With these three continuous ASL (CASL) techniques, magnetic labeling is achieved distal from the slice(s) of observation. This can either be accomplished by a dedicated labeling coil (two-coil CASL) or using the imaging coil itself (single-coil CASL). All the CASL techniques were shown in theory and experimentally to provide significantly better sensitivity than their pulsed ASL counterparts, although a quantitative assessment of blood flow appears less straight in absolute terms. Two-coil continuous techniques (17, 18) require dedicated labeling coils and drivers that are not commonly installed in small animal imaging setups. Also, in mice, the distance from the labeling coil to the imaging coil becomes small, making interference between both coils more difficult to avoid. Therefore, Williams et al. (1) have proposed to achieve the label with the imaging coil itself (single-coil CASL). In this case, the labeling induces off-resonance saturation of the magnetization outside the labeling slice including the image slice location, which influences the detected on-resonance signal by magnetization transfer (MT). This effect has to be compensated through an adequate control experiment. The control experiment can be accomplished with a labeling location on the opposite side of the imaging slice or at the same labeling location using dedicated amplitude-modulated rf emissions. The first solution is compromised in multi-slice studies for reasons of asymmetry and the second solution affects the labeling efficiency of the experiment.

For advanced studies, dynamic ASL (DASL) has been proposed (19) as a modification of CASL. It employs specifically adapted dynamic labeling functions so that beyond capillary blood flow, the involved relaxation times of blood and tissue can be extracted from the measurement data simultaneously.

More recently, pseudo-continuous (pCASL) techniques (20, 21) were proposed mimicking a continuous label with pulsed rf and gradient fields. The pseudo-continuous technique promises to become the most interesting technique since it has the potential to combine the advantages of PASL and CASL providing a good balance between tagging efficiency and sensitivity. It overcomes the problem of off-resonance saturation intrinsically and therefore has multi-slice capability without loss of efficiency.

In pulsed ASL (PASL), the magnetization of arterial blood is labeled using a short selective or global inversion pulse. Although a number of preparation modules and readout schemes have been proposed for pulsed ASL in clinical work, flow-sensitive alternating inversion recovery (22) is the most commonly used labeling scheme for pulsed ASL in animal studies. The CBF-dependent magnetization difference is produced by alternating acquisitions after global spin inversion of the largest possible volume of the animal and slice-selective inversion of the magnetization only in the imaging slice. In the latter case, blood entering the imaging slice during the inversion time TI was not inverted and therefore contributes to an accelerated observed relaxation in the slice. After global inversion, blood entering the imaging slice has been inverted as well and contributes to slow down the apparent tissue relaxation at field strengths at which blood T ₁ is longer than tissue T ₁. At the inversion time TI, a difference in longitudinal magnetization will be present between the two preparations. This difference is proportional to CBF, as shown in detail below.

PASL is less efficient than CASL due to long recovery periods, during which no signal is acquired, and due to a magnetic “arterial input function” that decays by longitudinal relaxation of blood (see also Note 3). Unlike in CASL techniques, the labeling zone for the FAIR technique is, however, placed symmetrically and close to the observation slice, which leads to better robustness with respect to arterial transit times (Note 4) and to an easier modeling (see also Note 5). Look–Locker inversion-recovery techniques were first proposed for rodent cardiac applications (23, 24), but also had their entry into brain studies. Despite higher sensitivity to bulk flow in larger vessels, they were shown to provide slightly better sensitivity than classical FAIR techniques (25, 26) and to open new ways to the exploration of the magnetization dynamics after labeling (27), including assessment of the shape of the magnetic “arterial input function.” Look–Locker techniques also provide an inherent and simultaneous measurement of T ₁, which enters CBF quantification, and which might be submitted to changes during the protocol.

Arterial spin labeling can be performed with any MRI setup provided that a homogeneous excitation radiofrequency coil is available. The necessary sensitive length of the excitation coil depends on the particular technique employed. Pulsed ASL techniques require coverage from brain to abdomen such that a sufficient amount of blood can be inverted. The reason for this necessity is that blood turnover in the rodent body is fast compared with humans so that non-inverted blood would enter the imaging slice before image acquisition, should the spatial coverage of the inversion be insufficient. This situation is shown in Fig. 2.

For continuous ASL techniques, in which labeling is provided during a longer time period, the rf excitation coverage has to include at least the animal’s neck for efficient labeling, which takes place in a thin slice usually placed at carotid level.

As outlined in the introduction of this chapter, high field strengths are advantageous since they provide higher MR signal and longer T ₁ values. Very high static fields, however, also impose more constraints on the robustness of rapid imaging techniques like EPI or spiral acquisition making the MR measurement more difficult in practice. Also, T ₂ relaxation is faster at very high field, which partly counterbalances the achieved signal gain, and which in addition leads to underestimation of perfusion due to the presence of blood in the imaging signal if not accounted for in the model used for data analysis (28). Efficient perfusion imaging using arterial spin labeling can be carried out in mice at field strengths as low as 4.7 T.

The theory of arterial spin labeling is based on the modified Bloch equations [2] which describe the tissue magnetization M _b (t) in the presence of inflowing spins.
where f is the blood flow, T _1b is the T ₁ of brain tissue, M _b ⁰ is the equilibrium brain tissue magnetization. M _a and M _v are the arterial and venous blood magnetization, respectively.

Quantification of blood flow with ASL relies on subtracting two images acquired with different states for arterial blood magnetization , so that equation [1] can be rewritten to

A standard kinetic model (29, 30) can be used to consider the magnetization difference ΔM _b as a tracer delivered by arterial blood flow (delivery function c(t)), which exchanges between blood and tissue compartment and which is cleared by a combination of T ₁ decay (relaxation function A(t)) and venous outflow (residue function r(t)). The signal difference ΔM _b can then be expressed as a convolution of the three previously defined functions,

Three main assumptions (see also Note 5) are made in the classical ASL model to calculate ΔM _b.