Jorge Fernández Villena, Athanasios G. Polimeridis, Yigitcan Eryaman, Elfar Adalsteinsson, Lawrence L. Wald, Jacob K. White, and Luca Daniel, Massachusetts Institute of Technology, University of Minnesota, Massachusetts General Hospital, Harvard-MIT Division of Health Sciences Technology, USA,
This work presents an efficient full-wave frequency-domain electromagnetic simulation tool for the analysis of realistic (highly inhomogeneous) human body models, including the virtual family population, coupled with arbitrary RF coil configurations.
The proposed methodology is based on a combination of surface and volume integral equation methods that naturally exploit the inherent characteristics of the different parts of an MRI system (coil arrays, shield and realistic human body models). It also incorporates a sophisticated numerical engine to rapidly characterize the electromagnetic performance of MR scanners. This characterization includes the computation of the un-tuned coil port parameters, the current distribution for the tuned coils, and the corresponding electromagnetic field distribution in the inhomogeneous body. The accompanying open-source software, running on Matlab and on a single standard Windows desktop, is able to perform a complete frequency sweep of a complex RF coil in ~3-5 min. per frequency point, and model the interactions between electromagnetic fields and highly inhomogeneous body in ~5-10 min. per port, depending on the model resolution and the prescribed error tolerance.
The analysis of multiple coil designs is further accelerated by introducing the pre-computed magnetic resonance Green functions, a numerical compressed model for a given realistic human body, that once generated can be re-used with any coil design. This allows considerable acceleration of the simulations that can reach up to two orders of magnitude, thus reducing the overall computation time of the full electromagnetic analysis of complex multi-channel systems to few minutes.
The collection of the novel algorithms has been incorporated into MARIE (MAgnetic Resonance Integral Equation suite), a Matlab-based prototype tool, freely available in http://thanospol.github.io/MARIE/.
Keywords: Magnetic Resonance Imaging, Electromagnetic Modeling, Integral Equations, Green functions