A Graphical Method for Designing Time-Optimal Non-Cartesian Gradient Waveforms
Abstract
One of the fundamental challenges for non-Cartesian MRI is the need of designing time-optimal and hardware-compatible gradient waveforms for the provided $k$-space trajectory. Currently dominant methods either work only for certain trajectories or require significant computation time. In this paper, we aim to develop a fast general method that is able to generate time-optimal gradient waveforms for arbitrary non-Cartesian trajectories satisfying both slew rate and gradient constraints. In the proposed method, the gradient waveform is projected into a space defined by the gradients along the spatial directions, termed as $g$-space. In the constructed $g$-space, the problem of finding the next gradient vector given the current gradient vector under desired slew rate limit and with desired direction is simplified to finding the intersection between a line and a circle. To handle trajectories with increasing curvature, a Forward and Backward Sweep (FBS) strategy is introduced, which ensures the existence of the solution to the above mentioned geometry problem for arbitrary trajectories. Furthermore, trajectory reparameterization is proposed to ensure trajectory fidelity. We compare the proposed method with the previous optimal-control method in simulations and validate its feasibility for real MR acquisitions in phantom and human knee for a wide range of non-Cartesian trajectories. The proposed method enables accurate and fast gradient waveform design, achieving significant reduction in computation time and slew rate overshoot compared to the previous method. The source code will be publicly accessible upon publication of this study.