Development of model-based and sensor-based programming techniques for optimizing directional deep brain stimulation therapy for movement disorders

Abstract

Deep brain stimulation (DBS) therapy is a programmable neurosurgical intervention that can significantly improve quality of life for individuals with medication-refractory movement disorders, such as Essential Tremor and Parkinson’s Disease. However, clinical outcomes with DBS therapy still vary across patients, and the clinical time and effort necessary to program the stimulation settings to each patient’s symptoms presents practical challenges in the clinic. With the advent of directional lead technology and independent multi-channel current-controlled stimulation, the scope of possible DBS configurations is now substantially larger than it was even five years ago. This has greatly increased the time to determine the most effective electrode configuration, and in reality, much of the stimulation parameter space is left unexplored during a clinical visit. This thesis addressed the gap between the directional lead technology and its clinical implementation by developing three promising techniques to program directional DBS lead systems. The first programming technique involved developing subject-specific computational models of DBS based on individual MRI/CT scans. Comparing model predictions to clinical outcomes from patients with Essential Tremor revealed that lateral and medial parcellations of the motor-thalamic afferents of the cerebellothalamic tract were differentially associated with stimulation-induced therapy and side effects, respectively. Second, sensor-based evaluation of DBS in Essential Tremor patients revealed that directional contacts were superior to ring-mode contacts in providing optimized tremor reduction with reduced dysarthria. The third programming technique involved using neurophysiological feedback to guide the selection of which electrode(s) to use during DBS. In Parkinson’s disease, for example, stimulation through electrodes with higher resting-state beta-band oscillatory power in the subthalamic nucleus generally results in better clinical outcomes. Using a non-human primate model, we tracked how beta-band power changed spatially and temporally between intraoperative and chronic time points and showed that the strongest variability occurred within the first two weeks after lead implantation. This suggested that neurofeedback-based programming may be most consistent after the immune tissue response settles. Together, these results showed how model- and sensor-based programming techniques can limit the parameter space for programming directional DBS enabling more efficient and effective clinical outcomes in the future.