Abstract
Functional magnetic resonance imaging (FMRI) is a non-invasive technique used to produce maps of brain activation based on the blood oxygen level dependent (BOLD) contrast. It is widely used in neuroscience research and, to a more limited extent, clinically for neurosurgical planning. A stimulus or task is conventionally used to evoke brain activity; however, functional connections in the brain can also be determined from correlations in the fluctuations of the BOLD signal observed in the absence of any external stimuli (resting-state FMRI).Gradient-echo echo-planar imaging (GE-EPI) is the most common technique for acquiring FMRI data because of its sensitivity to the BOLD signal changes and relatively high temporal resolution. GE-EPI images are however affected by signal dropout caused by magnetic field gradients arising from the differences in the magnetic susceptibilities of materials in the head. This hampers the detection of BOLD signal changes in areas of the brain close to air-bone interfaces such as the orbitofrontal and inferior temporal regions.
Theoretical calculations and numerical simulations were performed to determine the degree of signal recovery needed to detect task-evoked and resting-state BOLD signal changes in such areas of signal dropout. Three different approaches to reduc-ing signal dropout in GE-EPI images were then explored. The first two, z-shimming optimised to recover signal in grey matter and hyperbolic secant (HS) radiofrequency pulses aimed to reduce the dropout caused by through-plane susceptibility gradients. The third, using a combination of the HS pulse with compensatory gradients in the frequency encoding direction, aimed to further reduce the dropout correcting for in-plane susceptibility gradients. The impact of all three techniques on the ability to detect task-evoked and resting-state BOLD signal changes was investigated in a group of six healthy volunteers.
| Date of Award | 2013 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Gareth Barker (Supervisor) & Steven Williams (Supervisor) |