Help Privacy Policy Disclaimer
  Advanced SearchBrowse





Magnetic resonance spectroscopy: quantitative analysis of brain metabolites and macromolecules


Borbath,  T
Research Group MR Spectroscopy and Ultra-High Field Methodology, Max Planck Institute for Biological Cybernetics, Max Planck Society;
Max Planck Institute for Biological Cybernetics, Max Planck Society;

External Resource
Fulltext (public)
There are no public fulltexts stored in PuRe
Supplementary Material (public)
There is no public supplementary material available

Borbath, T. (2021). Magnetic resonance spectroscopy: quantitative analysis of brain metabolites and macromolecules. PhD Thesis, Eberhard-Karls-Universität Tübingen, Tübingen, Germany.

Cite as: http://hdl.handle.net/21.11116/0000-0009-AA77-2
Proton magnetic resonance spectroscopy (1H-MRS) allows non-invasive quantification of the human brain's metabolism in vivo. 1H-MRS measures the interaction of the 1H- hydrogen isotope with oscillating electromagnetic fields in the presence of a strong electromagnetic field. The measured MRS signal of the 1H-hydrogen atoms reflects the concentration of the metabolites present in the tissue. Metabolites are small molecules reflecting the metabolism. Each 1H-hydrogen atom present in a metabolite has a specific resonance frequency, which depends on the chemical structure of the metabolite. The ensemble of the resonance frequencies of all metabolites present in the measured tissue creates the MRS signal. The MRS signal is Fourier transformed, producing an MRS spectrum, where each resonance frequency appears as a distinct peak. The most abundant molecule in the human tissue is water. The resonance frequency of water is suppressed in 1H-MRS to permit the quantification of other metabolites, which are present with significantly lower concentrations. In the MRS spectrum, protons with lower resonance frequencies than water form the upfield spectrum, whereas protons with higher resonance frequencies form the downfield spectrum. This work focused on the modelling of the MRS spectrum. The first part is focused on the accurate determination of metabolite concentrations. The upfield spectrum contains most brain metabolites of clinical interest. However, there is a severe spectral overlap between the metabolite resonances, and therefore dedicated software calculates the contributions of individual metabolites. The modelling of the individual metabolite contributions to the measured spectrum is referred to as spectral fitting. Through this spectral fitting, the metabolite concentrations needed for clinical diagnostics are determined. The most significant overlap in MRS spectra originates from the signals underlying the metabolite resonances, referred to as the macromolecular spectrum. The macromolecular spectrum contains the resonance frequencies of protons in proteins and peptides, which have a slightly faster signal decay than the smaller molecules (metabolites). Other contributors to the spectral overlap are residuals of the not entirely suppressed water signal or lipid signals originating from outside the volume of interest. A spline baseline is typically used in the fitting software to model these contributors. This work firstly investigated the impacts of different macromolecular spectra and spline baselines used in spectral fitting. Significant effects in the quantified metabolite concentrations were noticed, when the spline baseline flexibility was altered in the community “gold standard” software, LCModel. Therefore, the newly developed fitting algorithm proposed in this work, ProFit-v3, incorporates an automatic adaptive baseline flexibility determination. The ProFit-v3 software was then systematically evaluated to different perturbations and baseline effects. The quantified concentrations were compared to the ground truth (when known) and the LCModel software results. The second part of this work focuses on the modelling of the less investigated regions of the MRS spectrum. The downfield spectrum contains many resonance peaks unassigned to metabolite contributions. In this work, downfield spectral peaks were used to quantify intracellular pH. Additionally, for all downfield peaks T2 relaxation times, peak linewidths, and concentrations were calculated. Lastly, based on the quantified peak properties combined with previous literature measurements, the contributing molecules to the downfield peaks were assigned. The macromolecular spectrum was attributed by previous literature to contributions of amino acids in proteins and peptides, based on in vitro measurement of dialyzed cytosol. Moreover, the resonance frequencies of protein amino acids have been extensively collected into a protein database by the NMR community. Hence, this work proposes a modelling approach to quantify the in vivo measured macromolecular spectrum to individual amino acids. In conclusion, the investigation results and the proposed fitting software ProFit-v3 from this work should lead to improved quantification of 1H-MRS spectra. Lastly, the peak assignments in the downfield spectra and the proposed amino acid model promises possible future biomarkers for disease.