Improving fMRI analysis methods for the measurement of cerebrovascular function

Beckerleg, Ryan 2023. Improving fMRI analysis methods for the measurement of cerebrovascular function. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF - Accepted Post-Print Version Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (27MB) | Preview
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (268kB)

Abstract

This work endeavoured to improve current methods of investigating cerebrovascular function using functional magnetic resonance imaging (fMRI). The main areas identified for improvement were retrospective motion correction (Chapter 3) and data-based quantification of cardiac pulsatility and heart rate variability (Chapters 4 and 5). Chapter 3 demonstrated that conventional motion correction techniques result in erroneous motion estimates in scans of cerebrovascular function. The severity of this was investigated using an external camera and novel methods were introduced to improve motion estimates. The ICA-based method more accurately estimated motion when compared with the other methods. However, this didn’t work for multi-PLD pseudo-continuous ASL scans. Additionally, the ICA-based methods performed the best when quantifying measures of cerebrovascular function. Therefore, I would recommend the use of an ICA in the calculation of motion parameters for scans of cerebrovascular function. Chapter 4 aimed to develop data-based methods to quantify cardiac pulsatility using resting-state fMRI (rfMRI). Ultimately, these methods failed because the cardiac signal was aliased and could not be accurately located. This chapter also showed that a regression-based approach using cardiac-related components as regressors would be better in the estimation of cardiac pulsatility. In Chapter 5, two novel methods of estimating cardiac pulsatility were introduced. Both methods created training datasets by isolating independent components that were cardiac-related. These were then used to train FSL’s ICA-based Xnoisifier (FIX). The first method (HRV method) used quality physiological traces to achieve this. Whereas the second method (Frequency method) used only the frequency data. Then, FIX was used to isolate cardiac components for all rfMRI datasets. Estimates of cardiac pulsatility were produced and compared to a gold standard. Results showed that both methods correlated highly with this gold standard. The HRV method showed slightly higher correlations than the Frequency method and is the recommended method.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Physics and Astronomy
Subjects:	Q Science > QC Physics
Uncontrolled Keywords:	Magnetic Resonance Imaging MRI Functional Magnetic Resonance Imagining fMRI fMRI analysis Motion correction Motion reduction Cerebrovascular function Pulsatility Cardiac Pulsatility Heart Rate Variance HRV Independent component analysis ICA
Funders:	EPSRC
Date of First Compliant Deposit:	13 February 2024
Last Modified:	13 Feb 2024 15:42
URI:	https://orca.cardiff.ac.uk/id/eprint/166281

Actions (repository staff only)

Edit Item