Magnetic Resonance Fingerprinting for multi-component estimations

Quantitative MR imaging can provide improved specificity and reproducibility of the clinical evaluation, increasing the diagnostic capability of the technique. However, most quantitative techniques assume the presence one single tissue within each voxel. When violated, this assumption can lead to quantification errors. To overcome this issue, the usual strategies are to perform high resolution imaging and/or to use specialized acquisitions to suppress signal of undesired components. Inspired by recent advances in multi-parametric transient-state quantitative MR imaging such as MR Fingerprinting (MRF), the aim of this thesis work was to develop intra-voxel quantification techniques based on multi-component signal models. The work was carried out at the Laboratory of Medical Physics and Magnetic Resonance of IRCCS Stella Maris and IMAGO7 Research center (Pisa, Italy). The implemented techniques included a fat-water separation method based on MR Fingerprinting. Based on a two-component signal model accounting for the simultaneous presence of fat and water, the technique provided accurate estimations of fat fraction and T1 relaxation time of both fat and water. The technique was extended by including a bone component in the signal model, and potential applications for attenuation correction in PET/MR imaging were discussed. Finally, similar multi-component signal models were used to perform myelin fraction mapping and CSF/flowing blood quantification. It was also shown that these additional parameters can be used to improve quality of synthetic MRI images, potentially increasing its applicability in clinical routine.