Electromagnetic Modeling and Sensitivity-Based Optimization of Medical Devices

Electromagnetics is a fundamental part of biomedical engineering and modern healthcare due to the electromagnetic nature of several important processes in the human body and the interactions of electromagnetic fields with the human body. As a consequence, electromagnetics is exploited for diagnostic and therapeutic purposes by a multitude of medical devices. The biomedical engineering society strives to develop and design new methods, as well as, to improve existing methods for diagnosis and therapy. In addition, electromagnetic compatibility of both electromagnetic and non-electromagnetic medical devices must be assessed. These tasks can be complicated since electromagnetic measurements in the human body can be difficult and in some cases harmful to the patient. Furthermore, the human body is highly heterogeneous, which makes predictions of its interaction with electromagnetic fields demanding. In this thesis, these problems are mitigated by means of accurate, unbiased, and automatized electromagnetic modeling that feature a number of disciplines: (i) detailed electromagnetic modeling based on Maxwell’s equations; (ii) mathematics with particular emphasis on numerical analysis and optimization; and (iii) large-scale parallel computations on computer clusters. Progress in these three areas enables larger and more difficult problems to be addressed. In particular, this methodology is applied to three biomedical problems in this thesis. First, the electromagnetics of pacemaker lead heating in MRI is modeled with emphasis on the multi-scale characteristic of the problem. The results show the resonant nature of the problem and that detailed modeling is essential to accurately describe this phenomenon. Second, a method for optimization of sensor positions in magnetic tracking systems is proposed. The method uses powerful mathematics to alleviate the difficulties and computational burden associated with experimental or computational trial-anderror procedures. Third, the estimation procedure in EEG-based source localization is facilitated by exploiting electromagnetic reciprocity during the modeling. This reduces the demands for tailored estimation procedures and removes one obstacle for real-time source localization.