University of Minnesota
Developing predictive models to optimize patient-specific deep brain stimulation therapy for Parkinson’s disease
Abstract
dc:description.abstractDeep brain stimulation (DBS) is an effective neurosurgical procedure for treating movement disorders such as Parkinson’s disease (PD). However, clinical outcomes of DBS therapy on the axial motor signs of PD, including gait dysfunction, are highly variable. This variability stems from a combination of factors including (1) long therapeutic time constants leading to challenges with optimizing stimulation settings during a clinical visit, (2) lack of understanding of which specific axonal pathways that, when modulated, induce a therapeutic effect on axial motor signs. Here, we developed patient-specific computational models and applied optimization algorithms to identify theaxonal pathway targets that underlie changes in gait dysfunction with DBS in the subthalamic nucleus (STN). In building these models, we first investigated the neural pathways implicated in the effects of DBS on reducing resting tremor, arresting stimulation-induced dyskinesias, and avoiding changes in mood with STN-DBS. We then further expanded the computational models to include seven distinct neural pathways in and around the STN in a retrospective analysis of changes in step length in PD patients with pronounced gait dysfunction (n=20 leads). Increased step length (therapy) was associated with activation of pedunculopallidal pathway, while decreased step length (anti-therapy) was associated with pallidopeduncular pathway. These results suggested a differential effect on the pedunculopontine nucleus depending on stimulation settings with antidromic excitation (orthodromic inhibition) leading to positive (negative) effects on step length. We further conducted a prospective study (n=14 leads), in which we applied a multi-objective particle swarm optimization algorithm to differentially target the pedunculopallidal axis with DBS and compare clinical outcomes on step length in the same subjects. Again, targeting the pedunculopallidal pathway was significantly better in improving step length than targeting the pallidopeduncular pathway. However, there was no significant difference between targeting the motor region of the STN compared to the pedunculopallidal pathway on step length. Together, these results show that optimization algorithms coupled with personalized computational models of DBS can help to more selectively stimulate specific pathways in the brain and improve our understanding of their role in clinical outcomes.
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Goftari, Mojgan
Subjects
dc:subject × 6Rights
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Citation
- Goftari, Mojgan. (2022). Developing predictive models to optimize patient-specific deep brain stimulation therapy for Parkinson’s disease. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/280271.