Magnetic Resonance Elastography: An Analysis of Image Acquisition Sequences and Inversion Algorithms for Elasticity Imaging of Healthy Liver Tissue

Abstract

Liver fibrosis is a form of liver disease that if left untreated can progress to potentially fatal liver cirrhosis. The gold standard technique for testing patients with potential liver fibrosis is a liver biopsy. This technique has numerous drawbacks. It is highly invasive in nature, expensive and prone to sampling errors. Non-invasive techniques capable of examining the liver for signs of fibrosis are highly sought after by medical professionals. One such technique is known as Magnetic Resonance Elastography (MRE). When a liver becomes fibrotic there is a significant change in the elastic properties of the tissue. This alteration of tissue elastic properties forms the basis of contrast for diagnosing fibrosis using MRE. MRE functions by placing a mechanical wave driver on the surface of the subject, thus introducing shear waves to the subject medium. The shear wave propagation is then imaged using a specialised MRE pulse sequence. This obtained wave image is then passed through an inversion algorithm which produces a map of internal shear modulus. The end goal is for medical professionals to use these stiffness maps to confidently diagnose patients without having to perform a liver biopsy. In this thesis three different inversion algorithms are analysed and compared with the current Siemens Product Inversion (PI) algorithm. It was concluded that, while all three techniques showed potential, further investigation and work is required before these algorithms could be used as an alternative to the PI algorithm. The second section of this thesis compares three Work in Progress (WIP) MRE imaging sequences with the current Siemens product imaging sequence. The investigated pulse sequences all performed well when imaging healthy liver tissue. However, the WIP sequences were designed to perform MRE on more difficult anatomical structures such as the heart. Further study would be needed to investigate how well all pulse sequences could handle more difficult scenarios.