Rapid and robust microstructural imaging with diffusion MRI

Thesis

Diffusion magnetic resonance imaging (dMRI) provides an non-invasive way to detect tissue microstructure information by probing the Brownian motion of water molecules and estimating the diffusion properties using signal analysis models. dMRI is broadly used in neuro disease detection and basic neuroscience studies. The most widely used dMRI acquisition technique is a two-dimensional (2D) single-shot echo planar imaging sequence due to its rapid acquisition speed. Recently, many advanced diffusion analysis models have been proposed which offer more specific estimates of tissue biophysical properties from acquired dMRI signals compared to simpler models like diffusion tensor imaging (DTI). However, these analysis models require acquisition of large numbers of diffusion volumes, which increases the dMRI scan time significantly and hinders their wider in vivo applications. Many efforts have been made to develop acquisition and reconstruction methods for fast dMRI scans, yet the trade-off between scan time, spatial resolution and model specificity still remains as a major challenge for dMRI studies.

The works in this thesis aim to develop acquisition and reconstruction methods that allow for rapid and robust brain microstructural imaging with dMRI. Firstly, a joint k-q reconstruction method based on Gaussian process is developed for accelerating multi-shell dMRI acquisition. Secondly, a computationally efficient eddy current and motion robust joint k-q reconstruction method is developed to increase the robustness of the joint reconstruction methods when the subjects may be uncooperative during scans. Thirdly, we extend the previous methods to accelerate diffusion-relaxometry imaging for microstructural imaging with higher specificity by using a k-q-TE joint acquisition and reconstruction method. The methods developed in this thesis seek to reduce the dMRI scan time while preserve the image quality and analysis accuracy, which offers potential to enable advanced microstructural imaging within clinically feasible time.

Authors: Ye, X

• DOI: 10.5287/ora-b75ogoxax
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: microstructural imaging; MRI reconstruction; diffusion MRI
• Subjects: Magnetic resonance imaging