Individualization of a surrounding tissue model in deep brain stimulation

Deep Brain Stimulation is an established therapy that consists of sending mild electrical pulses to the brain via a surgically implanted electrode. It is used to treat the symptoms of neurodegenerative diseases such as Parkinson's Disease and Essential Tremor. The stimulation effect is however patient-specific and difficult to predict. Further, impedance measurements indicate changes in the medium around the lead over time that are generally challenging to account for. Although mathematical models to characterize the extent of the stimulation have been developed in recent years, it is imperative that the model parameters are realistic enough to produce correct results. This study aims at developing a system identification approach to capture the changes in the medium around the lead by measuring the potential on non-active contacts during stimulation. Due to infeasibility of assigning properties to the medium around the lead in vivo, synthetic data are used for analysis instead. Results suggest that a transfer function with one zero and one pole that corresponds to a circuit composed of two RC contours in cascade is sufficient to describe the measurements in silico. Furthermore, the identified parameters show an injective relation to the properties of the medium. Continuous Least Squares and Laguerre domain identification are utilized for obtaining parameter estimates.