Test-retest reliability and reproducibility of long-label pseudo-continuous arterial spin labeling

Arterial spin labeling MRI can quantify the cerebral blood flow (CBF) without exogenous tracer. However, the variation of arterial transit time across different brain regions introduces bias for measuring local CBF, especially for those subjects with long arterial transit time (ATT). Long post-labeling delay (PLD) or multi-PLD methods could mitigate the problem of heterogenous ATT at the expense of the signal-to-noise ratio (SNR). Long-label ASL might address the low SNR problem by increasing the amount of labeled arterial blood. Thus, we hypothesized that with the same relatively long PLD, long-label pCASL may be more robust and reproducible than standard-label pCASL in population with potentially prolonged ATT. The purpose of the study was to investigate the reliability and reproducibility of long-label pCASL in the whole brain and vascular regions of interest in an elderly population, compared with standard-label pCASL.

Twenty adult volunteers (14 males, 6 females; age, 56.6 ± 17.2 years) were scanned twice on one 3.0 T scanner by standard-label pCASL (label duration (LD) = 1500 ms, PLD = 2000 ms) and long-label pCASL (LD = 3500 ms, PLD = 2000 ms). The intraclass correlation coefficient (ICC), within-subject coefficient of variation (wsCV), random noise and signal coefficient of variation(CoV) were used to assess global and regional reliability and reproducibility. Measurement agreement and difference were compared in different brain regions using correlation coefficient plots and Bland-Altman plots respectively.

CBF value measured by long-label pCASL was overall higher than standard-label pCASL in all ROIs. Long-label pCASL had higher ICC than standard-label pCASL in most ROIs, and lower wsCV, random noise and CoV in all ROIs. Regardless of ASL method used, anterior circulation flow territories (ICC, 0.93-0.97; wsCV, 0.03-0.06) had higher CBF reliability and reproducibility than posterior circulation flow territories (ICC, 0.89; wsCV, 0.06-0.08). In all ROIs, the correlation analysis showed higher test-retest agreement (r_long-label > r_{standard-label}) and the Bland-Altman plots demonstrated lower measurement difference in long-label pCASL.

The study demonstrated good reliability and reproducibility of long-label pCASL in anterior brain regions in the elderly population. To further improve CBF quantification in a long-ATT population while proper PLD is already used, increasing the label duration may help.