Parkinsonian rest tremor can be detected accurately based on neuronal oscillations recorded from the subthalamic nucleus

•Tremor can be detected accurately in short segments of STN local field potential recordings.•Four power estimates from a single DBS electrode contact pair suffice for successful detection.•High frequency oscillations are the most useful feature for tremor detection.

ObjectiveTo investigate the possibility of tremor detection based on deep brain activity.MethodsWe re-analyzed recordings of local field potentials (LFPs) from the subthalamic nucleus in 10 PD patients (12 body sides) with spontaneously fluctuating rest tremor. Power in several frequency bands was estimated and used as input to Hidden Markov Models (HMMs) which classified short data segments as either tremor-free rest or rest tremor. HMMs were compared to direct threshold application to individual power features.ResultsApplying a threshold directly to band-limited power was insufficient for tremor detection (mean area under the curve [AUC] of receiver operating characteristic: 0.64, STD: 0.19). Multi-feature HMMs, in contrast, allowed for accurate detection (mean AUC: 0.82, STD: 0.15), using four power features obtained from a single contact pair. Within-patient training yielded better accuracy than across-patient training (0.84 vs. 0.78, p = 0.03), yet tremor could often be detected accurately with either approach. High frequency oscillations (>200 Hz) were the best performing individual feature.ConclusionsLFP-based markers of tremor are robust enough to allow for accurate tremor detection in short data segments, provided that appropriate statistical models are used.SignificanceLFP-based markers of tremor could be useful control signals for closed-loop deep brain stimulation.