SpinWalk: A Monte Carlo simulator for MR-signal formation in inhomogeneous tissue.

Abstract

Monte Carlo simulation is extensively utilized in functional magnetic resonance imaging (MRI) research to examine the behavior of an MR sequence in the presence of diffusion within complex microstructures. These simulations necessitate a substantial number of diffusing particles and time steps to be modeled to achieve convergence and produce robust and reliable results, which is computationally intensive. Incorporating additional parameters to enhance the realism of the simulations further intensifies this computational burden, particularly when simulating steady-state sequences, which require a long period of time to be simulated. To address this, we present SpinWalk, a high-performance Monte Carlo simulator for functional MRI. SpinWalk is free and open-source software, designed to offer a high-performance framework for facilitating the simulation of custom sequences. SpinWalk enables popular sequences in functional MRI to be efficiently simulated and ensures that results can be consistently reproduced. Key sequence and tissue parameters can be set, making SpinWalk flexible in examining different factors that contribute in signal formation. This versatility is demonstrated by replicating simulations from several previous studies, including GRE, SE, bSSFP, GRASE, and STE sequences. Performance evaluations demonstrate that SpinWalk can significantly reduce computation times, making it feasible to perform extensive simulations within a reasonable time frame.