Proprioceptive vibratory stimulation modulates cortico-muscular postural response

doi:10.21203/rs.3.rs-1619780/v1

Download PDF

Article

Proprioceptive vibratory stimulation modulates cortico-muscular postural response

https://doi.org/10.21203/rs.3.rs-1619780/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

For decades, postural control has been believed to be dominated by reflexive mechanisms, but more recent evidence characterizes it as an articulate system which relies on multisensory integration (vestibular, visual, proprioceptive) and cortical coordination to preserve balance in the wake of external disruptions to the quiet stance. Corrective strategies to stabilise the body’s center of mass and prevent falls are planned at the central nervous system level and executed through the modulation of muscular activity, especially in the lower legs muscles: gastrocnemius, soleus and tibialis. In this study, we use localized proprioceptive vibratory stimulation to induce a perturbation of the subjects’ quiet stance. By integrating traditional force plate data with electrophysiological signals (EEG and EMG), we use cortico-muscular and inter-muscular coherence to investigate how brain and muscles interact to shape an effective response to a postural challenge. Our results show how the cortical drive is enhanced in beta [13-30]Hz and gamma [30-45]Hz bands, specifically for the soleus and gastrocnemius muscles, during the task execution. We also reveal a remodulation of the lower legs muscle network, with a distributed strengthening of muscle activity coupling.

postural control

electroencephalography (EEG)

electromyography (EMG)

cortico-muscular coherence (CMC)

inter-muscular coherence (IMC)

proprioceptive stimulation

Since the first early studies at the beginning of the 20th century, human postural control has been traditionally believed to be the result of low-level reflexive mechanisms and tonic muscle contractions¹. This paradigm held up for decades, until it was challenged by novel anatomical and behavioural observations. Mounting experimental evidence eventually formed a consensus around a new theory that considers postural control as an articulated system which relies on multisensory integration and higher-level coordination. Indeed, the seemingly effortless act of maintaining balance in a quiet upright stance has been shown to require the active planning of corrective strategies in order to adapt to changes in the surrounding environment, external perturbations and altered sensorial conditions^2,3. Vestibular, visual, and somatosensory-proprioceptive systems have been shown to play a key role in providing the central nervous system (CNS) with the sensory feedback required to constantly monitor the postural condition^4,5. The integration of this sensory information at the cortical level is the basis for the development of a corrective strategy. The postural corrective strategy is then executed through cortical modulation of muscle contractions over time, with the aim to adjust the body’s centre of mass, thus reducing instability, and ultimately preventing falls. The muscles that are primarily involved in postural control are those of the lower legs (tibialis, soleus and gastrocnemius), which take part in the so-called ‘ankle strategy’ and help to stabilise the body’s position⁶.

Traditionally, human postural control has been investigated by posturographic means, i.e., by analysing the fluctuations of the centre of pressure (COP) over time during different balance tasks. In this study, we set out to use electroencephalography (EEG) and electromyography (EMG) to gain a better understanding of the physiological mechanisms shaping human postural control. In particular, we want to explore the relationship between cortical and muscular activity and the way brain and muscles coordinate to effectively respond to a postural challenge.

Here, the participant’s upright stance is disrupted by a randomised sequence of proprioceptive skeletomuscular vibrations applied to the calves. Mechanical vibrations applied to muscles and tendons are a widely used way to alter the proprioceptive afferent system ^3,7,8. The application of an external vibration induces the muscle spindles to synchronise to the vibratory stimulation, finally resulting in an increase of the motor unit firing rate^9–11. Specifically, the altered proprioceptive feedback tricks the CNS with the illusion of movement and muscles fibres being stretched, and results in a tonic reflexive response that shortens the muscle fibres while tilting the body in the direction of the vibration ¹².

To better understand how the efferent cortical drive responds to this kind of postural challenge and to investigate the muscular response, we used two tools: cortico-muscular coherence (CMC) and inter-muscular coherence (IMC)¹³.

CMC is a common approach to assess the level of interplay between neural signals and muscle activity. It is considered to reflect predominantly the descending efferent corticospinal pathway that carries the neural signal from the pyramidal neurons of the cortex to the spinal motor neurons and eventually controls the muscle fibres activity¹⁴. In addition, more recent literature also suggested a role of the ascending afferent system in shaping the CMC, with the proprioceptive signal stemming from muscle receptors and travelling to the somatosensory cortex¹⁵. This is supported by observations of beta band (13-30Hz) synchronisation in the motor and somatosensory cortices^16,17. Consistently, beta band activity has been shown to be a signature of enhanced proprioceptive feedback integration for the maintenance of the ongoing sensorimotor state -or status quo-, which here relates well to the attempt of preserving balance in the wake of a disturbance ^18,19.

