New methods for localizing brain activity with Magnetic Resonance Imaging

Abstract

Blood-oxygen-level-dependent functional magnetic resonance imaging (BOLD fMRI) is the most frequently used non-invasive method for localizing brain activity in vivo with good spatial and fair temporal resolution. It is shown in a number of studies that BOLD fMRI has clinical potential, such as in neurosurgery, in stroke rehabilitation or in moderate to severe traumatic brain injury. However, for major clinical benefit further improvements of the method and technology are necessary. Indeed, BOLD measures changes in blood oxygenation levels secondary to neuronal activity. BOLD fMRI provides therefore an indirect measure of brain activity, which implies a poor specificity to neuronal activity of the depicted functional signal. The main motivation of this thesis was to improve localization of brain activity using MRI to increase our knowledge of the brain and its functional organization, as well as to provide better tools for studying and curing brain diseases in a long-term perspective. In this thesis, we investigated diffusion functional MRI (DfMRI) as a tool for depicting brain activity more directly. DfMRI is based on the theory that neuronal cell swelling during brain activity produces changes in diffusion properties of the space surrounding the neurons, which can be measured using diffusion weighting functional MRI using strong diffusion weighting. This method has the potential to measure more precisely brain activity with excellent specificity to neuronal activity of the signal. However, we demonstrated that DfMRI only measures SE BOLD related changes and does not detect brain activity more directly. Thanks to the recent development of 3D imaging and parallel imaging, it is possible to improve the BOLD fMRI sensitivity and specificity to neuronal activity. The best candidate sequences to perform 3D imaging combined with parallel imaging are the PRESTO and SSFP sequences. In this thesis, we compared these methods and assessed which was best to perform fMRI studies. We showed that 3D PRESTO combined with parallel imaging produces activity maps highly sensitive and specific to grey matter activity, while SSFP depicted mostly the BOLD signal from the veins draining the activated cortex. The PRESTO sequence was therefore best for fMRI studies. Our capacity to produce activation maps more specific to brain activity also relies on our understanding of the BOLD signal and the underlying physiological mechanisms accompanying it. At the moment, our knowledge of the BOLD effect is incomplete. However, thanks to the unique properties of the SSFP sequence we could depict velocity changes induced by neuronal activity in the large arteries feeding the activated cortex. The similarity between the BOLD signal and the SSFP signal showed that the BOLD signal might be explained by arterial CBF changes. The combination of the PRESTO and SSFP activity maps allowed a full overview of the BOLD signal from the large arteries to the capillaries and the draining veins, giving optimum information for studying neuronal activity.