William A. Grissom1,
Chen Dong1, Laura Sacolick1, Mika W. Vogel1
1GE Global
Research,
Determining optimal phase encoding locations for three dimensional parallel excitation spokes pulses is a non-trivial problem. Current algorithms can be classified as either sparse approximation-based methods, or methods that locally optimize the encoding locations. The two approaches have complementary strengths and weaknesses: sparse approximation-based methods approach global optimality when the target excitation phase is fixed and time-dependent effects (such as off-resonance) are ignored, while local methods can be performed jointly with target phase optimization and can account for time-dependent effects, but are only locally optimal. We introduce a new algorithm similar which interleaves greedy sparse approximation-based phase encoding selection with local gradient and target phase optimization. The new method is demonstrated using multiband pulse designs.