Principal component analysis to identify the major contributors to task-activated neurovascular responses

BACKGROUND: Consensus on the optimal metrics for neurovascular coupling (NVC) is lacking. The aim of this study was to use principal component analysis (PCA) to determine the most significant contributors to NVC responses in healthy adults (HC), Alzheimer's disease (AD), and mild cognitive impairment (MCI). NEW METHOD: PCA was applied to three datasets: 1) 69 HC, 2) 30 older HC, 34 AD, and 22 MCI, 3) 1&2 combined. Data were extracted on peak percentage change in cerebral blood flow velocity (CBFv), variance ratio (VR), cross-correlation function peak (CCF), and blood pressure, for five cognitive tasks. An equamax rotation was applied and factors were significant where the eignevalue was ≥1. Rotated factor loadings ≥0.4 determined significant NVC variables. RESULTS: PCA identified 12 significant factors accounting for 78% of variance (all datasets). Contributing variables loaded differently on the factors across the datasets. In datasets 1&2, peak percentage change in CBFv contributed to factors explaining the most variance (45–58%), whereas cognitive test scores, fluency and memory domains contributed the least (15–37%). In the combined dataset, CBFv, CCF and fluency domain contributed the majority (33–43%), whereas VR and attention the least (6–24%). CONCLUSIONS: Peak percentage change in CBFv and the visuospatial task consistently accounted for a large proportion of the variance, suggesting these are robust NVC markers for future studies.