IMC is a helpful tool to shed light on the motor control that the CNS exerts through muscles. In particular, it has been hypothesised that the CNS works by flexibly controlling combinations of muscular groups, resulting in so-called muscle synergies^20,21. The degree to which groups of muscle work together towards a specific task can be observed in the oscillatory patterns of their electrical activity, which reflect common oscillatory inputs from the CNS as well as afferent pathways¹³. Graph analysis techniques, usually employed in neural network scenarios, proved to be an innovative tool to enhance IMC analysis by providing an intuitive representation of the muscular response organization, in the form of a ‘muscle network’ ^22,23. IMC offers a way to evaluate the functional connectivity between muscles (nodes of the network), allowing the investigation of muscle activity modulation across tasks and experimental conditions, thus highlighting muscle synergies ^24,25.

In this study, we want to verify whether the postural challenge induced by localised proprioceptive vibrations leads to a shift towards high-level postural control strategy, with enhanced interplay between cortico-muscular signals and a modulation of muscular activity coupling, paving the way to new approaches to postural control interventions.

Participants and experimental design

Experimental data were acquired on 11 young healthy volunteers at Aston University’s ALIVE Lab, under the approval of the University Research Ethics Committee (ref: #1432). All participants were recruited among the student population (age: 24.4 ± 5.8; 5 males and 6 females). Criteria for participants’ recruitment included a minimum age of 18, no neuromuscular or balance disorders, no physical impediments to the maintenance of a prolonged upright stance and no medications or alcohol consumption during the previous 24 hours.

The bellies of the soleus (SOL), gastrocnemius lateralis (GL) and tibialis anterior (TA) muscles were identified following the SENIAM guidelines²⁶. The skin area around the identified locations was then shaved and cleaned with alcohol wet wipes. For the recording of surface EMG, two bipolar Ag/AgCl electrodes were applied to each muscle following the presumed direction of the underlying muscle fibres. A reference electrode was also applied to the medial malleolus. A 64-channel EEG cap was adjusted on the participant’s head and electroconductive gel was applied to reduce the impedance between the scalp and the Ag/AgCl wet electrodes of the cap, arranged according to a standard 10–20 system montage. The EEG cap was connected to a portable amplifier which the participant carried inside of a small backpack. The EMG bipolar electrodes were also connected to the same amplifier through a series of cascaded bipolar adaptors.

Finally, two vibrators were secured on the participant’s calves and connected to a custom-made actuator box attached to a belt and worn around the waist. The vibrators consisted of two eccentric rotating mass DC motors encapsulated in plastic cylinders (diameter: 30mm, length: 62mm), which delivered an 85Hz vibration. The actuator box, in addition to controlling the vibratory stimulation sequence, also acted as a trigger box connected to the amplifier and the force plate, enabling the synchronisation of the recorded signals with the stimulation epochs. Participants were finally asked to stand still on a force plate, in a natural upright posture, with arms relaxed along the body and facing straight ahead, gazing at a marker on the wall at about 2-meter distance. Figure 1 report a schematic of the study set-up.

Figure 1 about here (all figures are at the end of the file)

Figure 1 : Schematic representation of the experimental set-up. The participant was asked to stand on a force platform in a quiet, upright stance. After 30s of resting state (Baseline), a proprioceptive vibratory stimulation sequence was delivered through vibrators applied to the gastrocnemius muscles of the calves for 10min. The 64-channel EEG data, the EMG signals acquired on the TA, SOL and GL muscles, and the posturographic signal, were synchronized with triggers at the offset of each vibration. These triggers allowed for the definition of 1 second task epochs (highlighted in the figure), which were then concatenated into four quartiles.

