K Wake, N Varsier, S Watanabe, M Taki1, J Wiart, S Mann, I Deltour, and E Cardis (2009). The estimation of 3D SAR distributions in the human head from mobile phone compliance testing data for epidemiological studies. Phys. Med. Biol. 54: 5695–5706.
Several studies have evaluated the relationship of mobile phone use and brain tumours, where it is plausibly expected that the more strongly exposed location in the brain is likely to be the location of the tumor if a causal relationship really exists. Including exposure gradients at the location of the tumour is expected to improve exposure assessment. Given the difficulties of actual exposure measurements from multiple phones in an individual’s past, this study proposed a method to estimate the 3D distribution of the specific absorption rate (SAR) in the human head using existing non-3D compliance measurements for mobile phones.
Initial data was acquired in a phantom for compliance testing of mobile phones. SAR distributions have been measured for each model of phone since the late 1990s with a standard procedure. The data are not measured in the whole head as 3D data. Compliance testing allows for measurements made for the limited part of the head in the region and small volume around the location of the maximum SAR. Here, interpolation and extrapolation were employed and SAR distributions in the whole head were estimated from the limited measured data. Two models were used to estimate 3D SAR distributions, one being the head part of the TARO model based on MRI data of an averaged sized Japanese male, and the other being the Gridmaster model based on MRI data of a French male. One straight-type phone and one flip-type phone in an 800 MHz band were used. Extrapolation was based on an assumed exponential decay factor.
The method was validated both experimentally and numerically. In the former, the authors compared the inter-/extra-polated 3D SAR distribution estimated from measured compliance testing data by the proposed method with the SAR distribution obtained by actual 3D measurements for the same phantom model. Correlation coefficients between the estimated and measured SARs were 0.99 and 0.98 for the phones. Numerical validation consisted of comparing SAR distributions between numerically calculated 3D SAR data and the estimated data with the proposed method, including conversions for the shape of the head. The estimated SAR values in the whole brain and higher SAR region of the brain, which are most important for SAR estimation in this study, showed a strong correlation of more than 0.9 with the directly calculated data but tended to be somewhat underestimated. The proposed method was confirmed to be reasonable to estimate 3D SAR distributions from the limited data obtained from the compliance testing measurements. Regression coefficients for estimated SAR to calculated SAR were close to 0.75.
Discussion and conclusion
It was confirmed that the proposed method provided good and reasonable estimation of 3D SAR distribution in the head, and especially in the brain, which is the tissue of major interest in epidemiological studies, using the limited data obtained from standard compliance testing. The authors suggest that such information can be used to improve the relevance of exposure indices in epidemiologic studies.