Numerical electromagnetic techniques for dosimetry and biomedical applications

(EM) fields are increasingly being used for many new and rapidly expanding applications. Along with the growing applications of EM fields, increasingly sophisticated modeling, experimental and modeling techniques have been developed for evaluating and understanding the biological interaction of human tissues and body with EM fields. In this dissertation, we expanded the use of these EM techniques to some new and novel dosimetric and biomedical applications in the following four subjects: (1) We developed a broadband waveguide-based experimental setup and numerical procedures using FDTD method to determine the peak 1- and 10-g SARs, and designed a validation system for the SAR measurement system and/or for E-field probe calibration for the 802.11a frequency band 5.15 to 5.825 GHz. (2) We presented a numerical procedure for compliance testing of EAS systems with ferromagnetic cores, which utilizes the duality of magnetic and electric circuits and a procedure based on three-dimensional (3-D) admittance method to calculate the magnetic fields exposure of such devices. (3) We solved the bioheat equation for anatomically based thermal models of the human heads to determine the heating of various tissues due to EM radiation of typical cellular telephones. Also investigated are the thermal implications of the SAR limits suggested in the various safety guidelines and the relationships between brain temperature elevations and the various 1- and 10-g SARs. (4) MEG and its inversion techniques have a great potential in the noninvasive characterization of brain function and activities. Many reconstruction approaches employ ECDs (Equivalent Current Dipoles) for modeling highly localized stimuli-induced neural current sources in the brain and head shapes modeled with single or multiple concentric spheres for the forward problem. In this thesis, we used a shaped head model and 3-D impedance method to calculate the forward magnetic fields so that the inversion algorithm is based on a more accurate lead field matrix. A statistically-averaged approach of Tikhonov regularization and FOCUSS (FOCal Underdetermined System Solution) iterations were applied to forward-calculation "measurement" data corrupted by high level Gaussian noise (Signal-to-Noise Ratio as low as 2), to locate two to four possible ECDs in the brain with high accuracy.