Cerebrovascular diseases represent one of the most prevalent subsets of all cardiovascular diseases, the number one cause of mortality worldwide. In particular, ample evidence exists indicating that cerebral small vessel disease (SVD) is the most prevalent neurological disorder, underlying multiple neurological conditions such as stroke and dementia. Cerebral SVD denotes a range of pathological processes that affect the small brain vessels, leading to progressive, often age-related, cognitive decline. Although the contribution of cerebrovascular disease in dementia is likely still underestimated, increasing evidence supports the role of SVD in many forms of dementia, including Alzheimer’s disease, and also in healthy aging.

Imaging of cerebrovascular physiology thus presents great potential to provide suitable biomarkers of healthy aging as well as various neurological conditions. A large number of novel techniques have appeared in the past decades for imaging not only cerebral blood vessels, but also and most importantly the dynamics of blood flow, mostly based on MRI. Besides brain perfusion, or cerebral blood flow (CBF), which describes the blood supply of oxygen and nutrients to the tissues, closely reflecting tissue health, it is also important to measure cerebrovascular reactivity (CVR), which describes the ability of blood vessels to dilate or contract in order to adjust CBF in response to increases or decreases in the arterial blood CO2 pressure.

Arterial spin labelling (ASL) provides a non-invasive method for imaging perfusion, by magnetically labelling the water in the arterial blood, measuring the magnetization in the tissues after a certain time delay, and then subtracting this from the magnetization measured in non-labelled (control) images. If ASL data are sampled across multiple post-labelling delays (PLD), suitable kinetic models can be estimated to quantify several perfusion-related parameters. Unfortunately, ASL’s inherently low signal-to-noise ratio (SNR), together with the complex underlying kinetics, still poses tremendous challenges to ASL’s practical implementation. Our contributions in this field have included the optimization of the sampling strategy and model estimation of multiple-PLD ASL to allow voxelwise measurements with improved accuracy (Santos et al., 2011), as well as the comparison of different acquisitions and modelling approaches in terms of the inter- and intra-subject variability of the resulting perfusion measurements (Sousa et al., J. Magn. Reson. Imag. 2014).


The commonly used blood oxygen level dependent (BOLD) contrast is thought to closely reflect CBF changes, and therefore provides an indirect but convenient mechanism for measuring CVR, due to its easy implementation and relatively high SNR. The manipulation of respiratory gases inducing hyper/hypocapnia has been used for this purpose, but its complexity and discomfort may hinder the wider applicability of the technique. Our contributions in this field are based on exploring non-invasive vasoactive respiratory protocols, and we have already shown that they can produce reliable CVR measurements provided that the underlying haemodynamics is suitably modelled (Sousa et al., NeuroImage 2014)(Pinto et al., NeuroImage 2016).


Despite its undeniable convenience, the BOLD signal provides only an indirect and qualitative measure of CBF, which may be problematic when studying populations with altered cerebrovascular physiology, and/or longitudinal changes in response to plasticity mechanisms, disease progression or therapeutic interventions. In these cases, a more direct and quantitative measure may be obtained using ASL. We have contributed a series of studies establishing the added value of using ASL in fMRI studies of brain activation; besides CBF quantification, we showed that ASL-fMRI also yields better spatial localization and lower inter-subject variability than BOLD-fMRI (Tjandra et al.,  NeuroImage 2005)(Pimentel et al., Hum. Brain Mapp. 2013).


Principal Investigator: Patrícia Figueiredo