Abstract
Background:
Studies have found that pairing vagus nerve stimulation (VNS) with motor activity accelerates cortical reorganization and, and this synchronous pairing may enhance motor recovery.
Objective:
To develop and validate a motor-activated auricular vagus nerve stimulation (MAAVNS) system as a potential neurorehabilitation tool.
Methods:
We created MAAVNS and validated its function as part of an ongoing clinical trial investigating whether taVNS-paired rehabilitation enhances oromotor learning. We compared 3 different MAAVNS EMG electrode configurations in 3 neonates. The active lead was placed over the buccinator muscle. Reference lead placements were orbital, temporal or frontal.
Results:
The frontal reference lead produced the highest sensitivity (0.87±0.07 (n=8)) and specificity (0.64±0.13 (n=8)). Oral sucking reliably triggers MAAVNS stimulation with high confidence.
Conclusion:
EMG electrodes placed on target orofacial muscles can effectively trigger taVNS stimuli in infants in a closed loop fashion.
Introduction
Vagus nerve stimulation (VNS) synchronously paired with motor training has shown promise in reorganizing the primary motor cortex after brain injury in animal models and improving functional rehabilitation in stroke patients.1,2 This restoration of pathologically deficient neural activity is likely due to neuroplastic mechanisms activated by VNS, as the synergistic effects only occur when active stimulation is delivered with paired therapy.3,4 The vagus nerve can be stimulated non-invasively via the auricular branch in the ear by transcutaneous auricular VNS (taVNS), which has shown similar afferent neural activation patterns as traditional cervical VNS.5 Recently, taVNS paired with motor rehabilitation has been explored to treat motor function impairment post-stroke.6
Recently our group has demonstrated the initial safety and feasibility of pairing taVNS with neonatal occupational therapy to improve oromotor function in infants with feeding deficits.7 However, infant bottle sucking behavior can be difficult to continuously monitor visually and thus effectively pair consistently with neurostimulation. A trained provider must constantly monitor the feeding and then trigger the stimulation manually. This is labor intensive, costly, and introduces variation in different observers. A closed-loop system could provide a more accurate stimulation delivery method while reducing the number of operators required.
Oral feeding uses several facial muscles in concert, including the buccinator, masseter, and temporalis, which can be measured using electromyography (EMG).8,9,10,11 An EMG-paired taVNS system, triggered by muscles important for functional rehabilitation, has not yet been explored. Functional plasticity in the motor cortex may require intricate timing between targeted movement and applied stimulation. EMG provides indication of muscle activity with high spatial and temporal resolution – pairing these muscle-specific movements in real-time with taVNS may accelerate motor learning and facilitate restoration of deficient neural processes.
We designed and built a closed-loop motor-activated auricular vagus nerve stimulation (MAAVNS) system and evaluated whether MAAVNS is sensitive enough to detect sucking behavior, and specific enough to discriminate coordinated sucking motor activity from noise and other movements in a neonatal population. For determination of optimal settings, we prioritized sensitivity for nutritive over non-nutritive sucking movements.
Techniques and Methods
Overview of Motor Activated Auricular Vagus Nerve Stimulation (MAAVNS) system
Our closed-loop MAAVNS system uses EMG detection of movement of orofacial muscles to trigger taVNS stimulation when these target muscles were activated in a suck-swallow oromotor sequence (Figure 1a, 1b). We enrolled 3 neonates in a sub-study comparing 3 different EMG configurations to investigate the utility of MAAVNS and determine the optimal EMG configuration to reliably pair taVNS stimulation with nutritive sucking during bottle feeding.
Figure 1:
a) EMG lead placement for position C. Active lead on buccinator, reference lead on frontal eminence, ground lead in center of forehead. b) Overview of MAAVNS set up. EMG signals from facial muscles (1) were processed (2-5) and used to trigger stimulation (6, stim). c) Raw EMG signal was processed by amplifying, rectifying and integrating the signal. An activation threshold was set and calibrated to a visual suck, delivering a TTL output when a suck was detected.
EMG Lead Configuration and Muscle Selection
Three different EMG configurations were tested (Figure 2a). Small round 20mm EMG electrodes (Natus) were placed on the left side of the subject’s face and connected to a pre-amplifier (NeuroLog NL844). The active EMG lead was placed over the buccinator approximately 5 cm horizontally from the corner of the mouth for all configurations. The reference EMG lead location varied depending on the configuration: A) Orbital Rim, B) Temporal Ridge, C) Frontal Eminence. The common ground electrode was placed in the center of the forehead, approximately 2.5 cm above the brow ridge, for all configurations. Each configuration was tested independently on different treatment days over 3 neonates (n= 3(A), 6(B), 8(C)).
Figure 2:
a) EMG lead placement for Configurations A, B, and C. b) Sensitivity and Specificity of EMG lead Configurations A, B, and C. This data demonstrates that both Configurations B and C had a sensitivity greater than 86%, however, Configuration C had a significantly greater specificity than Configuration B (64% compared to 40%).
