The Set Increment with Limited Views Encoding Ratio (SILVER) Method for Optimizing Radial Sampling of Dynamic MRI

Abstract Purpose To present and assess a method for choosing a fixed increment between spokes in radially sampled MRI that results in higher SNR than other radial sampling increments. Theory and Methods Sampling uniformity contributes to image SNR when reconstructed using linear methods. Thus, for a radial trajectory, uniformly spaced sampling is ideal. However, uniform sampling lacks post-acquisition reconstruction flexibility, which is often needed in dynamic imaging. Golden ratio-based increments are often used for this purpose. The method presented here, Set Increment with Limited Views Encoding Ratio (SILVER), optimizes sampling uniformity when a limited number of temporal resolutions are required. With SILVER, an optimization algorithm finds the angular increment that provides the highest uniformity for a pre-defined set of reconstruction window sizes. SILVER was tested over multiple sets and assessed in terms of uniformity, noise amplification, and SNR both in simulations and in acquisitions of a phantom and healthy volunteers. Results The proposed algorithm produced trajectories that, for the optimized window sizes, had higher uniformity and lower image noise than golden ratio sampling both in a simulated single-coil system and in a multi-coil system, assessed using simulation, phantom, and in vivo experiments. The noise in SILVER optimized trajectories was comparable to uniformly distributed spokes whilst retaining flexibility in reconstruction at multiple temporal resolutions. In a resting state fMRI experiment, tSNR increases at different spatial and temporal resolutions ranged from 21-72%. Conclusion SILVER is a simple addition to any sequence currently using golden ratio sampling and significantly increases sampling efficiency and SNR.