Quantitative Gradient-Echo Imaging with Macroscopic B0 Field Variations in the Brain

Martin Söllradl

Publikation: StudienabschlussarbeitDissertation

Abstract

Gradient-echo based magnetic resonance imaging (MRI) sequences are widely employed for T1-weighted morphological and functional imaging. When acquiring multiple gradient-echo images with different echo times, the T2* decay allows insight into the tissue microstructure. Examples of different investigations include the T2* anisotropy in white matter nerve fibers, the determination of the water compartment in the myelin sheaths, or the visualization of abnormally high iron concentrations in deep gray matter.
Despite the improvement of MRI systems in the essential components, such as the main magnetic field, the shim and the gradient systems, macroscopic inhomogeneities of the magnetic field remain a major source of errors in the quantification of R2* (=1/T2*) relaxation rates. In 2D slice-selective measurement techniques, the signal dephasing is particularly pronounced in slice-direction because the slice thickness is usually much larger than the in-plane dimensions; consequently, the signal dephasing is strongly influenced by the excitation profile. All in all, this makes the exact quantification of tissue-specific parameters considerably more difficult.
To minimize the influence of these macroscopic field inhomogeneities, a signal model is presented, which allows for the description of macroscopic field inhomogeneities on the 2D multi-echo gradient-echo (mGRE) signal for arbitrary radiofrequency excitation pulses. The longer repetition time in 2D mGRE measurements with an interleaved slice acquisition than in 3D measurements is particularly suitable to reduce the influence of longitudinal relaxation. This is especially important in multi-compartmental analyses of the signal decay, such as the determination of the myelin water fraction (MWF).
To benefit from an increased signal-to-noise ratio (SNR) at optimized flip angles, the model uses a numerical solver for the Bloch equations. Its advantage is that it is not limited to small flip angles compared with the analytical solution. It has been shown that applying the model leads to less influence of macroscopic field gradients on R2* and MWF values in comparison with signal models that do not account for macroscopic field variations.
In a second approach, an adaptive, slice-specific “z-shimming’’ method was developed, which uses slice-specific compensation moments between the gradient-echo acquisitions. The compensation moments remarkably reduce the influence of macroscopic field gradients compared with conventional mGRE sequences. Moreover, an improved SNR compared with a slice-independent “z-shimming” approach could be achieved.
The presented signal model, in combination with the new adaptive “z-shimming” approach, led to substantial improvements in the quality of R2* maps, assessed by the median and the interquartile range in different deep gray matter and white matter regions
Originalspracheenglisch
QualifikationDoktor der Technik
Gradverleihende Hochschule
  • Technische Universität Graz (90000)
Betreuer/-in / Berater/-in
  • Stollberger, Rudolf, Betreuer
PublikationsstatusVeröffentlicht - 2020

Dieses zitieren