EMG Signal Processing
EMG raw signal was passed through a 4 channel AC pre-amplifier (NeuroLog NL844) and into an Isolator (NeuroLog NL820A) for signal amplification (×10000) (Figure 1b). The signal was filtered using low pass and high pass filters (NeuroLog NL136 & NL144) to remove signals outside the target frequency. The filtered signal was passed through a signal conditioner (NeuroLog NL530) to add gain and offset controls. Using an integrator (NeuroLog NL703) the amplified EMG response is converted into an analog voltage signal corresponding to the amount of motor activity at a sampling rate of 100 ms. This analog signal is full-wave rectified to ensure the voltage output is positive. The signal is then passed through a gated amplitude discriminator (NL201 Spike Trigger) which enables the calibration of an activation threshold by setting the “window height”. Adjustments to the “window height” allow for the fine tuning of trigger sensitivity. The window height was set to approximately 0.4 volts and was used to calibrate stimulation trigger sensitivity prior to each trial. The amplitude discriminator converts analog EMG signal spikes (target muscle activation) into trains of digital pulses that pass to a delay module (NL405 without delay) that produces adjustable TTL Logic output used to trigger taVNS stimulation. Fiber optic wires connect all the modules to minimize latency of signal transmission.
Administering taVNS
The TTL output signal generated from the EMG signal processing steps (prior section) is relayed to a TTL Train Generator (Digitimer Model DG2A) which triggers the constant current stimulator (Digitimer DS7AH) and initiates stimulation at pre-set parameters (3 second trains of 25 Hz stimulation at a 500μs pulse width). Stimulation is delivered at 0.1 mA below individual perceptual threshold via custom neonatal taVNS electrodes described in Badran Jenkins et al 20197 (under review – Frontiers in Human Neurosciences). Perceptual threshold was determined prior to each treatment by increasing the current intensity in 0.1 mA increments until the physical acknowledgment of the stimulation by the subject was observed (head turn, wince, etc.). The amount of stimulation that was delivered during each treatment session was dependent on the activity of the subject during the feed. If the subject did not have an effective feed, then the amount of stimulation that was delivered during that feed was decreased due to lack of oromotor activation. Refer to supplemental video for operational demonstration.
Determination of Optimal EMG Lead Configuration
At the beginning of each trial, the window height of the EMG Signal Processing system was calibrated such that taVNS stimulation was triggered when a nutritive suck was visually observed. The window height ranged from 0.3 to 0.6 V depending on the strength of the subject’s oromotor function associated with a nutritive suck. A designated observer was tasked with identifying nutritive sucks and whether they were correctly paired with the EMG-driven stimulation. Nutritive sucks that failed to trigger stimulation were also recorded. A positive nutritive suck/stimulation pairing was also indicated if the suck occurred within an ongoing 3 second train of stimulation.
Sensitivity and specificity scores were used to determine the relative effectiveness of each EMG lead configuration. Sensitivity was defined as the number of correctly paired stimulations divided by the total number of visually-confirmed sucks recorded. Sensitivity scoring reflected whether the EMG lead configuration could trigger taVNS stimulation even with sucks that appeared weaker but were related to a swallow.
Specificity was defined as the number of correctly paired stimulations divided by the total number of stimulation trains. Specificity scoring reflected whether the configuration could determine a nutritive suck from other movement/noise. An unpaired t-test was performed comparing the sensitivities and specificities of configurations A, B and C.
Results
Demographics
The 3 neonates enrolled had a mean gestational age (GA) at birth of 33.57±3.37 weeks (mean±SD). Mean GA at enrollment was 40.43±2.85 weeks. 2 subjects were black, and 1 subject was white. All 3 subjects were female.
Sensitivity and Specificity
We recorded from 17 independent 30-minute taVNS-paired feeding sessions. Each session, on average had a mean of 268 suck-related events that were recorded and analyzed for sensitivity and specificity. The sensitivity of the 3 tested EMG configurations were (mean±SD) A: 0.77±0.16 (n=3), B: 0.86±0.10 (n=6), and C: 0.87±0.07 (n=8). The specificity of the 3 tested EMG configurations were (mean±SD) A: 0.49±0.32 (n=3), B: 0.40±0.14 (n=6), and C: 0.64±0.13 (n=8) (Figure 2b). All sites captured the majority of visually confirmed sucks with greater than 75% confidence (Figure 2b), however Placement C (frontal eminence) provided the highest specificity while maintaining sufficient sensitivity in capturing weaker sucks. The specificity of Placement C was significantly different compared to Placement B (P=0.0018). Sensitivity of Placements A, B and C were not significantly different.
Discussion
We have designed and tested a closed-loop motor activated auricular vagus nerve stimulation (MAAVNS) system that uses real-time EMG signals from orofacial muscles to trigger taVNS in a motor rehabilitation setting. Three different EMG configurations demonstrated that MAAVNS was sensitive enough to capture small, visible suck-swallow sequences. The EMG reference electrode placed on the frontal eminence was the most specific.
Using real-time EMG activation paired with taVNS in a closed-loop fashion as described may facilitate motor learning and restoration of aberrant neural circuitry. This closed-loop system provides reliability, consistency, and reproducibility throughout treatments minimizing human error and ultimately leading the way towards automated clinic- and home-based neuromodulation therapy. MAAVNS is relatively inexpensive and was developed as an open-source platform for researchers to build laboratory-based systems in future trials.