Each session comprised two identical trials, performed respectively with open and closed eyes (OE, CE), in randomised order and with a 5-minute pause in between. Each trial started with 30s of baseline recording during an unperturbed quiet stance. Subsequently, the participant’s posture was disrupted via means of proprioceptive stimulation of the skeletal muscles of the calves. The stimulation was delivered for a total duration of 10min, in a randomised binary sequence of vibration and resting periods, with duration uniformly distributed between 1-6s

Data pre-processing

The EEG signal was acquired at a sampling frequency of 1000Hz and subsequently pre-processed using custom-made scripts in Matlab®R2019a (The Mathworks, Inc., Natick, MA). The pre-processing followed the semi-automated pipeline proposed in CARTOOL²⁷. First, the signal was filtered with a zero-phase band pass FIR filter between 1-45Hz and the DC offset was removed. The signal was then segmented into epochs of 1s duration, according to the triggers which marked the end of each vibration period. An EOG regression based on three frontal electrodes (Fp1, FpZ, Fp2) was performed to remove eye blink artifacts from the segmented epochs²⁸. For each epoch, channels with low signal-to-noise ratio were identified as those whose standard deviation (SD) exceeded the average SD of a factor greater than 1.7 and replaced with an interpolation of the adjacent channels. Epochs in which more than 15% of the channels were classified as ‘bad channels’ were discarded. The signal was finally re-references to common average.

The EMG signal was also pre-processed using custom-made Matlab scripts. First, the signal was filtered between 20-250Hz with a zero-phase band-pass Butterworth filter²⁹ and the 50Hz power line noise was removed with a notch filter. Then, the signal was rectified using a Hilbert transform³⁰. Finally, the acquired signal was segmented into 1s epochs, in the same way as the EEG signal, i.e., according to the triggers which denoted the end of each vibration period, in order to ensure data free from vibration-induced artefacts.

The COP trajectory recorded through the force plate was first detrended and filtered between 0.2-20Hz with a Butterworth filter³¹. Then, the signal was normalized to unit variance³² and segmented into epochs in the same fashion as the synchronised EEG and EMG signals.

Finally, the central 20s of the initial resting state recording were used to define the baseline epoch, while task epochs of each signal were concatenated into quartiles: Q1, Q2, Q3, Q4.

Posturography metrics

The average distance (DIST) and mean velocity (MV) of the COP were extracted from the posturographic data for baseline, Q1 and Q4. Given the anterior-posterior (AP) and medial-lateral (ML) components of the COP sway signal, over a time period of length T, comprising N time points, the average displacement of the COP from the center of the stabilogram is defined as³³:

$$DIST = \frac{\sum _{n=1}^{N}\sqrt{{AP}_{n}^{2}{+ML}_{n}^{2}}}{N}$$

The average velocity of the COP sway is defined as:

$$MV = \frac{\sum _{n=1}^{N}\sqrt{{({AP}_{n+1}-{AP}_{n})}^{2}+{({ML}_{n+1}-{ML}_{n})}^{2}}}{T}$$

A two-tailed Wilcoxon signed-ranked test was performed to check for statistically significant variations in these metrics, between baseline and Q1, and between Q1 and Q4. The computed p-values were then corrected for multiple comparisons with FDR (correction over 2 metrics and 2 tests)

CMC

The magnitude-squared coherence spectrum between EEG and EMG signals was computed to quantify the similarity of their frequency and phase content. For every subject, CMC was computed in beta band [13–30] Hz, for all the possible pairs of EMG and EEG channels (6 and 60, respectively), in baseline, Q1 and Q4, for the two experimental conditions (OE and CE). The coherence signal was computed as follows:

$${C}_{xy}\left(f\right)=\frac{{\left|{P}_{xy}\left(f\right)\right|}^{2}}{{P}_{xx}\left(f\right){P}_{yy}\left(f\right)}$$

Where ${P}_{xy}$ is the cross-spectral density between the two signals (Fourier transform of the cross-correlation), ${P}_{xx}$ is the power spectral density (PSD) of the EEG signal and ${P}_{yy}$is the PSD of the EMG signal. The PSD was estimated via Welch’s periodogram method³⁴, segmenting the epochs using non-overlapping Hamming windows with duration of one second³⁵.

To verify the presence of statistically significant changes in the coherence spectra following the administration of the proprioceptive vibratory stimulation, a cluster-based permutation test was performed between the coherence spectra computed during the resting state period (baseline) and during the first quartile of the postural task (Q1)³⁶. To investigate potential changes in cortico-muscular strategies over the course of the trial, as a result of habituation processes, a cluster-based permutation test was also performed between the first and last quartile of the task trial (Q1 and Q4, respectively).

The cluster-based permutation test was performed in the channel and frequency dimensions, to identify clusters of neighbouring EEG electrodes whose coherence with a specific muscle showed a statistically significant difference during the execution of the postural task compared to the resting state. The test was carried out in Fieldtrip³⁷ using 2000 permutations and comparisons were run in the beta and gamma frequency ranges.

