Effect of the Temporal Sampling Strategy for the Assessment of Cerebral Perfusion in Multidelay Arterial Spin Labeling

ABSTRACT Multidelay arterial spin labeling (ASL) MRI requires balancing temporal resolution with scan duration. The goal of the current study is to evaluate the influence of the temporal sampling strategy, implemented through temporal downsampling, on the quantitative accuracy of cerebral perfusion parameters obtained using multidelay ASL. In this prospective single‐center study, 12 healthy volunteers (mean age, 37 ± 9 years; seven men and five women) underwent multidelay pseudo‐continuous ASL on a 3‐T MRI system. Data were acquired using a variable time repetition scheme with 61 time points at a 75‐ms sampling interval. Cerebral blood flow (CBF) and arterial transit time (ATT) maps were calculated using seven temporally downsampled datasets with sampling intervals of 75 (gold standard), 150, 300, 450, 600, 750, and 900 ms. The normalized root mean square error (RMSE) between each sampling interval and the gold standard (75 ms) was calculated for both CBF and ATT. Additionally, subgroup analyses were performed for regions with ATT shorter or longer than the median, as determined from the 75‐ms interval dataset. The normalized RMSE of the CBF and ATT increased with more aggressive temporal downsampling. Across all sampling schemes, the normalized RMSE of CBF was significantly lower in regions with shorter ATT than in those with longer ATT ( p < 0.05). Temporal sampling configuration influences the quantitative reliability of CBF and ATT estimation. CBF bias is particularly pronounced in regions with prolonged ATT.