Parkinsonian Tremor Detection from Subthalamic Nucleus Local Field Potentials for Closed-Loop Deep Brain Stimulation

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Neuroscience

Background:

• Deep Brain Stimulation (DBS) is a common treatment for Parkinson's Disease (PD) symptoms.
• Current continuous DBS therapy presents challenges including side effects and high energy consumption due to fluctuating PD symptoms.
• There is a need for adaptive DBS systems that can adjust stimulation based on real-time patient needs.

Purpose of the Study:

• To investigate the efficacy of a logistic regression-based classifier for identifying rest tremor in PD patients using Local Field Potentials (LFPs).
• To explore the potential of machine learning for developing on-demand DBS therapy.
• To assess the feasibility of reducing side effects and energy consumption in DBS for PD.

Main Methods:

• Utilized Local Field Potentials (LFPs) recorded from Subthalamic Nucleus DBS electrodes in 7 PD patients.
• Applied a logistic regression classifier to analyze 36.1 minutes of data using 512 ms non-overlapping windows.
• Identified key frequency bands (31-45 Hz, 5-7 Hz, 21-30 Hz, 46-55 Hz, 56-95 Hz) with high discriminative power for tremor detection.

Main Results:

• The classifier achieved classification accuracy significantly above chance level across all patients.
• Area Under the Curve (AUC) values ranged from 0.67 to 0.93, indicating robust performance.
• Specific frequency bands in the LFP signals were identified as highly predictive of rest tremor.

Conclusions:

• A machine learning-based classifier can accurately detect PD rest tremor from LFP signals.
• This approach provides a foundation for developing on-demand DBS systems.
• On-demand DBS has the potential to optimize treatment efficacy, minimize side effects, and conserve battery power in PD patients.