Functional magnetic resonance imaging (fMRI) is an important tool for understanding neural mechanisms underlying human brain function. Understanding how the human brain responds to stimuli and how different cortical regions represent the information, and if these representational spaces are shared across brains and critical for our understanding of how the brain works. Recently, multivariate pattern analysis (MVPA) has a growing importance to predict mental states from fMRI data and to detect the coarse and fine scale neural responses. However, a major limitation of MVPA is the difficulty of aligning features across brains due to high variability in subjects’ responses and hence MVPA has been generally used as a subject specific analysis. Hyperalignment, solved this problem of feature alignment across brains by mapping neural responses into a common model to facilitate between subject classifications. Another technique of growing importance in understanding brain function is real-time fMRI Neurofeedback, which can be used to enable individuals to alter their brain activity. It facilitates people’s ability to learn control of their cognitive processes like motor control and pain by learning to modulate their brain activation in targeted regions. The aim of this PhD research is to decode and to align the motor representations of multi-joint arm actions based on different modalities of motor simulation, for instance Motor Imagery (MI) and Action Observation (AO) using functional Magnetic Resonance Imaging (fMRI) and to explore the feasibility of using a real-time fMRI neurofeedback to alter these action representations. The first experimental study of this thesis was performed on able-bodied participants to align the neural representation of multi-joint arm actions (lift, knock and throw) during MI tasks in the motor cortex using hyperalignment. Results showed that hyperalignment affords a statistically higher between-subject classification (BSC) performance compared to anatomical alignment. Also, hyperalignment is sensitive to the order in which subjects entered the hyperalignment algorithm to create the common model space. These results demonstrate the effectiveness of hyperalignment to align neural responses in motor cortex across subjects to enable BSC of motor imagery. The second study extended the use of hyperalignment to align fronto-parietal motor regions by addressing the problems of localization and cortical parcellation using cortex based alignment. Also, representational similarity analysis (RSA) was applied to investigate the shared neural code between AO+MI and MI of different actions. Results of MVPA revealed that these actions as well as their modalities can be decoded using the subject’s native or the hyperaligned neural responses. Furthermore, the RSA showed that AO+MI and MI representations formed separate clusters but that the representational organization of action types within these clusters was identical. These findings suggest that the neural representations of AO+MI and MI are neither the same nor totally distinct but exhibit a similar structural geometry with respect to different types of action. Results also showed that MI dominates in the AO+MI condition. The third study was performed on phantom limb pain (PLP) patients to explore the feasibility of using real-time fMRI neurofeedback to down-regulate the activity of premotor (PM) and anterior cingulate (ACC) cortices and whether the successful modulation will reduce the pain intensity. Results demonstrated that PLP patients were able to gain control and decrease the ACC and PM activation. Those patients reported decrease in the ongoing level of pain after training, but it was not statistically significant. The fourth study was conducted on healthy participants to study the effectiveness of fMRI neurofeedback on improving motor function by targeting Supplementary Motor Cortex (SMA). Results showed that participants learnt to up-regulate their SMA activation using MI of complex body actions as a mental strategy. In addition, behavioural changes, i.e. shortening of motor reaction time was found in those participants. These results suggest that fMRI neurofeedback can assist participants to develop greater control over motor regions involved in motor-skill learning and it can be translated into an improvement in motor function. In summary, this PhD thesis extends and validates the usefulness of hyperalignment to align the fronto-parietal motor regions and explores its ability to generalise across different levels of motor representation. Furthermore, it sheds light on the dominant role of MI in the AO+MI condition by examining the neural representational similarity of AO+MI and MI tasks. In addition, the fMRI neurofeedback studies in this thesis provide proof-of-principle of using this technology to reduce pain in clinical applications and to enhance motor functions in a healthy population, with the potential for translation into the clinical environment.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:754385 |
Date | January 2018 |
Creators | Al-Wasity, Salim Mohammed Hussein |
Publisher | University of Glasgow |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://theses.gla.ac.uk/30761/ |
Page generated in 0.0015 seconds