Quantitative MRI of the human brain : magnetisation transfer and magnetic field mapping

The ultimate goal of this thesis was to identify ways of combining the parametric maps produced by the use of multiple quantitative Magnetic Resonance Imaging (MRI) techniques. As a first stage towards this goal, this thesis focuses on magnetisation transfer (MT) imaging, and on the use of field mapping techniques to improve the reliability of other quantitative techniques such as functional MRI (fMRI) and diffusion tensor (DT) MRI. After summarising the basic principles of MRI, the MT phenomenon is described and a quantitative MT model is reviewed. A set of experiments is then described aiming at optimising the acquisition parameters for the measurement of the MT ratio (MTR). The interaction between T and MT is investigated, confirming that MTR acquisition protocols should be designed to minimize T -weighting. Next, the quantitative model of MT is used to optimise the white-to-grey matter contrast to noise ratio of a pulse sequence for MTR measurement, at both 1.5 T and 3.0 T. The following chapter is focused on the optimisation of quantitative MT for in vivo applications. First, the sensitivity to noise of the technique is investigated using simulations. Second, the implementation of a 3D pulse sequence for quantitative MT is described. The sequence is used at 1.5 T and at 3.0 T to collect data from healthy volunteers, providing normative values. Finally, the set of sampling points used to measure MT parameters is optimised using the Cramer-Rao lower bound, showing dramatic improvements in both precision and accuracy. Next, after a review of static field inhomogeneities and field mapping, the consequences of field inhomogeneities on quantitative MT are evaluated. The use of novel acquisition sequences for field mapping is investigated, the application of field-map based correction for fMRI and DT MRI data is considered, and its effects are discussed. Finally, an attempt to combine different parameters through multivariate analysis is presented, by using principal component analysis to identify patterns of association between MT parameters. Finally, an attempt to combine different tialysis is presented, by using principal componen ssociation between MT parameters.