Velocity‐driven adiabatic fast passage for arterial spin labeling: Results from a computer model

Abstract

Velocity-driven adiabatic fast passage (AFP) is commonly employed for perfusion imaging by continuous arterial spin labeling (CASL). The degree of inversion of protons in blood determines the sensitivity of CASL to perfusion. For this study, a computer model of the modified Bloch equations was developed to establish the optimum conditions for velocity-driven AFP. Natural variations in blood velocity over the course of the cardiac cycle were found to result in significant variations in the degree of inversion. However, the mean degree of inversion was similar to that for blood moving at a constant velocity, equal to the time-averaged mean, at peak velocities and heart rates within normal ranges. A train of RF pulses instead of a continuous RF pulse for labeling was found to result in a highly nonlinear dependence of the degree of inversion on RF duty cycle. This may have serious implications for the quantification of perfusion.Magn The development of arterial spin labeling (ASL) MRI has made it possible to study microvascular blood flow completely noninvasively, offering a means of investigating tissue perfusion in numerous functional and pathological states. Much work has been performed to assess and improve the accuracy of perfusion values from ASL (1-3). However, its inherently low sensitivity remains one of the most significant limitations of the technique, and the labeling process is a key factor in determining the sensitivity.Velocity-driven adiabatic fast passage (AFP) was applied initially for MR angiography (4). It is now widely regarded as the preferred method for continuous ASL (CASL) because it provides a continuous supply of blood that is labeled by inversion. However, the degree of inversion is limited by several factors, including the amplitudes of the magnetic field gradient and radiofrequency (RF) pulse used for AFP; the velocity of blood, which varies in arteries over the course of the cardiac cycle; and the spin relaxation in blood. In addition, the duration of RF pulses is often limited by RF amplifiers, especially on some clinical scanners, and the relatively long pulse required for labeling in CASL experiments (2 or 3 s) must be divided into a series of short pulses. It has been assumed that the effectiveness of AFP is proportional to the duty cycle of the resulting train of RF pulses (2,5).A computer modeling approach was adopted for this study in order to investigate the influences of several parameters on velocity-driven AFP individually, in a controlled manner. The aims were to establish optimal RF pulse parameters, to determine the effect of pulsatile variations in the velocity of blood, and to study the dependence of the degree of inversion on the duty cycle of a discontinuous RF pulse during labeling. The modeling was performed with one set of parameters for imaging humans in a clinical scanner, and a second set of parameters for imaging rodents in a small-bore system for animals. THEORYTo invert moving spins by AFP, an RF pulse, B 1 , is applied in the presence of a magnetic field gr...

Understanding and optimizing the amplitude modulated control for multiple‐slice continuous arterial spin labeling

“…Therefore, based on a peak amplitude of 220 Hz for the AM control, the labeling pulse should have an amplitude of 156 Hz. However, the optimum B 1 value for the labeling pulse was found to be approximately 110 Hz (20). This indicates that optimal mean B 1 for the AM control is about 40% greater than the optimum B 1 for labeling, given the same G and v. 240 mm s Ϫ1 , where there is no difference between m z with or without spin relaxation.…”

“…A higher frequency offset is expected to demand greater B 1 for optimal inversion by velocity driven AFP (20,22,23), but the dependencies of ␣ and ␣ AM on pulse sequence parameters and the relaxation times and velocity of blood are expected to exhibit similar patterns at 10 kHz to those shown using the computer model at 5 kHz. The labeling plane was 15 or 16 mm from the central imaging plane (to lie at the back of the brain), B 1 was 200 Hz for the label and 200*͌2 ϭ 283 Hz for the AM control, 56 image pairs were averaged, and acquisition time ϭ 8 min and 45 s. For the AM control, the modulation frequency was 250 Hz.…”

“…A detailed description of the model is given in ref. (20). Briefly, the change in the magnetization of a spin that follows B eff ϭ B 1 ϩ ⌬B, where the contribution of the frequency offset, ⌬B, at a position, z, given by (G.z) Ĝ , is calculated at 100,000 time points as the spin covers a specified distance, from Ϫz max to ϩz max .…”

“…In the computer model and in the results presented here, variables are expressed in the following units: Hz for B 1 , Hz mm Ϫ1 for G, Hz for f, mm s Ϫ1 for v, and seconds for T 1 and T 2 (for example, the centers of the AFPs are at z ϭ Ϯf/G). The model is designed to investigate CASL with the AM control, using parameters appropriate for imaging humans at 1.5 T and a separate set of parameters for imaging rats at 2.35 T (details of the human and animal models are given in Utting and coworkers (20)). A series of studies was carried out to assess the influence of each pulse parameter, spin velocity, and spin relaxation on the performance of the AM control by calculating m z in the imaging plane.…”

“…However, neither f nor is accounted for in ␤, meaning that it is of limited use for analysis of the AM pulse. Furthermore, spin relaxation affects the efficiency of the label (20,22,23) and the AM control pulses, and neither T 1 nor T 2 is incorporated into ␤. A more subtle appreciation of the pulse is required.…”