Perfusion MRI can be carried out non-invasively by labeling the hydrogen nuclei of the intravascular arterial water using one or two radiofrequency pulses. These pulses serve to invert the longitudinal magnetization of arterial blood just upstream of the region of interest (ASL: Arterial Spin Labeling, AST: Arterial Spin Tagging), whose measurable magnetization and relaxation time T1 will thus be modified.
During acquisition (in ultrafast echo planar type sequence), the signal thus includes the measurable magnetization of the region of interest to which is added the magnetization from by the labeled blood pool present in the explored volume. A second unlabeled acquisition serves as a reference to then calculate the perfusion images.
With pulsed labeling techniques (PASL: Pulsed Arterial Spin Labeling), a volume of blood is labeled upstream of the region of interest by a short RF pulse. Signal acquisition is carried out after a delay TI (of 1 to 2 seconds). The difference in signal between measurements with and without spin labeling reflect the amount of labeled blood arriving in the volume of interest during the delay TI, to calculate cerebral blood flow. The particularity of this method derives from the limited lifespan of the endogenous tracer, since the magnetization of the labeled blood bolus returns to equilibrium (depending on the time constant T1 of the blood). This limits the time delay between labeling and signal acquisition.
Due to imperfections in the spatial selectivity of the RF labeling pulse there has to be a gap between the region of interest and the labeling zone. The time needed to cross this space and arrive in the measurement zone will vary according to the velocity of the blood and the labeled bolus. These arterial transit time variations must be minimized in the context of flow quantification.
Adding a saturation pulse to the slice of interest will also reduce the effects of the imperfect inversion pulse slice profile.
With continuous labeling techniques (CASL: Continuous Arterial Spin Labeling), an adiabatic inversion pulse is applied to the blood upstream of the slice using a continuous radiofrequency of weak intensity associated with a gradient applied in the direction of flow. As a result of continuous spin labeling, the signal of the labeled slice of interest will reach steady state. The perfusion parameters are calculated by comparing with the signal from the same, unlabeled slice.
The drawback with this technique relates to the amount of RF energy delivered, and this has led to the development of sequences with pseudo-continuous labeling or the use of two coils (one for labeling, with a reduced volume of emission centered on the vessels, and one for signal reception).
The difficulties of these techniques relate to:
The drawbacks of this technique derive from its weak signal-to-noise ratio and the inability to explore the whole brain. It may nevertheless be suitable if only one specific area is studied, as in the case of a stroke, for example.