The Preliminary Application Of Pseudospectral Time-Domain Method In EMF Bioeffect Research

As electronic instruments that work at low frequencies or radio frequencies arewidely used in industries, telecommunications, and medical treatments etc., more andmore people are exposed to electromagnetic fields. The bioeffect of electromagneticfields on human body begins to draw attention from the biomedical society. Directexperimental research of EMF bioeffect on human body is generally difficult, so thecomputer modeling and simulation of EMF bioeffect becomes an important way ofdoing such researches, in which the electromagnetic model of human body andnumerical methods for electromagnetic fields are the basic tools.The finite difference time-domain (FDTD) method is stable and easy to be used, but due to the fact that it approximates the space derivatives in Maxwell's curlequations by central difference, FDTD usually requires more than 20 grids perwavelength to achieve acceptable accuracy. As a contrast, pseudospectraltime-domain (PSTD) method approximates the space derivatives by pseudospectralmethod, which guarantees the accuracy, theoretically requiting only 2 grids perwavelength. In this thesis, the numerical results obtained by FDTD and PSTDrespectively are compared, which have verified the advantage of PSTD in accuracywhen the number of grid per wavelength is small, and the disadvantage of itscomputation speed when the number of grid is same as in FDTD. The unwanted factthat PSTD requires its excitation source be specially modified is also pointed out. As away to accelerate the PSTD method, an algorithm of using real-to-real FFT/IFFT torealize pseudospectral method is proposed, and the result shows that the newalgorithm could save 50% of computation time.In the simulation of magnetic fields generated by gradient coils of MRI system, both FDTD and PSTD are applied. The results show that, in this problem, where thefrequency is relatively low and the geometric structure of excitation source iscomplicated, the traditional FDTD method yields better results than the PSTD methoddoes, either the computation speed or the accuracy. In the following simulation of induced current on human body, the results show that FDTD can achieve convergencestably, but PSTD cannot, and further more PSTD needs much more time to finisheach time step. The reason why PSTD cannot achieve convergence in this experimentis analyzed and discussed. In the last part of numerical simulation, PSTD is applied tocompute the induced electric fields and SAR on human body when the excitationsource is an electric dipole, which can be specially modified in PSTD. The resultsshow that, PSTD can achieve convergence in this problem, where the excitationsource is relatively simple and can be specially modified. PSTD has some advantagesin this kind of problems where the frequency of excitation source is very high and itsgeometric structure is simple, when only the coarse numerical results are needed.These simulation experiments demonstrate: in the problems where thewavelength is far longer than the size of computation region, FDTD always yieldsbetter results than PSTD does, because it is faster and the excitation source requiresno special treatment. The problem of simulating induced current on human body bygradient coils falls into this category. For electrically large problems, where thewavelength is far smaller than the size of computation region, PSTD can be applied toobtain the coarse results, provided that its excitation source can be specially modified.In this thesis, the problem of simulating the effect on human body excited by anelectric dipole belongs to this category.FDTD is a stable method and has been widely used in the EMF bioeffectresearch, while the application of PSTD has just begun and still faces manydifficulties. For wider applications of PSTD in this field, it must be taken intoconsideration that PSTD requires much less grid number as well as the difficulty ofexcitation source setup.