Assessment of Image Quality Requirements in Magnetic Resonance Imaging for Quantitative Brain Morphometry

Structural T1-weighted magnetic resonance imaging (MRI) provides sufficient anatomical details to measure and track changes in volumes of brain structures. The volumes of brain structures and changes in them can be used to study the effects of disease, treatment monitoring, aging, learning and brain development. The present thesis investigates the requirements for performing reproducible quantitative brain volume measurements with automated brain tissue segmentation tools and gives an error bound on the measurements under various experimental conditions. A short introduction into the challenges of performing reproducible brain volume measurements and the main issues that impede the adoption of quantitative volumetric measurements in clinical practice is given, followed by an overview of the acquisition, reconstruction and automated image segmentation methods used to perform quantitative brain volume measurements. The first part of this study was carried out to investigate the reproducibility of volumetric measurements preformed on different systems with a standardized ADNI protocol. Systematic biases in volume measurements were observed when there were changes in systems between the first scan and rescan. An important finding in the context of patient management was that neither repositioning nor a two-week gap between the measurements did significantly contribute to the uncertainty in volumetric measurements when compared to the uncertainty in a back-to-back scan-rescan scenario. In the second part of this study, the impact of new highly-accelerated acquisition protocols on automated brain tissue volume measurements was investigated. A single system was used to collect the data and acquisition time was varied at the expense of the SNR. An important outcome of this study was that for qualitative assessment accelerated protocols provided similar information. However, the automated volume measurements with the highly-accelerated protocols were found biased compared to the measurements with standardized ADNI protocol. In the final part of this study, scaling procedures were investigated as means for compensating for the observed differences in sequential automated brain volume measurements. A new image-property-based compensation strategy was proposed and compared to the current state-of-the-art protocol-based approaches. The main outcomes of this study were that there are limitations to the current state-of-the-art protocol-based approaches, namely that volume correction coefficients used in the protocol-based approaches can vary as a function of age, and there is an indication that the proposed image-property-based approach can be more robust to the age and contrast-dependent effects compared to protocol-based approaches.

Toward Reliable MR-Based Brain Structure Morphometry

Bénédicte Marie Maréchal

Dementing neurodegenerative disorders have more and more prominent public health implications with the most frequent cause being the currently non reversible Alzheimer's Disease (AD). However, over the past 15 years, an abundant literature has demonstrated the particular value of structural MRI from which brain morphological changes can be detected using automated volumetry and morphometry. These techniques offer promise to diagnose AD in early potentially-reversible stages as regional atrophy is currently "the" structural hallmark. Nevertheless, an essential question arises: to what extent are these quantitative measures dependent upon image quality? MRI quality can be mainly affected by machine-specific or patient-related artifacts, with motion being the most significant source of degradations. As we are currently not aware of any technique that reliably captures and corrects patient motion particularly for long high-resolution anatomical scans, rigorous quality assessment is hence of great importance to reliably derive qualitative or quantitative diagnostic information and consists in the first challenge addressed in this thesis. We developed a fully-automated method to classify the image quality based on a single MR image. Quality measures are derived by analyzing the air background of magnitude image and are capable of detecting image degradation from several sources, including bulk motion, residual magnetization from incomplete spoiling, blurring and ghosting. The method has been validated on 1670 multi-slice 2D, 3D T1- T2- PD- weighted 1.5T and 3T head scans acquired on Siemens and GE scanners operating with various software and hardware combinations. Results are compared against qualitative grades assigned by the Alzheimer's Disease Neuroimaging Initiative (ADNI) quality control center (taken as the reference standard). The derived quality indices are independent of the MR system used and agree with the reference standard with sensitivity and specificity higher than 90 %. With acceptable computational performance (less than 10 seconds), we incorporated our algorithm in the Siemens scanner software for routine clinical use. As for many other domains, quality criterion does not exhibit all-or-none properties (i.e., a unique quality cutoff value cannot discriminate high- versus low-quality scans) but is modulated by two main aspects addressed in this thesis: acquisition parameters (what kind of image?) and application (what for?). On the one hand, depending on acquisition parameters, more or less strong signal in the object is generated inducing more or less visible artifacts. Therefore, our quality criterion has to be adjusted accordingly. On the other hand, it appears quite natural that different quality criteria are required for algorithms segmenting global gray matter contrary to small brain structure such as hippocampus. Hence comes the second challenge addressed in this thesis: developing models that allow customizing quality criterion according to the acquisition parameters and the required performance of a target application. Taking up the first part of such challenge, we investigated certain acquisition parameters effects (i.e., echo time, bandwidth, signal suppression, field strength) and proposed an exploratory model with reasonable predictive performance (86.6 % sensitivity 92 % specificity). In order to customize quality cutoff levels according to the required performance of a target application, we developed a framework based on motion simulations. The idea is to progressively and realistically corrupt an image with motion and analyze the influence of such degradations (i.e., measured by means of our quality assessment technique) on the accuracy of a brain morphometry algorithm. To be faithful to the challenging research context of early diagnosis of AD, we targeted the automated morphometry of gray matter (GM) tissue (i.e., important imaging-based marker of pathological atrophy). Our framework reveals that 2 % error on the global volume of GM can be induced by a 1 mm patient motion halfway through the acquisition (critical low-frequency sampling of rectilinear k-space). However, this small movement is reliably caught by our quality index whose variation can be further characterized as an exponential function of the error. This model thus allows determining whether an image is of sufficient quality to warrant further quantitative analysis. In conclusion, this thesis underscores the importance of a rigorous quality assessment of MR images to support high-quality outputs of quantitative morphometry tools, ultimately increasing probability for higher diagnostic accuracy. Overall, we envision that such quality assessment could greatly improve clinical workflow through its ability to rule-out the need for a repeat scan while the patient is still in the magnet bore. Just as importantly, it would be an important step forward for incorporating morphometry analysis into regular brain-imaging protocols and thus, further exploit the clinical potential of MRI.

