The signal in Magnetic Resonance Imaging (MRI) has two constituents: magnitude and phase. In most clinical MRI studies, only the magnitude information is used. Recently, there has been rapid development of MRI methods which utilise the phase as it can be used to increase image contrast, correct image distortions or provide information about the tissue magnetic properties (e.g. in quantitative susceptibility mapping). Phase image processing is not trivial as the range of phase values which can be measured in MRI is limited to 2π radians. This means that any phase value larger than 2π is wrapped back into this range and phase unwrapping methods are necessary to recover the true phase values. Existing unwrapping methods are often time-consuming and fail in regions which are noisy or have strong phase variations. The initial aim of this project is to develop realistic numerical models of phase variations in the human brain. These models will then be used to train a convolutional neural network to give a fast and robust solution to the phase unwrapping problem and compare it against existing unwrapping algorithms. This project, in a fast-moving and exciting research area, will give you an insight into MRI signal properties, MRI image processing, Matlab programming and machine learning algorithms.
Characterising phantoms for elastography
Supervisors: Peter Munro and Pilar Garcia Souto
Student: Ana-Maria Nicolaie
Optical Coherence Elastrography obtains images of the mechanical properties of tissue and is performed by using Optical Coherence Tomography to image how tissue deforms in response to a mechanical load. A crucial part of this project is the development of phantoms with controllable mechanical properties. This project entails the construction of such phantoms along with their characterisation using an instron machine which provides gold standard characterisation of the mechanical properties of materials. This project is largely experimental, though an elementary understanding of continuum mechanics will be beneficial.