Studying functional magnetic resonance imaging with artificial imaging objects

Blood oxygenation level dependent (BOLD) functional magnetic resonance imaging (fMRI) is an indirect method for measuring information processing in the brain. The method has enabled mapping human brain function in an unprecedented variety of tasks and conditions, and with a spatial resolution of the order of 1 mm. In this dissertation, artificial imaging objects, or phantoms, with adjustable signal intensity were used to simulate and investigate the generation of fMRI signals. The objective was to characterise, and devise means to characterise, fMRI signal components that arise from methodological reasons, impeding the correct physiological interpretation of the signals. The first study involved building an fMRI phantom, where an electric current was applied to introduce magnetic field inhomogeneity within a magnetic resonance signal source. It was shown that the changes of field homogeneity and thus fMRI signal, largely corresponded to the human BOLD changes, even though the physical mechanisms were different. The mechanical properties of phantoms and brain however differ. Thus it was important to look into the attributes of phantoms that would make the fMRI signal from the phantom similar to brain scanning data. The second study examined geometric distortions in the echo-planar imaging method—commonly employed in both fMRI and diffusion tensor imaging—using a purpose-built structural phantom. In the third study, another fMRI activation phantom was built. There the induction wires were located outside the source of the fMRI signal, and thus the partial volume effect limiting the usability of the first fMRI phantom was abated. The phantom was applied to induce artificial activations that could be utilized to deduce periods when simultaneously measured brain activations would yield deviant activation levels due to unphysiological causes. In the last study, an fMRI phantom was used to show that transient fMRI signal components, often witnessed in brain activation data, could occur in the absence of corresponding physiological signal, resulting from the sole signal change