Abstract-In a previous paper, we proposed and tested a robust and efficient three-dimensional (3-D) subgridding algorithm for the FDTD solution method of the Maxwell's curl PDEs system. Its characteristic feature is the straight, non-recursive, embedding of Yee grids -refined by factors of 3, 5, 7 and even larger -within coarser ones. There, the algorithm's implementation was described with the traditional serial programming approach. In the present paper, we propose and test its parallel programming implementation. The goal is to make it suitable and efficient for large scale electromagnetic simulations.

No abstract

Realistic numerical models of human subjects and their surrounding environment represent the basic points of radiofrequency (RF) electromagnetic dosimetry. This also involves differentiating the human models in men and women, possibly with different body shapes and postures. In this context, the aims of this paper are, firstly, to propose a female dielectric anatomical model (fDAM) and, secondly, to compare the power absorption distributions of a male and a female model from 0.1 to 4 GHz. For realizing the fDAM, a magnetic resonance imaging tomographer to acquire images and a recent technique which avoids the discrete segmentation of body tissues into different types have been used. Simulations have been performed with the FDTD method by using a novel filtering-based subgridding algorithm. The latter is applied here for the first time to dosimetry, allowing an abrupt mesh refinement by a factor of up to 7. The results show that the whole-body-averaged specific absorption rate (WBA-SAR) of the female model is higher than that of the male counterpart, mainly because of a thicker subcutaneous fat layer. In contrast, the maximum averaged SAR over 1 g (1gA-SAR) and 10 g (10gA-SAR) does not depend on gender, because it occurs in regions where no subcutaneous fat layer is present.

This paper presents an automatic method to obtain tissue complex permittivity values to be used as input data in the computer modelling for hyperthermia treatment planning. Magnetic resonance (MR) images were acquired and the tissue water content was calculated from the signal intensity of the image pixels. The tissue water content was converted into complex permittivity values by monotonic functions based on mixture theory. To obtain a water content map by MR imaging a gradient-echo pulse sequence was used and an experimental procedure was set up to correct for relaxation and radiofrequency field inhomogeneity effects on signal intensity. Two approaches were followed to assign the permittivity values to fat-rich tissues: (i) fat-rich tissue localization by a segmentation procedure followed by assignment of tabulated permittivity values; (ii) water content evaluation by chemical shift imaging followed by permittivity calculation. Tests were performed on phantoms of known water content to establish the reliability of the proposed method. MRI data were acquired and processed pixel-by-pixel according to the outlined procedure. The signal intensity in the phantom images correlated well with water content. Experiments were performed on volunteers' healthy tissue. In particular two anatomical structures were chosen to calculate permittivity maps: the head and the thigh. The water content and electric permittivity values were obtained from the MRI data and compared to others in the literature. A good agreement was found for muscle, cerebrospinal fluid (CSF) and white and grey matter. The advantages of the reported method are discussed in the light of possible application in hyperthermia treatment planning.

Complex permittivity values have a dominant role in the overall consideration of interaction between radiofrequency electromagnetic fields and living matter, and in related applications such as electromagnetic dosimetry. There are still some concerns about the accuracy of published data and about their variability due to the heterogeneous nature of biological tissues. The aim of this study is to provide an alternative semi-automatic method by which numerical dielectric human models for dosimetric studies can be obtained. Magnetic resonance imaging (MRI) tomography was used to acquire images. A new technique was employed to correct nonuniformities in the images and frequency-dependent transfer functions to correlate image intensity with complex permittivity were used. The proposed method provides frequency-dependent models in which permittivity and conductivity vary with continuity--even in the same tissue--reflecting the intrinsic realistic spatial dispersion of such parameters. The human model is tested with an FDTD (finite difference time domain) algorithm at different frequencies; the results of layer-averaged and whole-body-averaged SAR (specific absorption rate) are compared with published work, and reasonable agreement has been found. Due to the short time needed to obtain a whole body model, this semi-automatic method may be suitable for efficient study of various conditions that can determine large differences in the SAR distribution, such as body shape, posture, fat-to-muscle ratio, height and weight.