Removing the effects of CSF partial voluming on fitted CBF and arterial transit times using FAIR, a pulsed arterial spin labelling technique.

FAIR, an arterial spin labelling technique, provides non-invasive, quantitative CBF values and arterial transit times deltat. This paper focuses on the negative impact of CSF partial voluming on FAIR results. To understand and solve this problem, we performed a theoretical analysis and a range of simulations. We then acquired FAIR data from a volunteer to illustrate our findings. We found that the determinant effect of CSF is a delayed zero-crossing during inversion recovery. The subtraction of magnitude inversion recovery data in FAIR generates erroneous negative data and distorted fit results: we simulated that for CSF percentages of 0-40%, CBF and deltat will be progressively overestimated by up to 50%. For higher CSF percentages the errors were found to increase steeply. We explored a straightforward solution: taking the magnitude of the FAIR data before fitting. This provided a remarkably strong antidote against the effects of CSF partial voluming: for CSF percentages of 0-40%, simulations now gave CBF values accurate within 1%, and deltat within 5%. The fit remained robust for high CSF fractions. Our analysis and simulations demonstrate that using magnitude FAIR data minimises the detrimental effects of CSF partial voluming. Data from a healthy volunteer illustrate these results.