Use this URL to cite or link to this record in EThOS:
Title: Structural and functional networks for components of language : network analysis of stroke aphasia patients
Author: Zhao, Ying
ISNI:       0000 0004 7657 4457
Awarding Body: University of Manchester
Current Institution: University of Manchester
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
The PhD project used a multimodality imaging database, which included T1, resting-state fMRI (rs-fMRI), and DTI images of stroke aphasia, to explore the relationships between the brain and behaviours. As network approaches for language processing have become popular, the current project was especially interested in functional and structural connections that are related to behavioural variations in post-stroke aphasia. Chapter 1 reviewed the findings and methods that are used in language network studies. Chapter 2 analysed patients' T1 images with principal component analysis with varimax rotation and uncovered the hidden vascular structure behind the lesions. On the behaviour side, four behavioural principal components were extracted: phonology, semantics, executive, and fluency. The behavioural components were used across Chapters, as this pattern is stable and replicated previous results found in the literature. When using the lesion components to predict the behavioural components in a regression model, the lesion component approach implicated additional regions that explained the extra behavioural variance. Chapter 3 analysed the patients' rs-fMRI data with regards to variation in the timing of the BOLD response. Researchers have previously reported that there can be regionally-specific hemodynamic time shifts in resting-state MRI. The chapter discovered that these time shifts (positive and negative) can be related to the patients' variation across the PCA-derived behavioural components, reflecting possible diaschisis or hemodynamic compensation of the brain post stroke. Finally, Chapter 4 and Chapter 5 analysed the functional connections in rs-fMRI and white matter connections in DTI, respectively. There were significant regularised regression models for predicting the behavioural components, which might suggest that the connections were an important aspect in explaining the behavioural variance. However, when regressing out the variance that can be explained by the lesion, there were no significant models. This indicated that, in stroke, the lesion is the dominating modality. Chapter 6 reports an explorative investigation which combined imaging modalities (T1 and FC data) using joint independent component analysis. The resultant joint components were biased by the covariance in T1 data as it had larger covariance among its features. The T1 components were similar to the vascular structure while the FC patterns were less interpretable. In conclusion, multimodality neuroimaging data helps to understand language mechanisms post-stroke from different angles. Nevertheless, the lesion itself is the predominant modality in explaining behavioural variation in stroke aphasia.
Supervisor: Lambon Ralph, Matthew ; Halai, Ajay Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available