Consistent Cerebral Blood Flow Covariance Networks across Healthy Individuals and Their Similarity with Resting State Networks and Vascular Territories.

Cerebral blood flow (CBF) represents the local blood supply to the brain, and it can be considered a proxy for neuronal activation. Independent component analysis (ICA) can be applied to CBF maps to derive patterns of spatial covariance across subjects. In the present study, we aimed to assess the consistency of the independent components derived from CBF maps (CBF-ICs) across a cohort of 92 healthy individuals. Moreover, we evaluated the spatial similarity of CBF-ICs with respect to resting state networks (RSNs) and vascular territories (VTs). The data were acquired on a 1.5 T scanner using arterial spin labeling (ASL) and resting state functional magnetic resonance imaging. Similarity was assessed considering the entire ASL dataset. Consistency was evaluated by splitting the dataset into subsamples according to three different criteria: (1) random split of age and sex-matched subjects, (2) elderly vs. young, and (3) males vs. females. After standard preprocessing, ICA was performed. Both consistency and similarity were assessed by visually comparing the CBF-ICs. Then, the degree of spatial overlap was quantified with Dice Similarity Coefficient (DSC). Frontal, left, and right occipital, cerebellar, and thalamic CBF-ICs were consistently identified among the subsamples, independently of age and sex, with fair to moderate overlap (0.2 < DSC ≤ 0.6). These regions are functional hubs, and their involvement in many neurodegenerative pathologies has been observed. As slight to moderate overlap (0.2< DSC < 0.5) was observed between CBF-ICs and some RSNs and VTs, CBF-ICs may mirror a combination of both functional and vascular brain properties.