Ultra-high field NMR spectroscopy of GABA in the human brain

Abstract

Magnetic resonance imaging (MRI) is a non-invasive and radiation-free imaging method that has been utilized in medicine for decades. Using magnetic fields, the nuclear spins within the tissue are perturbed, and the resulting magnetization is measured. A strong, homogeneous main field creates a net magnetization within the subject. This magnetization can be manipulated by applying RF pulses on-resonant to the nuclear magnetic resonance frequency. Switchable gradient fields allow to spatially and temporally alter the resonance frequency. Next to high-resolution anatomical scans, a wide variety of physiological processes, like perfusion and diffusion, can be visualized using different measurement sequences. <br /> The time evolution of the magnetization is affected by the molecular surrounding of the nuclear spin. Magnetic resonance spectroscopy (MRS) uses this effect to differentiate the signals of different chemical compounds and infer their concentrations. Although MRS is historically older than MRI, it is still rarely used in clinical practice. This is mainly because of the limited signal strength. MRI uses the signal from the hydrogen nuclei within the water molecules. The concentration of the compounds that are measured using MRS is much lower. Furthermore, elaborate data processing is needed to ensure reliable concentration estimates. <br /> A higher magnetic field strength leads to an increased signal. Additionally, the spectral resolution increased. Therefore, MRS could strongly benefit from the relatively recent introduction of clinical 7 Tesla MRI machines. The higher field strength does not come without challenges. The higher resonance frequency facilitates slice selection, which is needed to obtain a localized signal. Furthermore, the higher RF frequency leads to a more effective absorption within the tissue. For security reasons, the applicable RF power is limited. This must be taken into account when planning measurement sequences. Lastly, the magnetic fields are less homogeneous compared to lower field strengths. <br /> This thesis focuses on measuring the concentration of gamma-amino butric acid (GABA), the dominant inhibitory neurotransmitter in the human brain. Changes in GABA concentration are linked to multiple diseases. Due to its low concentration and the signal overlap of more prominent metabolites, a GABA-specific measurement sequence is needed. <em>J</em>-editing is a method that is based on measuring two slightly different spectra. The resonance shapes of the target compound differ in both spectra, while the resonance shapes of an overlapping compound are identical. Consequently, the signal of the target compound can be isolated by subtracting both spectra. MEGA-sLASER is a high-field sequence that uses this principle. <br /> The main part of this thesis focuses on the implementation, optimization, and validation of a MEGA-sLASER sequence for GABA concentration estimation. The main target region is the hippocampus, a brain region that is severely affected by Alzheimer’s disease. Strong field inhomogeneities are present in this region, hampering accurate concentration estimates. No hippocampal GABA concentration has previously been published. An inter-subject variation, comparable to published values in more accessible brain regions, was found. To achieve this reproducibility, the pulse sequence was optimized with simulated GABA resonances. Furthermore, a dedicated data processing pipeline was implemented. <br /> Additionally, an imaging module is added to the sequence. This allows measuring the spatial distribution of the GABA concentration inside the human brain. Within a single slice, the spatial distribution was imaged with a resolution of 1 mm. Because of the small voxel size and the low GABA concentration, quantification is very difficult. Despite strong noise in the concentration maps, higher GABA concentration in grey matter than in white matter was found. This is in agreement with several published studies.