IMC

IMC was also computed among EMG channels as a measure to estimate the degree to which the activity of couples of muscles is coordinated. This allowed to represent the participant’s lower leg muscles as a network in which each muscle represents a node (n = 6). The number of edges of the network is determined by the number of possible pairs (k = 2) of muscles as:

$${C}_{n,k}=\frac{n!}{k!\left(n-k\right)!} = 15$$

The PSD of the EMG signal was estimated on overlapping 1s windows using Welch’s method (75% overlap) and the magnitude-squared coherence was computed between [0–40] Hz²⁴. This process was repeated for each of the 15 pairs of muscles in baseline, Q1 and Q4, in both experimental conditions. To isolate the contribution of different frequency components, a non-negative matrix factorization (NNMF) approach was used. A NNMF iterative algorithm (5000 iterations) decomposed the coherence spectra into 5 frequency components and a set of associated weights. For every subject, then, the weights were used to define 5 $[n \times n]$ connectivity matrices, one for each frequency component, with n being the number of muscles.

The connectivity matrices so obtained were then used to perform a node-wise and edge-wise analysis on the lower leg muscle network.

For the node-wise analysis, 3 network metrics were extracted from each adjacency matrix, for each of the 6 nodes of the network: clustering coefficient (CC), local efficiency (LE) and strength (STR). For every muscle, differences in network metrics between baseline and Q1, and between Q1 and Q4 were tested with a two-tailed Wilcoxon signed-rank test. A false discovery rate (FDR) correction was applied on the resulting p-values to correct for multiple comparisons (procedure applied over 30 multiple comparisons, resulting from nodes/muscles times frequency components).

The edge-wise analysis consisted in a statistical comparison among connectivity matrices, to highlight significant changes in connectivity values in specific edges of the network. Differences were tested with a two-tailed Wilcoxon signed-rank test; p-values were corrected for multiple comparisons using FDR (procedure applied over 75 multiple comparisons, resulting from edges/muscle pairs times frequency components).

Posturography results

The posturography results are reported in Table 1 and illustrated in Fig. 2. The box plots show the distribution of DIST and MV values in baseline, Q1 and Q4, in the two experimental conditions (OE and CE). In both conditions, a significant increase in DIST and MV can be observed between baseline and Q1, with the associated p-values (p < 0.05) reported in the figure.

Table 1 about here (all figures are at the end of the file)

No significant variation in posturography metrics can be observed between the first and last quartile of the postural task, except for a slight increase in COP mean distance during the open-eyes trial.

Figure 2 about here (all figures are at the end of the file)

Figure 2. Posturography results: The box plots depict the distribution of 11 sample points for the posturography metrics: DIST (first row) and MV (second row), in baseline, first and last quartile (BASE, Q1 and Q4 respectively). Results during the closed-eyes trial are reported in the left column, results during open-eyes trial are reported in the right column. Centre lines show the medians; box limits indicate the 25th and 75th percentiles, sample points are represented by dots. Significant differences between BASE and Q1, or between Q1 and Q4 are indicated by a star. Significant p-values (p < 0.05) are also reported.

CMC results

Three muscles showed a significant increase in CMC in beta band during the first quartile of the CE trial compared to baseline, as observed in Fig. 3. The coherence between the left TA and a cluster of 14 central-occipital electrodes resulted significantly higher in Q1, in the frequency bin centred at 21.48Hz (p = 0.0235). The SOL muscles of both legs also showed a significant increase in Q1. Specifically, CMC between the left SOL and a cluster of 15 central electrodes was found to be significantly higher at 20.51Hz (p = 0.015). For the right SOL, significant differences were found on two adjacent frequency bins at [19.53–20.51] Hz, involving two overlapping clusters of central electrodes (p = 0.01).

Figure 3 about here (all figures are at the end of the file)

Figure 3. CMC results - Beta, Closed eyes: Each row depicts a pair of muscle and EEG electrodes cluster, whose CMC signal showed a significant difference between BASE and Q1 in one or multiple (consecutive) frequency bins. The first column shows the muscle electrode position on the participant’s lower legs. The second column depicts the location of the electrode clusters on the scalp, overlayed to the topographical distribution of CMC differences (BASE vs. Q1) in the specific frequency bin. If the p-value associated with the cluster is p < 0.05, electrodes are marked with an ‘X’; if p < 0.01, electrodes are marked with a ‘*’. The third column shows the CMC signals, averaged across the cluster’s electrodes, in Q1 and baseline, over the whole beta frequency band. The coloured area denotes the SD interval and a dashed line marks the frequency bin who resulted in a significant CMC difference.

