Validating Arterial Spin Labelling Cerebral Blood Flow measure with perfusion phantom

Introduction The maintenance of cerebral functions strongly depends on the sustained supply of blood and the ability to adapt cerebral blood flow (CBF) to its metabolic demands. Occlusion of afferent vessels bears a risk for ischemic stroke and permanent cerebral damage. Arterial spin labeling (ASL) technique [1] can be used to quantify CBF. However, the CBF measure extracted by this technique may contain artifacts [2] and moreover may be affected by large within-subject, between-subject and regional variability [3]. An unambiguous and reliable description of physiological features by this method may therefore represent a “mission impossible”. It is therefore mandatory to validate and optimize the CBF measure by independent methods. Validation via a perfusion phantom is the aim of the present study. Methods A perfusion phantom was constructed with 125cm3 volume filled with SiO2 spheres of different diameters [e.g. 0.01 mm 0.05 mm and 0.1 mm]. Due to this set of diameters an average pore size of 0.0138 mm was achieved. With this average pore size, we are a factor 2 larger than average diameter of capillaries of ~ 6 m. A pump (Harvard Apparatus Pulsatile Blood pump for hemodynamic studies) provides a circular flow into the phantom via the input- and output tubes. The pump allows exact setting of flow volume and flow rate. MR imaging was conducted in a 3T Siemens Prisma Scanner. The pseudo CASL sequence [1] were set with the following paramerters: TE = 30 ms, TR = 2 s, 3 mm 3 mm 3 mm voxel dimension, 64 64 matrix size, FOV 192 mm, 38 slices, and slice thickness = 3 mm. In order to assess variability and precision of CBF measure estimate we systematically changed the following ASL parameters [4]: Post-labeling delay [100 ms:200:ms:1500ms], Bolus duration # RF [ 20, 60, 80]; and the following pump parameters: flow ratio [0.5, 1, 1.5] and Flow volume*Flow rate [20:20:200]. Results The perfusion phantom is operating (Fig 1.A and B). Due to the large amount of different sequence and pump settings we present a limited set only. Setting the pump Flow volume*Flow rate at 200 [ml/min], a flow ratio of 1, post labeling delay of 500 ms revealed an extraordinary high precision of CBF estimation by the perfusion phantom: 198.3029 ± 0.76 [ml/100g/min] (Fig 1.C and Fig. 2). The signal to noise ratio (SNR) was 4.2. The 0expected value of CBF was therefore underestimated by 1.7 [ml/100g/min]. The error in CBF estimation is 0.5 % which is at same level as the accuracy of the pulsatile pump. Conclusion The perfusion phantom with the pulsatile pump showed high accuracy of CBF estimation that were reliable and reproducible. The observed CBF values showed high precision (i.e. low standard deviation) that allows to draw conclusion about the variance of the CBF measure originated by the ASL sequence only. Uncertainty in CBF estimation by ASL technique may therefore be assessed. This novel approach will finally allow clinicians, physicians and researchers to unambiguously estimate disease-related CBF effects. Therefore, it is relevant for patients as well as for socioeconomic aspects. The perfusion phantom will be generalized to be used within other institutions to enable quality controls on different scanners with different ASL sequences and different field strengths: since the phantom is stable and easy to send. References [1] Dai, W., et al., Continuous flow-driven inversion for arterial spin labeling using pulsed radio frequency and gradient fields. Magn Reson Med, 2008. 60(6): p. 1488-97