Measuring Resting Cerebral Perfusion using Magnetic Resonance Imaging (MRI)

Introduction: We aimed to demonstrate non-invasive measurements of regional oxygen extraction fraction (OEF) from quantitative BOLD MRI modeling at baseline and after pharmacological vasodilation. We hypothesized that OEF decreases in response to vasodilation with acetazolamide (ACZ) in healthy conditions, reflecting compensation in regions with increased cerebral blood flow (CBF), while cerebral metabolic rate of oxygen (CMRO₂) remained unchanged. We also aimed to assess the relationship between OEF and perfusion in the default mode network (DMN) regions that have shown associations with vascular risk factors and cerebrovascular reactivity in different neurological conditions.

Material and methods: Eight healthy subjects (47 ± 13 years, 6 female) were scanned on a 3 T scanner with a 32-channel head coil before and after administration of 15 mg/kg ACZ as a pharmacological vasodilator. The MR imaging acquisition protocols included: 1) A Gradient Echo Slice Excitation Profile Imaging Asymmetric Spin Echo scan to quantify OEF, deoxygenated blood volume, and reversible transverse relaxation rate (R₂^’) and 2) a multi-post labeling delay arterial spin labeling scan to measure CBF. To assess changes in each parameter due to vasodilation, two-way t-tests were performed for all pairs (baseline versus vasodilation) in the DMN brain regions with Bonferroni correction for multiple comparisons. The relationships between CBF versus OEF and CBF versus R₂’ were analyzed and compared across DMN regions using linear, mixed-effect models.

Results: During vasodilation, CBF significantly increased in the medial frontal cortex ($P=0.004$), posterior cingulate gyrus (pCG) ($P=0.004$), precuneus cortex (PCun) ($P=0.004$), and occipital pole ($P=0.001$). Concurrently, a significant decrease in OEF was observed only in the pCG (8.8%, $P=0.003$) and PCun ($8.7\%,P=0.001$). CMRO₂ showed a trend of increased values after vasodilation, but these differences were not significant after correction for multiple comparisons$.$ Although R₂’ showed a slightly decreasing trend, no statistically significant changes were found in any regions in response to ACZ. The CBF response to ACZ exhibited a stronger negative correlation with OEF ($\beta =-0.104\pm 0.027$; $t=-3.852,P<0.001$), than with R₂’ ($\beta =-0.016\pm 0.006$; $t=-2.692,P=0.008$).

Conclusion: Quantitative BOLD modeling can reliably measure OEF across multiple physiological conditions and captures vascular changes with higher sensitivity than R₂’ values. The inverse correlation between OEF and CBF across regions in DMN, suggests that these two measurements, in response to ACZ vasodilation, are reliable indicators of tissue health in this healthy cohort.

Introduction: Cerebral blood flow (CBF) is an important physiological parameter that can be quantified non-invasively using arterial spin labelling (ASL) imaging. Although most ASL studies are based on single-timepoint strategies, multi-timepoint approaches (multiple-PLD) in combination with appropriate model fitting strategies may be beneficial not only to improve CBF quantification but also to retrieve other physiological information of interest.

Methods: In this work, we tested several kinetic models for the fitting of multiple-PLD pCASL data in a group of 10 healthy subjects. In particular, we extended the standard kinetic model by incorporating dispersion effects and the macrovascular contribution and assessed their individual and combined effect on CBF quantification. These assessments were performed using two pseudo-continuous ASL (pCASL) datasets acquired in the same subjects but during two conditions mimicking different CBF dynamics: normocapnia and hypercapnia (achieved through a CO₂ stimulus).

Results: All kinetic models quantified and highlighted the different CBF spatiotemporal dynamics between the two conditions. Hypercapnia led to an increase in CBF whilst decreasing arterial transit time (ATT) and arterial blood volume (aBV). When comparing the different kinetic models, the incorporation of dispersion effects yielded a significant decrease in CBF (∼10–22%) and ATT (∼17–26%), whilst aBV (∼44–74%) increased, and this was observed in both conditions. The extended model that includes dispersion effects and the macrovascular component has been shown to provide the best fit to both datasets.

Conclusion: Our results support the use of extended models that include the macrovascular component and dispersion effects when modelling multiple-PLD pCASL data.