Optimisation of quantitative magnetisation transfer (QMT) MRI to study restricted protons in the living human brain

Magnetisation Transfer (MT) imaging exploits magnetisation exchange between 'free' protons and 'restricted' protons attached to macromolecules in biological tissue, to indirectly provide access to the restricted protons, which are invisible using conventional MR imaging techniques. The Magnetisation Transfer Ratio (MTR) is calculated from a pair of images with different MT "weightings", however it reflects a complex combination of biological and acquisition dependent factors. Quantitative MT (qMT) imaging allows the examination of fundamental parameters underlying the MT exchange process independently of sequence details. The effect of B errors on MTR measurements was investigated, both theoretically and experimentally, and a method for correcting for B errors was proposed, based on the collection of a B map in addition to MTR data. The temperature dependence of many quantitative MR properties may cause systematic errors in phantom Quality Assurance (QA) measurements, which could have an impact on the interpretation of quantitative changes observed in long-term clinical studies. Many traditional thermometry methods are unsuitable for use in an MRI scanner. Using localised 1H-MRS acquisition sequences routinely available on clinical MRI scanners, and commonly available analysis packages, internal thermometry in phantoms using DSS (sodium 3-(trimethylsilyl)propane-1-sulphonate)) as a chemical shift reference was shown to be realistic, with a minimum detectable temperature difference of 100 ( 20) mK. The qMT acquisition parameters (combinations of MT pulse amplitude and offset frequency) were optimised, via the minimisation of the Cramer-Rao Minimum Variance Bound (CRMVB). Compared to a conventional acquisition, the optimisation enables less data to be acquired, reducing acquisition time without compromising uncertainties in estimated parameters. Alternatively, for the same number of MT-weighted data points, the parameter map noise could be reduced. This analytical approach was verified numerically, using Monte Carlo simulations, and experimentally, and optimised acquisition schemes were shown to be applicable to a range of brain tissues