Time-efficient methods for non-invasive brain blood flow imaging using arterial spin labelling MRI

Accurate knowledge of brain tissue perfusion and clear visualisation of cerebral arteries are of great interest for the diagnosis and research of a range of diseases and conditions such as stroke, dementia, Moyamoya, and cancer, as well as in drug development. 15O-H20 PET imaging is considered the gold standard for measuring cerebral blood flow (CBF), while contrast-enhanced x-ray digital subtraction angiography is the gold standard for visualising cerebral arteries. However, both methods use ionising radiation and injectable contrast agents so carry some risks to the patient and are not appropriate for longitudinal monitoring or research. There is growing interest in arterial spin labelling (ASL) magnetic resonance imaging (MRI), which uses magnetically labelled blood as an endogenous and freely diffusible tracer and can be flexibly used to visualise the arterial tree as well as to measure CBF. However, ASL is an intrinsically low SNR technique, requiring signal averaging and long scan times. The main aim of this thesis is to improve the time-efficiency of ASL MRI perfusion and angiographic scans, increasing the accuracy and SNR of CBF measurements and vessel depiction in clinically relevant scan times. First, a statistical framework was developed and validated for optimising multi- delay ASL perfusion protocol timings to improve the accuracy of CBF and arterial transit time measurements. This flexible framework can be targeted to improve the accuracy of specific parameters and provides a more informed method for designing ASL protocols. Next, this framework was used to optimise and objectively compare a range of existing pseudo-continuous ASL (PCASL) perfusion protocol designs, including time- encoded preparations, from the literature and to explore novel protocol designs. These comparisons allowed the advantages and disadvantages of each approach to be explored, and recommendations about the best ASL experimental designs to be made. Finally, the benefits of time-encoded PCASL preparations for ASL angiography were investigated. Angiography visualises the pass-through of labelled blood, in contrast to perfusion imaging which relies on the accumulation of labelled blood in tissue, and so the short label durations used in time-encoded PCASL have a greatly reduced impact on the magnitude of the ASL signal. A variable flip-angle approach was also used to remove signal discontinuities and demonstrate a significant SNR improvement over conventional PCASL angiography.