Arterial spin-labeling magnetic resonance imaging of brain maturation in early childhood: Mathematical model fitting to assess age-dependent change of cerebral blood flow

Abstract

PurposeTo determine the trajectory of age-dependent cerebral blood flow (CBF) change in infants and young children by fitting mathematical models to the imaging data. MethodsIn this retrospective study, we reviewed the arterial spin-labeling imaging studies of 49 typically developing infants and young children at postmenstrual age (PMA) ranging from 38 to 194 weeks. All patients had normal structural MR imaging. Coregistration and gray matter segmentation were performed to extract whole-brain CBF values. Regional CBF values were obtained using manual region-of-interest placement. Curve estimation regression procedures with the corrected Akaike information criterion (AICc) were performed to determine the mathematical model best fitting the relationship between the CBF (whole-brain and regional measurements) and PMA of the patients. ResultsWhole-brain CBF trajectory was best fitted by a cubic model (AICc = 215.95; R2 = 0.566; P < .001). Whole-brain CBF at 1, 6, 12, and 24 months was estimated to be 36, 52, 58, and 55 mL/100 g/min, respectively. Regional CBF trajectory was also best fitted by a cubic model in the frontal (AICc = 233.63; R2 = 0.442; P < .001), parietal (AICc = 229.18; R2 = 0.614; P < .001), basal ganglion (AICc = 239.39; R2 = 0.178; P = .043), temporal (AICc = 236.01; R2 = 0.441; P < .001), and occipital (AICc = 236.46; R2 = 0.475; P < .001) regions. ConclusionsIn early childhood, the trajectory of CBF change was nonlinear and best fitted by the cubic model for the whole brain and all brain regions.