Design strategies for improved velocity-selective pulse sequences

Over the years, a variety of MRI methods have been developed for visualizing or measuring blood flow without the use of contrast agents. One particular class of methods uses flow-encoding gradients associated with an RF pulse sequence to distinguish spins in flowing blood from stationary spins. While a strength of these particular methods is that, in general, they can be tailored to capture a desired range of blood flow, such sequences either do not provide a sharp transition from stationary spins to flowing spins, or else are long, generating relaxation losses and undesirable SAR, and have limited immunity to resonance offsets and to RF inhomogeneity. This article provides design methods for improving these longer RF pulse sequences, especially to provide improved immunity to RF inhomogeneity, and also to improve immunity to resonance offsets, as well as to minimize RF sequence length. These design methods retain the flexibility to capture a desired range of blood flow, with sharp transitions between stationary spins and flowing blood. These improvement strategies are demonstrated through Bloch equations simulations of examples of these new sequences in the presence of blood flow. Examples of improved sequences that should prove suitable for use at 3.0Tesla are presented.