Adaptive feedback control in deep brain stimulation: a simulation study

Deep brain stimulation (DBS) is an effective electric therapy to treat movement disorders associated with chronical neural diseases like essential tremor, dystonia and Parkinson's disease. In spite of a long clinical experience, the cellular effects of the DBS are still partially unknown because of the lack of information about the target sites. Recent studies, however, have proposed the local field potentials (LFPs) in the targets as a useful tool to study the behavior before and after stimulation [Priori et al., 2006]. Our work investigates the relationship between DBS settings and LFPs in a detailed simulator of the electric activity in the Vim (one of the preferred surgical targets) under tremor conditions. A least-square approach is adopted to identify a functional, input-output ARX model structure for the Vim and evaluate the effects of the stimulation on its electric patterns. Based on it, an adaptive minimum variance control scheme is then proposed to restore the spectral features of the Vim's LFPs to reference values, i.e., as in subjects not affected by movement disorders. Results indicate good performances in tracking the reference spectral features through selective changes in the low (2-7 Hz), α (7-13 Hz) and β (13-35 Hz) ranges.