Quantification of cerebral vascular hemodynamics using MRI

Thesis

Cerebral perfusion is a promising biomarker for the early diagnosis of dementia. Additionally, the change in perfusion in response to vasoactive stimuli, cerebrovascular reactivity (CVR), may also prone to be a valuable dementia biomarker. Arterial spin labeling (ASL) MRI is a non-invasive technique to quantify perfusion and in turn CVR. However, the precise estimation of perfusion using ASL MRI is not only limited by partial volume effects, due to the relatively low resolution of the ASL data, but also affected by confounding factors including arterial transit time (ATT) and blood flow velocity. This thesis investigates the sensitivity of partial volume correction (PVEc) methods for gray matter (GM) perfusion estimation and the impact of confounding factors on CVR measurements.

Linear regression (LR) and spatially regularized PVEc methods were examined using simulated ASL data by comparing the estimation accuracy and the ability to preserve spatial variations in perfusion images. Results indicated that the LR approach was less sensitive to uncertainties in data while the spatially regularized method was superior in preserving spatial details.

The impact of PVEc on the repeatability of GM perfusion was investigated using data from a healthy cohort. The repeatability improved after PVEc using both LR and spatially regularized methods, implying that PVEc is beneficial to increase the reliability of GM perfusion measurements.

CVR quantification was investigated using the standard implementation of ASL (single-PLD PCASL) and a novel ASL technique (Turbo QUASAR ASL). CVR, ATT, and blood flow velocity were estimated before and after acetazolamide administration in healthy volunteers. Regional differences of CVR were found between the two ASL methods. Both PCASL and Turbo QUASAR ASL are limited by the change in blood flow velocity for the use in CVR estimation, while PCASL is also hindered by ATT. Accounting for flow velocity enhanced the CVR estimation for both techniques.

Authors: Zhao, Y

• DOI: 10.5287/ora-pm7vap750
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: Perfusion; Cerebral blood flow; Arterial spin labeling; MRI
• Subjects: Biomedical engineering; MRI; Clinical neurosciences