Highly-accelerated multi-shot diffusion-weighted imaging and multi-echo functional magnetic resonance imaging

Abstract

Single-shot EPI (ssEPI) is commonly used for data acquisition of diffusion-weighted imaging (DWI) and multi-echo functional magnetic resonance imaging (ME-fMRI). However, the geometric fidelity and the attainable spatial resolution in ssEPI are suboptimal and limited. Though multi-shot EPI (msEPI) can mitigate the geometric distortion and achieve higher spatial resolution, the prolonged scan time of msEPI hinders its feasibility in clinical routine scan. In this thesis, the pulse sequences based on msEPI were designed to achieve high in-plane acceleration in DWI and ME-fMRI for data acquisition. In addition, the compatible reconstruction frameworks were proposed to reconstruct images with high SNR and high resolution. The practical challenges on intravoxel incoherent motion (IVIM), such as low SNR and demanding acquisition, may hinder its adoption in clinical use. In the first study, a specific acquisition scheme is designed for IVIM to acquire highly-accelerated DWI data with multiple b-values. Furthermore, a reconstruction scheme based on parametric POCS-based reconstruction of multiplexed sensitivity encoding (POCSMUSE) and IVIM model fitting is proposed to produce images with high SNR and reduced geometric distortion for subsequent quantitative estimation. High angular resolution diffusion imaging (HARDI) can reveal the fibre crossing configurations while pronounced spatial distortion may hamper its clinical adoption. The second study is designed to improve geometric fidelity by a high acceleration factor in k-space without the limitation of the number of coil channels. The preliminary results demonstrates that the approximate orientation density function to the reference standard can be obtained. There is a trade-off between the achievable spatial resolution and the number of echoes in ME-fMRI. In the third study, a novel acquisition method with sliding window (SW) feature is proposed to achieve high spatial resolution in 4-echo fMRI. Two novel reconstruction flowcharts based on parametric POCSMUSE and MUSE respectively are designed to generate the aliasing-free and distortion-mitigated images. The SW operation for data composite in the third study may smooth the fluctuation of resting state fMRI. The fourth study is designed to estimate the effect of SW operation on the identification of resting state networks at two different TRs and then determine the feasible TR for the proposed MUSE-based reconstruction in resting state ME-fMRI.