Magnetic resonance measurement of blood and CSF flow rates with phase contrast--normal values, repeatability and CO2 reactivity.

BACKGROUND: Similarity in flow pulsatility has been proposed as a basis for semi-automated segmentation of vessel lumens for MR-based flow measurement, but re-examinations of salient aspects of the methodology have not been widely reported. METHODS: 12 normal control subjects underwent repeated (3*Baseline+1*5%CO2) phase contrast measurements of CSF flow through the cerebral aqueduct and foramen magnum, and CBF through the 6 large cranial vessels at the level of the 1st vertebra. Average flows were calculated for regions temporally correlated (0.3 < or = Rthreshold < or = 0.95) to user defined seed points and their 3 x 3 neighbours. RESULTS: Arterial CBF averaged 710ml/min, with low variability (+/- 4%/17%, intra-individual/group CV respectively) and was the only flow to respond significantly to 5%/mmHg CO2. Venous outflow was much smaller (298ml/min +/- 10%/ 72%), possibly due to the weak venous pulse and variable venous anatomy. Average CSF flows exceeded the classical 0.4ml/min CSF production rate and were highly variable--aqueduct: 0.6ml/min (+/- 50%/93%), foramen magnum: -2.7ml/min (+/- 158%/226%). CONCLUSIONS: This preliminary analysis identified procedural steps that can improve the accuracy and repeatability of MR flow measurements, but the process remains user-dependent for the weakly pulsatile foramen magnum CSF and venous flows where variability remains a significant confound even to relatively large perturbations such as CO2 administration.