Limitations
We observed a sensitivity-specificity trade off when determining the best window height adjustment. We determined that not missing a nutritive suck was a priority for our objective to optimize treatment for motor learning. This was reflected in the results of all three configurations. Another potential source of variability in specificity was that placement of the electrode may be at the junction of the masseter and the buccinator at our target location. However, both muscles are involved in the suck -swallow sequence.
Conclusion
A closed loop MAAVNS system is feasible with reasonable sensitivity and specificity. In the future, EMG could possibly be used to trigger other brain stimulation modalities that need to be paired with motor function. The MAAVNS system can be applied to any taVNS-paired motor rehabilitation paradigm - as long as the desired muscle action can be isolated, and EMG leads can be adhered to quantify activation.
Supplementary Material
Highlights:
A motor-activated auricular vagus nerve stimulation (MAAVNS) system effectively and consistently pairs stimulation with oromotor activation generated by a nutritive suck in neonates.
Using a closed-loop system to deliver transcutaneous auricular vagus nerve stimulation allows for more uniform delivery of treatment across multiple operators.
MAAVNS is feasible and may become a promising tool for both adult and pediatric neuromodulation-based rehabilitation.
Acknowledgments
Financial Support: This study was funded by the National Center of Neuromodulation for Rehabilitation (NC NM4R). The National Center of Neuromodulation for Rehabilitation (NC NM4R) is supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under award number P2CHD086844 and P20GM109040.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
For submission to: Brain Stimulation
Conflict of Interest Statement: DNC, MSG, DDJ, BWB are listed as inventors on pending patents assigned to the Medical University of South Carolina on the methods described in this manuscript.
References
- 1.Porter BA, Khodaparast N, Fayyaz T, Cheung RJ, Ahmed SS, Vrana WA, Rennaker RL 2nd, Kilgard MP 2012. Repeatedly pairing vagus nerve stimulation with a movement reorganizes primary motor cortex. Cereb Cortex 22:2365–2374. [DOI] [PubMed] [Google Scholar]
- 2.Dawson J, Pierce D, Dixit A, Kimberley TJ, Robertson M, Tarver B, Hilmi O, McLean J, Forbes K, Kilgard MP, Rennaker RL, Cramer SC, Walters M, Engineer N 2016. Safety, Feasibility, and Efficacy of Vagus Nerve Stimulation Paired With Upper-Limb Rehabilitation After Ischemic Stroke. Stroke 47:143–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Morrison RA, Hulsey DR, Adcock KS, Rennaker RL 2nd, Kilgard MP, Hays SA 2018. Vagus nerve stimulation intensity influences motor cortex plasticity. Brain Stimul. 12:256–262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Engineer CT, Engineer ND, Riley JR, Seale JD, Kilgard MP 2015. Pairing Speech Sounds With Vagus Nerve Stimulation Drives Stimulus-specific Cortical Plasticity. Brain Stimul 8:637–644. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Badran BW, Dowdle LT, Mithoefer OJ, LaBate NT, Coatsworth J, Brown JC, DeVries WH, Austelle CW, McTeague LM, George MS 2018. Neurophysiologic effects of transcutaneous auricular vagus nerve stimulation (taVNS) via electrical stimulation of the tragus: A concurrent taVNS/fMRI study and review. Brain Stimul 11:492–500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Redgrave JN, Moore L, Oyekunle T, Ebrahim M, Falidas K, Snowdon N, Ali A, Majid A 2018. Transcutaneous Auricular Vagus Nerve Stimulation with Concurrent Upper Limb Repetitive Task Practice for Poststroke Motor Recovery: A Pilot Study. J Stroke Cerebrovasc Dis. [DOI] [PubMed] [Google Scholar]
- 7.Badran BW, Jenkins DD, Cook DN, Thompson S, Dancy M, Devries WH, Mappin G, Summers P, Bikson M, George MS Transcutaneous Auricular Vagus Nerve Stimulation (taVNS)-paired Rehabilitation for Oromotor Feeding Problems in Newborns is Feasible, Safe, and Shows Potential Efficacy. (Under Review – Frontiers in Human Neuroscience). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tamura Y, Matsushita S, Shinoda K, Yoshida S 1998. Development of periorial muscle activity during suckling in infants: a cross-sectional and follow up study. Dev Med Child Neurol 40:344–348. [PubMed] [Google Scholar]
- 9.Martins CD, Furlan RM, Motta AR, Viana MC 2015. Electromyography of muscles involved in feeding premature infants. Codas 27:372–377. [DOI] [PubMed] [Google Scholar]
- 10.Gomes CF, Trezza EM, Murade ED, Padovani CR 2006. Surface electromyography of facial muscles during natural and artificial feeding of infants. J Pediatr 82:103–109. [DOI] [PubMed] [Google Scholar]
- 11.Nyqvist KH, Färnstrand C.Eeg-Olofsson KE, Ewald U 2001. Early oral behavior in preterm infants during breastfeeding: an electromyographic study. Acta Paediatr 90:658–663 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.