Fig. 2.

Schema for analysis technique used to remove the stimulus artifact. Individual DBS pulses are identified by a voltage amplitude-dependent peak detection algorithm. Note the reproducibility of the stimulus artifact in the superimposed individual traces (A1). The stimulus artifacts identified during peak detection are averaged together into an evoked potential (A2). This evoked potential is then subtracted from each occurrence of the stimulation pulse in the raw data (B1 and B2). Because of the large amplitude and variability of the initial sharp transient in the stimulus pulse, subtraction does not completely eliminate this portion of the stimulus artifact, so the raw data is set to zero or flattened during this brief interval. C and D: perineuronal analog histograms from 2 PD subjects with contralateral STN DBS in various stimulation conditions. The histograms are centered temporally on every occurrence of an individual neuron. Note that the morphology of the action potentials in the DBS off conditions and in the deartifacted data are identical, and that during both high (160 Hz) and low frequency (30 Hz) stimulation epochs, the stimulus artifact (gray arrows) is seen at a short latency (∼1 ms) before the action potential waveform (black arrows).