On Blood and Mechanical Motion Sensitization of Encoding and Decoding in Magnetic Resonance Imaging
OPEN ACCESS
Loading...
Author / Producer
Date
2022
Publication Type
Doctoral Thesis
ETH Bibliography
yes
Citations
Altmetric
OPEN ACCESS
Data
Rights / License
Abstract
The assessment of three-dimensional blood flow dynamics by means of Phase-Contrast Magnetic Resonance Imaging (PC-MRI) holds promising potential for the non-invasive diagnosis of diseases of individual vessels and entire structures such as the heart. A variant of PC-MRI, 4D Flow MRI, offers novel biomarkers, which are gaining importance for the classification of e.g. valvular heart disease. The evaluation of mean blood flow and turbulence, by additionally encoding the Reynolds stress tensor, may further enable the quantification of stenosis severity, the assessment of hemolysis and other factors. As these markers can influence clinical decision making, knowledge of the accuracy and precision of the estimated parameters is of utmost importance. Given the intrinsic limitations of PC-MRI in terms of spatial and temporal resolution and the employed hardware, estimated flow parameters are sensitized to the blood flow itself but also to other, undesired contributions. In addition, long acquisition times in 4D Flow MRI hamper its clinical applicability and patient acceptance.
In previous works, the echo planar imaging (EPI) readout technique has been suggested to reduce the time needed for acquisition of 4D Flow MRI exams. Here, it is shown that employing EPI for 4D Flow MRI results in misregistration, velocity estimation errors and degrading spatial resolution depending on blood flow patterns. Therefore, it is concluded that for shortening scan time other acceleration methods such as compressed sensing in conjunction with standard gradient echo (GRE) imaging are favorable.
Current turbulence encoding models for PC-MRI are based on assumptions regarding the time scales of the underlying flow field. In this work, the turbulence encoding model is derived and related to diffusion MRI and turbulence theory. Results of PC-MRI simulations employing large eddy simulation (LES) data as input show that current turbulence encoding models need to be revisited as they systematically underestimate turbulence parameters. Subsequently, a correction method based on probing the Lagrangian turbulence spectrum is presented and used to gauge the encoding model to reinstate the accurateness of turbulence parameter estimation. The method is demonstrated for PC-MRI of stenotic flows.
While encoding and reconstruction techniques can be further optimized, MRI hardware limitations remain. Based on a linear, time-invariant description of the MRI gradient system, the influence of mechanical resonances on PC-MRI data are highlighted. It is shown that residual background phases, which result in biased velocity estimation, can be reduced both in amplitude and spatial order by PC-MRI sequence optimization. The influence of mechanical resonances on spatial encoding is evaluated, demonstrating that EPI readouts are particularly vulnerable to undesired contributions due to mechanical motion of the gradient system. The gradient performance of lower-field systems is shown to benefit from reduced Lorentz forces, mitigating the influence of mechanical resonances. This benefit is contrasted with the reduced signal-to-noise ratio of lower-field systems. Based on a comparison of an MRI system operated at standard and at lower-field strength, increased gradient fidelity as well as reduced sound pressure levels are demonstrated for the lower-field configuration.
Permanent link
Publication status
published
External links
Editor
Contributors
Examiner : Kozerke, Sebastian
Examiner : Obrist, Dominik
Book title
Journal / series
Volume
Pages / Article No.
Publisher
ETH Zurich
Event
Edition / version
Methods
Software
Geographic location
Date collected
Date created
Subject
Organisational unit
09548 - Kozerke, Sebastian / Kozerke, Sebastian