Dynamic peripheral nerve stimulation can produce cortical activation similar to punctate mechanical stimuli

Abstract

During contact, phasic and tonic responses provide feedback that is used for task performance and perceptual processes. These disparate temporal dynamics are carried in peripheral nerves, and produce overlapping signals in cortex. Using longitudinal intrafascicular electrodes inserted into the median nerve of a nonhuman primate, we delivered composite stimulation consisting of onset and release bursts to capture rapidly adapting responses and sustained stochastic stimulation to capture the ongoing response of slowly adapting receptors. To measure the stimulation's effectiveness in producing natural responses, we monitored the local field potential in somatosensory cortex. We compared the cortical responses to peripheral nerve stimulation and vibrotactile/punctate stimulation of the fingertip, with particular focus on gamma band (30-65 Hz) responses. We found that vibrotactile stimulation produces consistently phase locked gamma throughout the duration of the stimulation. By contrast, punctate stimulation responses were phase locked at the onset and release of stimulation, but activity maintained through the stimulation was not phase locked. Using these responses as guideposts for assessing the response to the peripheral nerve stimulation, we found that constant frequency stimulation produced continual phase locking, whereas composite stimulation produced gamma enhancement throughout the stimulus, phase locked only at the onset and release of the stimulus. We describe this response as an "Appropriate Response in the gamma band" (ARγ), a trend seen in other sensory systems. Our demonstration is the first shown for intracortical somatosensory local field potentials. We argue that this stimulation paradigm produces a more biomimetic response in somatosensory cortex and is more likely to produce naturalistic sensations for readily usable neuroprosthetic feedback.

Keywords: biomimetic stimulation; gamma; perception; peripheral nerve stimulation; somatosensory feedback.

FIGURE 1

Stimulation patterns. Illustrated are the timing patterns associated with stimulation. The BB stimulation uses the 8th level of SP across the four levels of PP.

FIGURE 2

Analysis flow. Outline of the steps taken in analysis on a vibrotactile trial. (A) Transform evoked potentials into event related power, corrected by the standard deviation of the baseline power of each frequency, Z(t,f). (B) Use non-parametric analysis to determine significant time frequency points across stimulation conditions, Z_sig(t,f). (C) Calculate Phase Lock Value, PLV(t,f), across trials of each condition. (D) Determine time points of significant phase locking by calculating the Phase Lock Value Confidence Interval, PLV_CI. (E) Calculate the PLγ by isolating the average Z_γ (t) that is stimulation significant and PLV significant. (F) Calculate the NPLγ by isolating the average Z_γ (t) that is stimulation significant and not PLV significant. (G) Separate data into onset [–50 to 150 ms] and delayed [200 to 400 ms] epochs. (H) Integrate significant gamma states within onset and delayed time domains to compare to the perceptually appropriate standard.

FIGURE 3

Cortical significance to mechanical and peripheral nerve stimulation. Cortical array structure and representation illustrated in the top right. The N-Form cortical arrays consist of a 2 × 2 arrangement of probes, with seven recordings sites and a single shallow reference. Two arrays are implanted into somatosensory cortex, with direction defined in the figure. Included is an example response of the array to illustrate the medial-lateral and depth arrangement of the figure. On the left, significant cortical channels are shown for punctate, vibrotactile, and CP stimulation, but significant REF channels of the lateral array are indicated in gray. Below the cortical electrode maps, time-frequency significance masks are constructed by averaging the Boolean time-frequency mask (Figure 2B) for each significant channel. This shows common points of response across all significant cortical channels. All stimulation types evoke significant responses predominantly on the medial array.

FIGURE 4

Evoked cortical response of each stimulation mode. Evoked potentials for all modes of stimulation. Punctate stimulation exhibits evoked peaks after stimulation start and stop. This is mimicked by peripheral nerve stimulation only when the stimulation includes the phasic patterns. During both vibrotactile and constant frequency stimulation, cyclic entrainment is obvious. This diminishes at higher frequencies, but the evoked “off” response is absent.

FIGURE 5

Cortical responses to stimulation modes. From left to right: stimulation mode, evoked potential responses, onset gamma phase locking, delayed gamma phase locking, and the number of stimulation electrodes/sites that possessed the “appropriate” response in terms of gamma and evoked responses. The onset-release evoked potentials peaks are noticeable only under punctate, PP, and BB stimulations. Only one CP channel but all BB stimulation channels achieve an absence of delayed phase locked gamma. All channels of SP and PP fail to achieve the release evoked response. BB stimulation achieves the appropriate gamma patterns and evoked activity for 5/5 active stimulation channels.

FIGURE 6

Illustrative summary of bimodal biomimetic construction. Force applied during active grip tasks and object manipulations activates both the slow and fast adapting mechanoreceptors in the skin. The patterns observed in the peripheral nerve demonstrate onset-release firing bursts and sustained firing, both with some level of stochasticity. Firing patterns are never composed of determinate evenly spaced pulses, so stimulation should aim to mimic the physiological data. This study uses stimulation patterns that individually mimic the FA and SA traits, and combines them into a composite stimulation. Using this basic model of Bimodal Biomimetic stimulation, cortical responses more similar to punctate stimulation can be achieved.