Development of novel magnetic resonance imaging pulse sequences for clinical applications

Abstract

Diffusion is a type of translational molecular motion playing a central role in a large number of physical, chemical and biological phenomena, the latter encompassing in vitro as well as in vivo processes. Diffusion studies with nuclear magnetic resonance (NMR) have long proven to be a robust and versatile method in both fundamental and applied research. Despite providing extraordinary measurement precision for most single-component homogeneous liquid samples, the two basic measurement techniques in diffusion NMR imaging and spectroscopy, i.e., pulsed gradient spin-echo and pulsed gradient stimulated echo (PGSE and PGSTE) have a number of drawbacks. These often result in either an increased number of experiments, a longer single experiment duration or additional post-processing. Modifying the PGSE and PGSTE pulse sequences or their method of execution is known to help overcome some of the complications. This work addresses two reasons classic diffusion NMR experiments often prove ineffective. The first issue is responsible for longer durations of single experiments and results from the consistency between experiments being provided by starting the measurement from an equilibrium state of magnetization "" typically thermodynamic equilibrium. On the other hand, consistency can be achieved by using a different state of magnetization, characterized by dynamic equilibrium, i.e., the steady state. A steady state is known to form if the experiment is repeated rapidly enough, yet the diffusion manifestation in the steady state has not yet been well described even though it is the default execution mode in most diffusion NMR imaging studies. In order to provide such a description, the Bloch equations were solved for a PGSE experiment employing two radiofrequency pulses and an arbitrary magnetic field gradient waveform. The theoretical description was tested for the case of isotropic diffusion in a polyethylene glycol/water solution and for the anisotropic case using a lyotropic liquid crystal sample. Computer simulations were also performed for the isotropic diffusion case. The results of isotropic diffusion measurements differed from the reference values when steady state effects were not accounted for. The theoretical description and computer simulations developed on the other hand showed close agreement with the control measurements. In contrast, anisotropic diffusion measurements showed no difference between the two approaches, which can be attributed to the relaxation properties of the sample. The criteria for distinguishing between the cases of classic and steady state descriptions of NMR diffusion measurements were proposed based on these results. Another reason for complications in diffusion NMR imaging experiments appears when there are multiple diffusing components (usually two, although more can be present) in the studied system or sample. In such a case conventional data analysis which provides a single diffusion coefficient (or diffusion tensor if the diffusion is anisotropic) yields non-meaningful results. A number of methods exist to separate the components or to suppress one in order to acquire reliable data for the other. The most frequently used separation methods lead to an increase in the total experiment time (due to an increased number of data points that needs to be acquired) while the suppression effects can dissipate over time due to the sysytem returning to equilibrium state. A number of solutions were developed to improve both the suppression and the separation techniques. Separation experiments were performed by employing a low acquisition bandwidth, which results in a chemical shift based separation of the components in the image domain. Suppression experiments were performed with alternating polarity gradient pulses and composite radiofrequency inversion pulses. A theoretical method was developed for fast analytical computation of the b-matrices for bipolar gradient pulse modules and other diffusion-weighting gradient pulse modules. The results of the method application to bipolar gradient pulses were implemented as a part of Bruker ParaVision method structure. The latter simplified the setup of the supression experiment as well as the acquired data post-processing. Both the suppression and the separation methods showed the ability to successfully resolve multicomponent diffusion with precision and accuracy equivalent to or surpassing the conventional techniques.

Date of Award	2015
Original language	English