| Summary: | <p>The brainstem plays a crucial role in a great number of vital functions such as respiratory regulation, visual reflexes, and the perception of pain. The small size and close proximity of the nuclei requires high-resolution functional magnetic resonance imaging (FMRI). However, brainstem FMRI using conventional gradient-echo echo-planar imaging (GRE-EPI) techniques is challenging due to the increased signal dropout and geometric distortions in the brainstem.</p><p>The primary aim of the work presented in this thesis was to investigate alternative methods for brainstem FMRI in order to overcome some of the challenges associated with single-shot GRE-EPI techniques. Towards this goal 3D segmented sequences were explored, which have the advantage that the size of the geometric distortions is not proportional to the resolution at which is scanned. In particular, two sequences were investigated: balanced steady-state free-precession (bSSFP) and spoiled gradient echo (SPGR).</p><p>First, a set of experiments was conducted, in which each experiment aimed to isolate a limited range of sequence properties in order to characterize and assess the potential of the candidate sequences. It was found that bSSFP has better noise characteristics compared to GRE-EPI when applied with a 2D acquisition, but when 3D readouts were used the signal instabilities increased dramatically.</p><p>Based on these findings, experiments that investigate the influence of multishot acquisitions on signal instabilities caused by physiological noise were performed. The signal instabilities were found to mainly originate from regions of CSF and blood and were highly correlated to the cardiac cycle. Several correction methods were explored and one method was identified to be implemented <em>in vivo</em>.</p><p>A novel method that allows real-time cardiac synchronization of the k-space acquisition was developed. The developed methods used a custom parallel imaging reconstruction to allow for acquisition with a fixed volume frame rate, which is desirable for FMRI purposes. The method was found to reduce the signal instabilities in 3D SPGR and bSSFP significantly.</p><p>A comprehensive assessment of two currently available retrospective correction techniques was conducted and their practicalities were compared. Recommendations are made to improve the robustness of the investigated correction methods. A novel optimization method was implemented, which was developed to determine the optimal regressor set for retrospective corrections. The method can be applied to image based as well as k-space based methods.</p>
|
|---|