EPFL2011

Fluid and white matter suppression with the MP2RAGE sequence

Giulio Gambarota, Tobias Kober, José Pedro Rebelo Ferreira Marques

Purpose: To develop a magnetic resonance imaging (MRI) sequence (fluid and white matter suppression, FLAWS) for generating two sets of images from a single acquisition: one with contrast similar to a T1-weighted magnetization-prepared rapid gradient-echo sequence (MPRAGE) for structural definition; the other with nulled white matter (WM) signal intensity, similar to the fast gray matter T1 inversion recovery (FGATIR) sequence, for improved delineation of subcortical brain structures. Materials and Methods: The recently proposed MP2RAGE, which is a modification of the MPRAGE and generates two image sets at different inversion times, was employed to generate the FGATIR-like contrast (FLAWS1) and MPRAGE-like contrast (FLAWS2). Five healthy volunteers were scanned at 3T and brain tissue contrast and contrast-to-noise were compared. Results: FLAWS1 and FLAWS2 exhibited similar tissue contrast and contrast-to-noise as the `` reference'' sequences, FGATIR and MPRAGE, respectively. Synthetic minimum value images generated from FLAWS1 and FLAWS2 provided a gray matter-dominant image. Conclusion: FLAWS provides two coregistered 3D volumes, one with nulled WM signal intensity and another with nulled cerebrospinal fluid. The coregistered nature of the two datasets allows for generating images that might be helpful in segmentation algorithms and clinical diagnosis. Key Words: brain; MRI; MPRAGE; brain tissue segmentation; globus pallidus; deep brain stimulation J. Magn. Reson. Imaging 2012; 35: 1063-1070. (C) 2011 Wiley Periodicals, Inc.

