layout: page —

Deep Brain Stimulation Programming 2.0:\

Future Perspectives for Target Identification and Adaptive Closed Loop Stimulation
Franz Hell, Carla Palleis, Jan H. Mehrkens, Thomas Koeglsperger, and Kai Bötzel
Doi: 10.3389/fneur.2019.00314

Summary

The most of the article is dedicated to a comprehensive overview over recent research on the target structures and targeting strategies in Deep Brain Stimulation (DBS). The second part focuses on on closed-loop stimulation and associated biofeedback signals. (The following summary covers only the latter topic)

Problem statment. Open-loop DBS are prone to reduction of efficiency over time due to chanhes in patients’ condition. Closed loop adaptive DBS involves two distinctive parts: feedback signals (biomarkers) and mechanisms of control. The fisrts step is necessay to describe relation to a clinical symptom. Adaptive control mechanisms are used for adjustment of a stimulation.

Biomarkers

  1. Electrophysiological measurements. Eg. Local field potentials (LFP) which use beta frequency amplitude as a mechanism to trigger the stimulation; Electromyography; Kinematic sensors; ECoG

  2. Neurochemical sensing. Recently several devices were developed: detectors wich sence changes in dopamine concentration in rodents; WINCS Harmoni®: implanted neurochemical sensor which measures neurotransmitter concentration. One should notice, that these devicesn have een used only in preclinical DBS studies!

  3. External mechanistic sensors Accelerometers, EMG sensors. State-of-the-art devices incorporate several sensors. E.g. a smart glove contains two touch sensors, two 3D-accelerometers and a force sensor to assess tremor, rigidity and bradykinesia of hand and arm [https://ieeexplore.ieee.org/abstract/document/5999113]. (ToDo: add Heldman et al. [https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4760891/], [https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4372473/])

Control Mechanisms

  1. Beta threshold targeting. Stimualtion is turned on, when the amplitude of oscillatory activity in the β-band exceeds a certain threshold, or only when pathological long β bursts are detected. But, unfortunately, beta oscillations and beta oscillatory characteristics are sensitive to medication and patient’s behavior.

  2. Noise cancellation. This control mechanism is based on the effect of noise cancellation: when specific phases of the tremor sensed by an accelerometer attached to a hand of a parkinsonian patient, the stimulation applied on the talamus.

  3. Stimulation on demand. E.g. cortical electrodes sense β-band desynchronization in patients with essential tremor in the beginning of a movement. Then the stimulation applied only in a movevent, but stays tirned off during restinf states. However, this approach reduces tremor only with a delay, as it is not capable (yet) to predict a movement before it occurs.

  4. Coordinated reset stimulation. The method implies brief high-frequency pulse trains when, specific structures known to produce clinical symptomth produce rhythmic synchronized oscillations. This leads to reduction of symptoms such as tremor within patients with Parkinson desease.

Future Perspectives

  • Learning disease symptoms severity and underlying neural activity with ML.
  • Integrating parameters for ML from different kinds of biomarkers.
  • Decoding and predicting patiens neural and clinical states.
  • Application of reinforcment learning to learn and control stimulation.
  • Inferring the causal structure of brain networks to reduce amount of time and cost needed to gather enough data for ML. (Methods such as Granger causality , dynamic causal modeling, structural equation modeling, and causal Bayesian networks have already showed their effectivness in revealing neronal dynamics of several brain regions).