Multiparametric mapping of brain oxygen consumption with resting state calibrated functional MRI

BOLD and cerebral blood flow (CBF) signal perturbations induced by isometabolic vasodilation enable the estimation of BOLD and CBF cerebrovascular reactivities (CVRs) and calibration of the BOLD signal through inference of its maximum change (M). We developed a BOLD and oxygen-transport modelling approach that uses a hypercapnic estimate of M to map the oxygen extraction fraction (OEF) and cerebral metabolic rate of oxygen (CMRO2). Inducing hypercapnia requires CO2 inhalation or volitional breath-holding (BH). We present a calibrated fMRI framework aiming to overcome the limitations of induced hypercapnia that exploits endogenous resting-state (RS) modulations in brain hemodynamics. This approach was compared against BH. We derived a fitting regressor representing a non-metabolically demanding vascular signal from the average grey matter (GM) BOLD obtaining similar parametric maps between BH and a 10-min RS. Associations between average GM values were M: r=0.70, OEF: r=0.88, CMRO2: r=0.94 (p-values<10-4) with slight underestimation of parameters derived from RS (∼10%) compared to BH. The most informative frequency range to extract a vascular regressor was in the high-frequency portion of the RS spectrum (oscillation times <20 s), where modulations in systemic pressure induced by breathing occur. RS fMRI estimation of CMRO2 appears feasible, and it holds promise for research and clinical application.