Improved ultrashort time echo and dynamic contrast enhancement magnetic resonance imaging based on stack-of-stars golden angle radial sampling scheme

Abstract

Magnetic resonance imaging (MRI) is a radiation-free medical imaging technique and it is widely used for clinical diagnosis. MRI is developed from the physical phenomenon of nuclear magnetic resonance (NMR). MR signals are excited, spatially encoded and acquired by the corresponding radio-frequency (RF) pulse, gradient system and receiving coil system respectively. Compared to traditional medical imaging techniques like X-ray and Computed Tomography (CT), MRI requires a relatively long data acquisition time to gradually fulfill the k-space to form an image. Aggressive acceleration is applied to MRI to satisfy the clinical requirements for dynamic imaging. Meanwhile, MR signals from the short T_2 tissues experience significant decay before data acquisitions, these tissues are invisible in MRI images. Ultra-short time echo (UTE) protocols are essential for imaging short T_2 tissues.
This thesis explores the methods of imaging short T_2 tissues and dynamic imaging using MRI. Experiments are performed using simulations, multiple resolution phantoms and human subjects. Several frameworks with a hybrid three-dimensional (3D) sampling scheme are developed for UTE imaging and highly accelerated dynamic MRI imaging respectively.
A significant contribution presented in this dissertation is the development and assessment of the 3D stack-of-star central out golden angle protocol for UTE imaging. Echo Time (TE) is reduced in the proposed protocol by applying the adaptive phase encoding gradient. The intrinsic sensitivity to the hardware imperfection in the non-Cartesian UTE imaging is alleviated by employing trajectory measurement and iterative density compensation function (DCF). The self-gating property is effectively utilized to achieve motion resolved lung imaging, enabling investigation of lung functions incorporated with oxygen-enhanced MRI.
This thesis also presents a significant contribution for improved reconstruction of dynamic contrast enhanced (DCE) MRI based on the stack-of-stars golden angle sampling scheme. In this thesis, Low Rank plus Sparse (L+S) with two sparsity constraints temporal total variation (TV) and temporal fast Fourier transform (FFT) is developed and assessed. The additional sparsity constraint temporal FFT is employed to alleviate the temporal blurring caused by temporal TV and recover the dynamic contrast. The proposed method achieved high spatial-temporal resolution, high reconstruction efficiency and improved dynamic contrast simultaneously when comparing with other methods in reconstructing several simulated phantom datasets and free-breathing liver DCE-MRI datasets. Another contribution in this thesis is the soft-weighting for motion corrected DCE-MRI. A soft-weighting matrix is integrated into the proposed reconstruction framework for motion corrected DCE-MRI. Compared to the motion subdivision, a better motion compression is achieved by soft weighting function without increasing computational complexity.

Date of Award	Jul 2022
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Chengbo Wang (Supervisor), C.F. Kwong (Supervisor), Penny Gowland (Supervisor) & Paul Glover (Supervisor)