Bio-electromagnetic model of deep brain stimulation

Functional stimulation is one of the most fascinating applications of bioelectromagnetism. It deals with the stimulation of excitable biological tissues by electromagnetic fields. One of its most impressive medical applications is the subthalamic nucleus deep brain stimulation (DBS). It consists in the insertion of an electrode into the deep brain, delivering electric pulses to treat Parkinson's disease and other movement disorders. But despite its wide use throughout the world for almost twenty years, the understanding of the mechanisms of action remains unclear. To help clinicians to better understand the mechanisms of DBS, its limitations and implications from an electrical point of view, electrical models of the head can be used to predict the electric potential distribution generated by the electric pulse. With the development of medical imaging techniques, the information on biological tissues that can be used to build these electrical models has never been so detailed. The diffusion tensor magnetic imaging (DT-MRI) is able to provide the orientation of the fibers within the cerebral tissues. Thus, the high inhomogeneity and anisotropy of the head can be modeled through anisotropic electrical conductivity tensors to set up realistic models of the patient's head. This thesis aims to provide to clinicians an accurate prediction of the potential distribution generated by the electric pulse. With this purpose, a finite element (FE) model is set up using electric conductivity values based on DT-MRI data. Special care has been taken to model more realistic boundary conditions than the ones commonly encountered in literature. A great effort has also been put to model the tissues surrounding the stimulation. The results show that these two aspects are impacting significantly the potential distribution. To predict the neural extent of the stimulation, electrical equivalent models of axons are combined with the obtained potentials. Volume of tissues activated (VTA) are thus obtained. Results show that the VTA are also impacted by the decision on how to model the boundary conditions. They show that the usual choice assumed in literature up to now leads to an overestimation of 30% of the VTA.