Wiley-Blackwell2012

Methods for Molecular Magnetic Resonance Imaging and Magnetic Resonance Spectroscopy using Hyperpolarized Nuclei

Riccardo Balzan

Since the introduction 10 years ago of the dissolution method, Dynamic Nuclear Polarization (DNP) became a widely applied and powerful technique to enhance nuclear magnetic resonance (NMR) signals of low naturally abundant, insensitive nuclear spins for analytical chemistry and biomedical research. The aim of DNP is to obtain a very high degree of polarization on the nuclei of interest in cryogenic conditions (i.e. 1 K ), a hundred or more times higher than the thermal polarization. Then, trough a fast dissolution process the molecules hosting the nuclear spins are mixed in a room temperature solution and, since their polarization is conserved if their relaxation times are long enough, they can be used for in vitro or in vivo experiments with a SNR enhanced by 104 times or more. Although the DNP technique was established about 60 years ago, the hardware and cryogenic equipment needed to perform DNP and the subsequent dissolution, still are a technological challenge. In this thesis we will cover different DNP aspects, ranging from hardware installation, software development, solid state measurement, in vitro and in vivo experiments and a possible application of DNP to MRI studies of granular materials. A new DNP cryostat was developed and tested to determine its cryogenic performances and then optimized for fast cooldown, helium holdtime and overall minimal operational consumption. We will show the characteristics of the cryostat and the performances on standard operations like cooling and DNP at 1 K . Another important aspect the development tackled was automation. The management of the system was condensed in a single electronic box capabile to handle the interface with all the cryostat instruments. This box is driven by a single common USB port trough a custom made software interface we developed. Although a few manual operations are still needed, we achieved a high degree of automation. The solid state polarization enhancement ε is defined as the ratio between the DNP enhanced and the thermal polarization signals. It can be determined by measuring the DNP enhanced signal after the system stabilizes in the polarized state (i.e. after about 5 buildup time) and thermal polarization signal after the spins have fully relaxed to equilibrium (i.e. after about 5T1). As a first application of the new system, we present a method that exploits the behavior of steady states magnetization produced with trains of evenly spaced pulses, at constant flip angle. This method allows to precisely determine the thermally polarized signal of a sample with known T1 and a given flip angle in a small fraction (as low as half of a T1 or less depending on the requested accuracy) of the time needed in the usual "wait 5 T1 and pulse" scheme. The main drawback of the dissolution DNP is that the large polarization achieved will ineluctably relax towards its thermal equilibrium value in a few times T1, that for typical 13C labeled molecules is shorter than a minute. On the other hand some nuclei as 6Li and quaternary 15N show a very long T1 on the order of few minutes in biological conditions, allowing measurements as long as several minutes. 6Li is very sensitive to contrast agents such as Gd-doped contrast agents, O2 and de-oxygenated hemoglobin. We studied the effect of the oxygenation of human and rat blood on the 6Li T1, showing clear changes between oxygenated and de-oxygenated samples. Thus we showed that 6Li could be used to map blood oxygenation levels in vitro, and suggest possible in vivo applications, even in brain due to its capability to easily pass the blood brain barrier. 15N labeled choline has a very long 15N T1 and can be used to probe rate of transport or metabolism in brain. Although it has a long T1, the spectral dispersion between the precursor, choline, and its metabolites is very small. The spectral dispersion is much larger for 1 H spins and thus being able to partially transfer repeatedly the 15N to 1H before detection would greatly increase the spectral separation between metabolites while still being able to measure the time course of the metabolic process, which is not the case with a full transfer like with INEPT. We demonstrate that a Hartmann–Hahn polarization transfer can be used to partially transfer the hyperpolarization of 15N to surrounding 1H before detection on 1H. This method can then be applied to in vivo DNP experiments. Even with a very high signal it is very difficult to quantify metabolites that are very close in the NMR spectrum, compared to the actual in vivo linewidth. To improve the spectral resolution it is worth performing the experiments in high magnetic fields. We show the first in vivo rat muscle DNP measurement of acetate at 14.1 T , to resolve e.g. the glutamate peak from acetate substrate. Finally we discuss a segmentation algorithm used to precisely reconstruct granular samples composed of up to 104 spheres. We used a standard Gradient Echo imaging sequence on a Siemens 7 T MRI human scanner to obtain high resolution images of a phantom composed of a cylinder filled with Cu2+ doped water and plastic spheres. Then the 3D images where segmented trough a thresholded Hugh transform method and the beads centers determined. The datasets were studied to extract structural and geometrical information.