Laakso I. 2009. Assessment of the computational uncertainty of temperature rise and SAR in the eyes and brain under far-field exposure from 1 to 10 GHz. Phys Med Biol. 54(11):3393-404.
Morphological differences in computational models are an important source of uncertainty and variations in computation of specific absorption rate (SAR) distribution and temperature rise under exposure to electromagnetic fields.
The focus of this study was to assess the magnitude of uncertainty related to grid resolution of the computational model.
In this study, finite-difference time-domain (FDTD) simulations of SAR values in the human head were performed. Plane-waves were used as the source of exposure. High 0.5 mm resolution adult male and female voxel models were applied, and simulations were performed for frequencies in the range 1 GHz -10 GHz. Computations of temperature rise due to power absorption were performed by the bioheat equation using a multigrid method solver. Calculated cubically averaged 10 g SAR in the eyes and brain and eye-averaged SAR were compared with the corresponding temperature rise and exposure guidelines. The computational accuracy was assessed by comparing the results of calculations with resolutions of 0.5mm, 1 mm and 2mm.
Results and interpretation
It was shown that results for 2 mm resolution at frequencies below 2.5-3 GHz differed from 0.5 mm results by less than 10%. These differences were smaller for 1 mm resolution up to the frequency of 5 GHz. At frequencies higher than 5 GHz the results diverged from the 0.5 mm results. Uncertainty in SAR and temperature values due to resolution (typically smaller than ±10%), though were not negligible, were generally smaller than uncertainty from other possible sources, such as morphological differences between the models. Incident power density smaller than 100 W m-2 ensured that the temperature rise in the lens of the eye was less than 1°C in the whole frequency range studied. The temperature rise in the brain in all the studied cases was less than 0.6°C.
The results suggest that 2 mm resolution should be used for frequencies below 2.5 GHz and 1 mm resolution – below 5 GHz. Variations related to morphological differences in models are greater than those related to grid resolution. The ICNIRP reference levels and the IEEE maximum permissible exposure limits are conservative in that the temperature rise in the eye and in the brain at these levels was less than 1°C.