Physical Sciences for Health CDT
Thesis project - "The development of a novel MRI based method for measuring blood perfusion in neurovascular damage"
Professor Andrew Peet, Institute of Cancer and Genomic Studies
Dr Jan Novak, Institute of Cancer and Genomic Studies
Dr Niloufar Zarinabad, Institute of Cancer and Genomic Studies
Dr Hamid Dehghani, School of Computer Science
Whilst water is the most abundant molecule in almost all tissue, the motion of water in tissue is complex and poorly understood. The aim of this study is to improve our fundamental physical understanding of water molecular motion in tissue through magnetic resonance imaging (MRI). It is well known that the motion of water molecules can be detected and imaged using MRI. The technique is commonly termed diffusion weighted imaging with the presumption that the major source of water motion in tissue is through purely diffusive processes. However, there are other influences on the motion of water, most notably the bulk motion of water molecules in small blood vessels which is related to blood perfusion of the tissue. These two types of water motion, diffusive motion and bulk motion, are usually quantified using the intravoxel incoherent motion (IVIM) model which assumes an exponential decay of the MRI signal with gradient strength for each component giving a biexponential fit. However, IVIM is an imperfect model for water molecule motion in tissue through its numerous complex physical environments. This project aims to gain a more complete understanding of the water motion in tissue by developing a method which does not depend on the simplified biexponential IVIM model.
We have previously shown that diffusion weighted imaging data can be fitted using the Auto-Regressive Moving Average (ARMA) model where there is no prior assumption to the number of components in the signal (with each component defining a unique physical phenomena- such as diffusion). Other unknown processes may also affect the detected signal from the tissue such as tubular flow or glandular secretion. Previous studies have also suggested that the diffusion phenomenon possesses multi-exponential behaviour. Hence a model with the ability to optimise the number of components could provide a new insight into the physical properties of water motion in tissue.
Whilst this variable component method gives excellent fits to the data, we do not yet know the physical processes which the metrics describe. It will be important to determine the relationship of the fitting parameters to traditional perfusion and diffusion metrics but also extend this to other processes governing water motion in tissue. An appropriate phantom model that exhibits both diffusion and perfusion effects will need to be designed and built in order to investigate the phenomena. The phantom model will allow observations of the subtle variations in the parameters upon the signal. This phantom can then be used to validate the ARMA model. The project will also require further development of the ARMA model’s formalism as well as investigating its properties in virtual simulations and testing in volunteers prior to application in patients.
Diffusion tensor imaging measures magnitude as well as direction of the diffusion in tissue providing greater understanding of its structure. Techniques applied to diffusion weighted imaging will be integrated into diffusion tensor imaging to give a more complete understanding of water motion in tissues
The motivation behind this work lies in the importance of water motion as a sensitive indicator of tissue disruption through a number of disease processes including trauma. MRI is routinely used in these patients and so can provide an important modality for translation of these findings to a clinical arena. However, current imaging, including diffusion weighted imaging, is a poor predictor of patient outcome. An optimised multicomponent model which better characterises tissue damage could provide better biomarkers of patient wellbeing.