Analysis of improvements to post-stroke visual field mapping using prospective motion correction in fMRI

There are approximately 110,000 stroke cases in England each year with the lifetime risk of a stroke for a middle-aged man being 1-in-6 and for a middle-aged woman being 1-in-5. It is estimated that 60% of stroke survivors suffer from some form of visual impairment immediately after a stroke. Of those, one in three still suffer from visual impairment three months post-stroke onset. This visual impairment can have a marked impact on quality of life, as it leads to a lack of mobility, loss of confidence and increased rates of collisions and accidents.
In ongoing work we have been using magnetic resonance imaging (MRI) to identify spared cortex surrounding the lesions in the occipital lobe with capacity to support vision. With functional MRI, we have been estimating the visual field coverage of different areas within the spared visual cortex and using this information to train participants on a difficult visual task.
In addition to functional MRI, there is a powerful set of additional techniques that can provide important additional information. Despite this however, these additional measurements are really never use in combination in the clinical setting.
- diffusion weighted imaging
- angiography/perfusion measurements
- high resolution anatomical scans
One of the main reasons for these technologies are not often combined is the amount of time a participant has to spend in an MR scanner and problems and artefacts related to motion. For example, due to the co-occurrence of movement disorders such as chorea and tremors post-stroke, many stroke survivors may not be able to benefit from such a targeted intervention, since the involuntary movements caused by these disorders can cause motion artefacts in MRI) and therefore also all the imaging-dreived data.
A detailed study of the benefits (and limits) of different approaches to improving data quality (prospective motion correction, parallel imaging, multiband, compressed sensing) for fMRI derived data is therefore crucially important and likely to lead to tangible improvements in what kind of imaging protocols are used to assess these patients.
Aims and objectives
To investigate and quantify the improvements attainable with new technologies aimed at speeding up and stabilising magnetic resonance imaging (MRI) measurements. In particular, we will investigate and quantify how currently used protocols for functional MRI, diffusion weighted imaging and angiography/perfusion measurements can be improved by
- use of parallel imaging (e.g SENSE) and multiband acquisition
- prospective correction for motion based on navigator echoes
- using undersampled data acquisition (compressed sensing)
- (time permitting) use of machine learning techniques to improve quantification of imaging-derived measures for angiography, etc.
Methodology
The plan for the start of the PhD program is as follows. The student will:
- compare the currently used setup and imaging protocol (at 3T) for imaging a cohort of stroke survivors and healthy controls.
- investigate the quality improvements (in healthy participants) of moving from currently used scanner (3T Achieva) to an updated hardware platform that includes many technical improvements (3T Ingenia)
- assess the trade-off between imaging speed-up and data quality with 2 kinds of complementary parallel imaging approaches (SENSE and Multiband)
- investigate further speed-ups by also using computationally intensive approaches to using undersample data (compressed sensing, reconstruction of sparse signals), which will be particularly beneficial for angiography
- assess the improvements in patient comfort achieved by speeding up and/or reducing imaging related acoustic noise. This is particularly important for our development of Patient and Public Involvement (PPI).
Alignment to EPSRC's strategies and research areas
- specific thematic call, relating to Imaging technologies for vulnerable subjects.