacts d’imagerie et reconstruire l’hémisphère entier à l’aide de stratégies de recalage difféomorphe avancées. L’objectif de ce projet est l’obtention d’un jeu de données IRM de très haute résolution spatio-angulaire de l’hémisphère gauche. Ce jeu de données anatomique et de diffusion unique permettra à terme de constituer un nouvel atlas IRM mésoscopique de la structure, de la connectivité et de la cytoarchitecture du cerveau humain.

International audience; PurposeTo accelerate cardiac cine at 7 tesla using simultaneous multi‐slice (SMS) acquisition with self‐calibration to resolve misalignment between calibration and imaging data due to breathing motion.MethodsA spoiled‐gradient echo cine sequence was modified with radiofrequency phase‐cycled SMS excitations. A Fourier encoding strategy was applied along the cardiac phase dimension to allow for slice untangling and split‐slice GRAPPA calibration. Split‐slice GRAPPA was coupled with regular GRAPPA (SMS‐GRAPPA) and L1‐SPIRiT (SMS‐L1SPIRiT) for image reconstruction. 3‐slice SMS cine MRI was evaluated in ten subjects against single‐slice cine MRI in terms of SNR and contrast‐to‐noise ratio and slice leakage.ResultsSNR decreased significantly from 10.1 ± 7.1 for single‐slice cine to 7.4 ± 2.8 for SMS‐GRAPPA (P = 0.02) and was recovered to 9.0 ± 4.5 with SMS‐L1SPIRiT (P = 0.02). Contrast to noise ratio decreased significantly from 14.5 ± 8.1 for single‐slice cine to 5.6 ± 3.6 for SMS‐GRAPPA (P < 0.0001) and increased slightly but significantly back to 6.7 ± 4.4 for SMS‐L1SPIRiT (P = 0.03). Specific absorption rate restrictions imposed a reduced nominal flip angle (−37 ± 7%, P = 0.02) for 3‐slice SMS excitations compared to single‐slice acquisitions. SMS slice leakage increased significantly from apex (8.6 ± 6.5 %) to base (13.1 ± 4.1 %, P = 0.03) in the left ventricle.ConclusionThree‐fold acceleration of cine at 7T was achieved using the proposed SMS technique. Fourier encoding self‐calibration and regularized image reconstruction enabled simultaneous acquisition of three slices without significant SNR decrease but significant CNR decrease linked to the reduced nominal excitation flip angle.