Measuring motor-training induced neuroplasticity using diffusion and functional MRI: a sensitive, pathology-robust approach

Reid, Lee Bremner (2017). Measuring motor-training induced neuroplasticity using diffusion and functional MRI: a sensitive, pathology-robust approach PhD Thesis, Faculty of Medicine, The University of Queensland. doi:10.14264/uql.2017.797

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
s4328115_phd_final.pdf Thesis (open access) application/pdf 8.58MB 0
Author Reid, Lee Bremner
Thesis Title Measuring motor-training induced neuroplasticity using diffusion and functional MRI: a sensitive, pathology-robust approach
School, Centre or Institute Faculty of Medicine
Institution The University of Queensland
DOI 10.14264/uql.2017.797
Publication date 2017-06-02
Thesis type PhD Thesis
Supervisor Stephen Rose
Roslyn Boyd
Ross Cunnington
Total pages 224
Language eng
Subjects 080106 Image Processing
111403 Paediatrics
110903 Central Nervous System
Formatted abstract
Brain injury constitutes a significant health burden worldwide, and outcomes are generally poor. To improve treatment, we must better understand how injury affects brain function, how the brain responds to learning pressures (neuroplasticity), and the interaction between the two. This is difficult, however, because of current technological shortcomings.

This thesis focusses on the measurement of neuroplasticity in response to motor training in situations such as rehabilitation following brain injury. Although the principles and methods discussed generalise to many forms of brain pathology, discussion is generally focussed on unilateral cerebral palsy (UCP) because altered brain-development in UCP makes analyses particularly difficult. This begins with a literature review (Chapter 2) that outlines UCP, rehabilitative approaches, and the relationship between atypical brain organisation and early-life brain injury. Also discussed is how novel rehabilitation strategies may stem from studying brain changes induced by rehabilitation, and why the most informative findings are likely to come from studies utilising multiple imaging modalities.

Blood Oxygen-Level Dependent Functional Magnetic Resonance Imaging (fMRI) measures brain activity indirectly and has been used to study neuroplasticity in both healthy participants and those with brain pathology. Although fMRI confounds are well understood, no published work has explored their cumulative effect when brain injury is present. Resultantly, confounds have largely been ignored, leading to heterogeneous findings that have provided little toward understanding brain changes, biologically speaking. Chapter 3 examines the cumulative effect of fMRI confounds, assumptions, and other issues, when attempting to measure neuroplasticity in participants with brain injury. This work concludes that, although fMRI can be expected to locate approximate regions of brain activation in an individual scan, numerous confounding factors make its interpretation, in terms of neuroplasticity, difficult without additional information acquired through alternative means.

Diffusion MRI (dMRI) is an imaging modality that is typically used to identify the brain's white-matter pathways and calculate metrics that are influenced by factors such as myelination or axonal density. Standard dMRI analyses are ill suited to studying brain injury due to their reliance on brain 'atlases' which indicate relationships between brain regions and brain functions. Brain pathology can prevent reliable atlas registration, and alter relationships between structure and function. Chapter 4 details a novel dMRI analysis designed to measure neuroplasticity, addressing the need for a sensitive and reliable alternative to standalone-fMRI. This method utilises surface-based fMRI analyses, rather than brain atlases, to locate regions involved in motor execution. Corticospinal (CST) and thalamocortical tracts emanating from these regions are delineated using dMRI tractography and machine learning. This method was applied to children with UCP - the majority of whom presented with significant brain pathology. Relationships between dMRI measures of CST microstructure and clinical measures of motor function were identified. In these and other measures, the present method outperformed a cutting-edge voxelwise fMRI+dMRI technique applied to the same data.

Studying plasticity in participants with brain injury requires first understanding the neuroplastic processes of healthy people. Imaging literature in this area is largely fragmented: studies are typically mono-modal and have incomparable methodologies. Chapter 5 and Chapter 6 address this through a multimodal investigation into motor-learning driven neuroplasticity. Twenty-four healthy adults practised a finger-thumb opposition sequence with their non-dominant hand daily. After four weeks, task performance improved and fMRI activation associated with trained-sequence execution decreased. In 'trained' motor areas, transcranial magnetic stimulation mean evoked potentials (MEPs; at the pollicis brevis) increased; MEP map volume did not. These responses overlapped spatially with fMRI and potential cortical thickness increases. These results imply that a long-term-potentiation-like process occurred predominantly in regions already responsible for conducting the task, improving task-processing efficiency. Potential cortical thickness increases were also found in the dorsolateral prefrontal cortex (dlPFC). Diffusion changes were seen in striatal regions connected to the dlPFC and connections between these regions. The aforementioned fMRI-driven dMRI method revealed changes to the CST consistent with myelination.

Chapter 7 describes an attempt to take this analysis further: to measure brain changes in 14 children with UCP who participated in 20 weeks of virtual reality therapy. Clinical improvements were limited and no changes in associated white matter microstructure were detected. The fMRI-driven dMRI method demonstrated a high degree of test-retest reliability in the 13 enrolled controls, and so it is likely that statistical power was limited by degrees of clinical improvement. The number of enrolees required to achieve meaningful power for such measurements, however, was unexplored in the literature.

These results make clear that a barrier to investigating neurorehabilitation-induced neuroplasticity is uncertainty surrounding statistical power. Chapter 8 describes power analyses for paediatric UCP neurorehabilitation trials measuring cortical thickness and dMRI changes. Calculations were based on results presented in the preceding chapters. The results can be utilised to formally plan study numbers, preventing underpowered trials. This analysis also allowed a number of recommendations to be made that may further reduce barriers to successful neurorehabilitative trials.
Keyword Diffusion-weighted magnetic resonance imaging
Blood-oxygen level dependent functional magnetic resonance imaging
Cerebral palsy
Cortical thickness
Magnetic resonance imaging
Multimodal imaging

Document type: Thesis
Collections: UQ Theses (RHD) - Official
UQ Theses (RHD) - Open Access
Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Created: Thu, 25 May 2017, 12:15:30 EST by Lee Bremner Reid on behalf of Learning and Research Services (UQ Library)