Simulation of RF Electromagnetic Fields for MRI: Methods and Applications

Christopher Collins¹
¹New York University School of Medicine, United States

Synopsis

Keywords: Physics & Engineering: RF Safety, Physics & Engineering: Hardware, Physics & Engineering: Physics

Simulations of electromagnetic fields have many valuable uses for design and evaluation of RF coils with consideration of both performance and safety. What methods are best depend on the intended application, available resources, and preferences of the researcher. Successful use of RF simulations for MRI requires good understanding of the simulation process, the expected behavior of RF coils and the fields they produce, and how these fields relate to signal and/or heating during MRI. In this presentation, we will cover important concepts and methods for simulating RF coils, fields, and their effects in MRI.

Introduction:

RF fields in MRI are applied to manipulate nuclear magnetic moments, but in doing so cause heating of tissues. RF coils are used to detect signal from nuclear magnetic moments, but are also sensitive to noise induced by random motion of ions and molecules throughout the body. While we typically desire a homogeneous field for nuclear excitation, this is increasingly difficult to achieve as we move to higher frequencies in search of better SNR. And while we desire a configuration of receive coils with high sensitivity to signal and field distributions optimal for SNR and accelerated imaging, design of such arrays can be confounded by the variation of field distributions with subject geometry.
Simulation of RF fields in MRI is increasingly used to understand observed effects, ensure patient safety (including in the presence of active and passive metallic implants), design transmit and receive coils and arrays, and design pulses and sequences that maximize desired image characteristics, minimize undesired effects and artifacts. There are a wide variety of approaches to simulation and interpretation of results, and the best methods are often dependent on the particular application.
In this presentation we will review a wide variety of methods for field calculation, representation of coils and the human body, interpretation of results, and compromises between these approaches for a wide array of engineering problems in MRI.

Consideration of Commonly-used Simulation Methods and Settings for MRI:

The following figures summarize some considerations in selecting methods and settings for RF field simulations. Simulation methods included here are Finite Difference Time Domain (FDTD), Finite Integration Technique (FIT), Finite Element Method (FEM), Boundary Element Method (BEM), and Method of Moments. When applicable, commercial tools in each category are listed in alphabetical order by manufacturer name.
Regarding commercially available simulation methods, FDTD and FIT methods (Microwave Studio, Sim4Life, xFDTD) most often use a rectilinear grid and can simulate high-resolution multi-tissue anatomical models with relatively little computational resources (memory and time), whereas FEM (HFSS), BEM (COMSOL), and MoM (Feko), not being constrained to rectlinear representations, can simulate conductive surfaces in coils, shields, and implants with arbitrary orientation, but generally have limitations with the anatomical detail that can be simulated with reasonable computational resources. Between FEM, BEM, and MoM, generally FEM has higher memory requirements for a given problem, but also better ability to adapt to non-sparse problem geometries, including human body models with a surface-based format.
Regarding coil driving methods, if coil current and/or voltage distributions are well understood and the goal of the simulations is to understand field distributions in the sample or patient, directly implementing those current or voltage distributions (especially with sources across gaps where capacitors would be placed in the physical coil) can save significant time. On the other hand, if the purpose of the simulation is more about designing the coil or array to perform in a desired manner or to relate the forward power from the amplifier to the power absorbed in the subject, modeling the coil circuitry very exactly becomes important. Depending on the number of electrical components (especially capacitor values) to be optimized, simulating the coil with multiple ports in the place of all capacitors followed by use of a circuit co-simulator can be very valuable.
Often a major consideration regarding time and memory requirements for a simulation, as well as the storage space for results, is the spatial resolution of the grid or mesh used. It often takes a significant amount of experience with a software package to design efficient grids or meshes. When it comes to evaluating the results of a simulation, while the various available software packages include a differing degree of tools for analysis relevant to MRI (sometimes including temperature simulation tools or even a Bloch-based MRI simulator), often advanced analysis is better accomplished in other environments, such as with Matlab. In these cases, the ability to easily export data for use in another environment is an important feature.
Besides the commercial packages listed above, a number of freely-available options exist (with the understanding that ease of use – especially regarding GUIs for setting up simulations) exist. Limited versions of some of the commercial packages are also available for use at academic institutions free of charge.

Acknowledgements

No acknowledgement found.

Synopsis

Introduction:

Consideration of Commonly-used Simulation Methods and Settings for MRI:

Acknowledgements

References