Abstract
BACKGROUND: Deep brain stimulation (DBS) of the ventral intermediate nucleus (VIM) of the thalamus (TH) is an effective therapy for suppressing tremor. One of the critical challenges to optimizing VIM-DBS therapy is the lack of robust neural biomarkers that correlate well with tremor. OBJECTIVE: To quantify biomarkers for tremor using intraoperative TH local field potential (LFP) recorded from DBS electrodes. Further, we used computational modeling to understand the biophysical basis of the recorded LFP signal. METHODS: We simultaneously recorded intraoperative TH LFP and tremor from the hand dorsum (32 participants) and during DBS at different frequencies (16 participants). Then, we simulated the effects of DBS and spatial distribution of tremor cells on calculated LFPs in a TH model. RESULTS: There was a moderate correlation between tremor and LFP spectral power in the theta and alpha bands (r = 0.445 and 0.389, respectively). There was a strong correlation between tremor and peak coherence between LFP and tremor signal (r = 0.559). Postural tremor was decoded from the LFP signal with an area under the curve of ∼0.7. High frequency DBS reduced spectral power in the theta and alpha bands and tremor could be decoded from the LFP spectral power in the presence of DBS (0.429 goodness of fit R(2)). The theta power in the simulated LFP signal varied substantially with the specific location of the bipolar contact pair of the DBS electrode used for the LFP recordings as well as the spatial distribution of tremor cells. CONCLUSIONS: Theta power alone was not sufficient for prediction of tremor control. Simulations indicated that the number and distribution of tremor cells surrounding the DBS lead may explain the lack of a strong correlation between tremor and theta power.