252 Dual Device Beat Frequency Artifact Impairs Local Field Potential Analysis
Nabeel Diab, Nicole R. Provenza, Nisha Giridharan, Anthony Kaspa Allam, Garrett P. Banks, Sameer Vikram Rajesh, Eric Storch, Wayne K. Goodman, Jeffrey A. Herron, Sameer Anil Sheth · Neurosurgery · 2025
INTRODUCTION: Beat frequency artifacts (BFAs) previously described in dual sensing-enabled DBS systems are high amplitude contaminations of underlying neural signals that occur at regular intervals. The analysis of experiments with associated local field potential (LFP) data depends on reliable neural data across trials. BFAs risk compromising time frequency and spectral analysis of any experiment. METHODS: We developed a wavelet convolution pipeline to average neural responses across 120 trials of a task in which provoking or neutral images are presented on a screen followed by a patient response. Trial epochs containing visible BFA amplitude deviations were grouped and averaged separately. RESULTS: We found an average of 14% of trials in the provocation task contained some BFA amplitude deviation across patients (n=5). Conventional outlier removal of trials containing spectral power greater than 3 standard deviations from the mean failed to remove all contaminated trials in any of the patients. These contaminated trials varied in relative spectral power and timing from trials without any BFA present. CONCLUSIONS: In a cohort of 5 patients with dual DBS systems, all had contaminated LFP recordings during an experimental task. More rigorous statistical approaches are needed to remove these epochs of data from large datasets.