Comparison of SAR calculation algorithms for the finite-difference time-domain method

Abstract

Finite-difference time-domain (FDTD) simulations of specific-absorption rate (SAR) have several uncertainty factors. For example, significantly varying SAR values may result from the use of different algorithms for determining the SAR from the FDTD electric field. The objective of this paper is to rigorously study the divergence of SAR values due to different SAR calculation algorithms and to examine if some SAR calculation algorithm should be preferred over others. For this purpose, numerical FDTD results are compared to analytical solutions in a one-dimensional layered model and a three-dimensional spherical object. Additionally, the implications of SAR calculation algorithms for dosimetry of anatomically realistic whole-body models are studied. The results show that the trapezium algorithm-based on the trapezium integration rule-is always conservative compared to the analytic solution, making it a good choice for worst-case exposure assessment. In contrast, the mid-ordinate algorithm-named after the mid-ordinate integration rule-usually underestimates the analytic SAR. The linear algorithm-which is approximately a weighted average of the two-seems to be the most accurate choice overall, typically giving the best fit with the shape of the analytic SAR distribution. For anatomically realistic models, the whole-body SAR difference between different algorithms is relatively independent of the used body model, incident direction and polarization of the plane wave. The main factors affecting the difference are cell size and frequency. The choice of the SAR calculation algorithm is an important simulation parameter in high-frequency FDTD SAR calculations, and it should be explained to allow intercomparison of the results between different studies.

On the Use of Conformal Models and Methods in Dosimetry for Nonuniform Field Exposure

“…The goal was to evaluate the induced field strength around the surface for localized or nonuniform exposure. A dielectric sphere of radius 8 cm centered at the origin is exposed to a magnetic dipole source The analytical approach to analyze a multilayer sphere with arbitrary isotropic material parameters (ε and σ) exposed to magnetic and/or electric dipoles based on Mie theory provides a full-wave solution and works at any frequency (previously confirmed at both 50 Hz and in the gigahertz range [37]). …”

SAR exposure from UHF RFID reader in adult, child, pregnant woman, and fetus anatomical models

“…As a final remark, one should note that besides the uncertainty due to the dielectric properties, there are several other factors intrinsic to the computational simulation that can influence the reliability of the estimated SAR values. These sources of error are of various origins, mainly related to numerical limitations of the modeling of the source and anatomical body, discretization, resolution of the FDTD grid at a specific frequency, and choice of the algorithm for SAR calculation [Huber et al, ; Kuster et al, ; Bakker et al, ; Laasko et al, ]. According to Bakker et al [], the worst‐case total expanded uncertainty of the numerical modeling with a 95% confidence interval ( k = 2) can reach 58% for the psSAR 10 g .…”

New Algorithms for the Specific Absorption Rate Numerical Evaluation Based on Spherical Averaging Volumes

“…As a matter of fact, the number of available algorithms for the numerical evaluation of SAR is quite large [9][10][11][12][13][14][15][16][17] and in this section, a brief overview on the most commonly adopted ones is presented and commented. One of the usually adopted approaches [11,12], for instance, computes the SAR rm on a certain point by considering the contributes coming from the cells which belong to a cube centered in that point.…”