Applications of susceptibility-based brain imaging using 7T MRI

MRI is a powerful non-invasive technique for imaging soft tissues and for providing information on chemical composition. In this thesis, the effects of magnetic susceptibility from iron contained in ferritin and haemoglobin, and of myelinated tissue were studied. These effects were measured using a range of imaging techniques, including susceptibility weighted imaging (SWI) and quantitative susceptibility mapping (QSM). Initially, simulations of static and defusing spins were performed to gain an understanding of the behaviour of different systems, which agreed well with comparison to the results from analytical models. For dipoles randomly placed in a spherical volume and in the x-y plane, the R2* and ∆ω values were approximately constant, but increased for dipoles restricted to a line parallel to the z axis and decreased for dipoles in a line in the x-y plane. The fitting of the phase variation of a simulated cylinder was then used to calculate the susceptibility values of a ferritin-doped cylindrical agar phantom, which correlated well between 3T and 7T, and increased with iron concentration. Post-mortem imaging allows MR parameters to be studied in greater detail than in vivo imaging, but it is also confounded by fixation and storage conditions. An experimental setup was assembled to allow temperature-controlled scanning of a post-mortem brain. The R1, R2* and susceptibility values all decreased with increasing temperature, due to the reduction in the susceptibility of iron dominating the effect of the increased rate of diffusion of water. This was similar for a ferritin-doped cylindrical agar phantom, but the R2* values increased, due to the uniformity of agar. A longitudinal imaging study of the post-mortem brain was performed with respect to fixation time. The volume of the tissue decreased. The R2* and susceptibility values decreased and the R1 values increased overall, but were also dependent on the location and depth within the tissue. The thalamus relays information between different areas of the brain. In 1.5T and 3T MRI, the contrast between the thalamic nuclei is limited, thus motivating an evaluation of 7T post-mortem and in vivo MR imaging. A range of imaging sequences were optimised and the thalamic nuclei were manually delineated in the histology and corresponding post-mortem MR images, with a high degree of confidence, with reference to the Morel atlas. These were corroborated with the use of k-means clustering, Canny edge detection and hierarchical clustering. The nuclei in the in vivo images were also delineated with a reasonable degree of confidence, by comparing the boundaries across a range of imaging contrasts. For the groupings obtained from hierarchical clustering, the QSM clusters allowed the most nuclei to be identified on average, followed by the combined FLASH (fast low angle shot) magnitude and QSM clusters. The brain has a high level of metabolic activity, which requires a significant supply of blood. Disruption to this can result in the loss of function, as cerebrovascular diseases. The cerebral blood vessels were imaged with optimised gradient echo and time of flight sequences, to calculate venous and arterial maps, overlap maps and a venous atlas. The distribution of vessels between subjects was similar, with little overlap between the venous and arterial maps. The venous atlas detailed the deep grey matter vessels better than the manually segmented Ward atlas. Finally, oximetry measurements were inferred from the susceptibility values, which were higher within grey matter compared to white matter, and were observed to increase up to a vein radius of 1mm and then decreased.