Xi Lin Chen, Stefan Benkler, Nicholas Chavannes, Valerio De Santis, Jurriaan Bakker, Gerard van Rhoon, Juan Mosig, and Niels Kuster, Bioelectromagnetics, Volume 34, Issue 5, pp. 375-384, July 2013, online February 12, 2013
The objective of this paper is to investigate the dependency of dosimetric compliance assessments on applied methodology and anatomical modeling for human brain tissues exposed to low-frequency (LF) magnetic fields. This study has addressed several critical open questions pertaining to the LF numerical dosimetry of B-field induction, e.g., the impact of discretization errors on the spatially averaged E-fields, the effect of tissue conductivity contrast on peak E-field intensity, and the consistency and applicability of the ICNIRP and IEEE exposure limits.
The scientific and technical impact of this study can be summarized as:
- Frequency scaling should be applied with caution when significant variations in tissue conductivity contrast with respect to frequency shifts are expected.
- The peak induced E-field derived in a homogeneous head model is not conservative enough to represent the worst-case induced E-field in the human brain.
- Reference levels of the IEEE C95.6  standard can be in contradiction with the basic restrictions by a factor of more than 5, and revision of the limits is recommended.
- An arbitrary filtering approach, such as the 99th percentile value (ICNIRP 2010), could lead to significant underestimation of the peak E-field and may significantly impair the reproducibility of exposure assessments.
- Spatial averaging with a grid size of 0.5 mm or smaller sufficiently reduces the impact of artifacts regardless of tissue size, and it is, therefore, recommended to replace the 99th percentile with such a requirement.