NOTICE: This event will be held via Blue Jeans. The link will be posted below.

Blue Jeans link: https://bluejeans.com/6788336326

We propose a novel optimization algorithm for radiofrequency (RF) pulse design in magnetic resonance imaging (MRI), that regularizes the magnitude and phase of the target (desired) magnetization pattern separately. This approach may be useful across applications where the relative importance of achieving accurate magnitude or phase excitation varies; for example, saturation pulses "care" only about the magnitude excitation pattern. We apply our new design to the problem of spin "prephasing" in 3D functional MRI using blood-oxygen-level-dependent (BOLD) contrast; spin prephasing pulses can mitigate the signal loss observed near air/tissue boundaries due to the presence of local susceptibility gradients. We show that our algorithm can improve the simulation performance and recover some signal in some regions with steep susceptibility gradients. In all cases, our algorithm shows better phase correction than a conventional design based on minimizing the complex difference between the target and realized patterns. The algorithm is open-source and the computation time is feasible for online applications. In addition, we evaluate the impact of the choice of (initial) excitation k-space trajectories, both in terms of trajectory type (SPINS vs extended KT points) and overall pulse duration.

Chair: Dr. Jon-Fredrik Nielsen