During the OE trial, the only significant result emerged in beta band, for the right TA muscle, as shown in Fig. 4. The CMC spectra in Q1 and baseline, computed between the right TA and a cluster of 9 occipital electrodes, mostly overlaps throughout the beta band. In the 25.39 Hz frequency bin, though, a significant decrease in CMC during Q1 was detected (p = 0.013).

Figure 4 about here (all figures are at the end of the file)

Figure 4. CMC results - Beta, Open eyes

The figure shows the CMC results for the right TA muscle (first column), whose CMC signal with the cluster of electrodes depicted in the topography (second column) showed a significant difference between BASE and Q1 at 25.39Hz. The first column shows the muscle electrode position on the participant’s lower legs. The second column depicts the location of the electrode clusters on the scalp, overlayed to the topographical distribution of CMC differences (BASE vs. Q1) in the specific frequency bin. If the p-value associated with the cluster is p < 0.05, electrodes are marked with an ‘X’; if p < 0.01, electrodes are marked with a ‘*’. The third column shows the CMC signals, averaged across the cluster’s electrodes, in Q1 and baseline, over the whole beta frequency band. The coloured area denotes the SD interval and a dashed line marks the frequency bin who resulted in a significant CMC difference.

The cluster-based permutation test performed in gamma band [30–40] Hz, highlighted only one significant comparison, during the CE trial, as shown in Fig. 5. The CMC of the left GL muscle was found to significantly increase in Q1 compared to baseline, in two adjacent frequency bins: [41.02 -42] Hz (p = 0.005). As shown by the topographies the results involve two overlapping clusters of central electrodes.

Figure 5 about here (all figures are at the end of the file)

Figure 5. CMC results - Gamma, Closed eyes

The figure shows the CMC results for the left GL muscle (first column), whose CMC signal with the cluster of electrodes depicted in the topographies (second column) showed a significant difference between BASE and Q1 at 41.02Hz and 42Hz (first and second row respectively). The first column shows the muscle electrode position on the participant’s lower legs. The second column depicts the location of the electrode clusters on the scalp, overlayed to the topographical distribution of CMC differences (BASE vs. Q1) in the specific frequency bin. If the p-value associated with the cluster is p < 0.05, electrodes are marked with an ‘X’; if p < 0.01, electrodes are marked with a ‘*’. The third column shows the CMC signals, averaged across the cluster’s electrodes, in Q1 and baseline, over the whole beta frequency band. The colored area denotes the SD interval and a dashed line marks the frequency bin who resulted in a significant CMC difference.

IMC edge-wise results

The results of the IMC edgewise analysis are summarized in Figs. 6 and 7, for the closed eyes and open eyes trials respectively.

The first frequency component obtained from the NNMF is dominated by very low frequencies around 0 Hz and, because of its amplitude and scale, it is more likely to reflect an artefactual pattern rather than a physiological mechanism. The second component is the smallest in amplitude, but it is widespread across the whole frequency band analyzed, with greater values in beta and early gamma band. The last three components have comparable amplitudes and are characterized by clearer frequency bands. The third one peaks around 5Hz and 15 Hz. The fourth component is dominated by a 2 Hz contribution, while the fifth one is centred around 10 Hz.

Figure 6 about here (all figures are at the end of the file)

Figure 6. IMC edgewise results - Closed eyes: Each row corresponds to one of the five frequency components resulting from the NNMF of the coherence spectra. The first column shows the frequency components themselves, defined between [0–40] Hz. The second and third columns display the adjacency matrices, averaged across subjects, during baseline and Q1 respectively. The fourth column depicts the muscle network resulting from the statistical analysis, which highlights edges that showed a significant variation in connectivity values between baseline and Q1 (p < 0.05, FDR corrected). Specifically, green edges indicate increased connectivity in Q1 compared to baseline, while red edges indicate decreased connectivity. The thickness of the network edges is proportional to the percent of variation.

The results in both experimental conditions show, in the first two frequency components, a generalized decrease in network connectivity in Q1. Specifically, the reduced connection weights are appreciable across all muscles, both infra- and intra- leg (see Figs. 6 and 7, first and second rows). In the remaining three frequency components, on the other hand, the results show a generalized increment of connectivity across the network in Q1 compared to baseline. Consistently across OE and CE trials, the connection weights between muscles belonging to the same leg do not change between baseline and Q1, while the connections weights between bilateral muscles increases steadily across the three components.

Figure 7 about here (all figures are at the end of the file)

Figure 7. IMC edgewise results - Open eyes

Each row corresponds to one of the five frequency components resulting from the NNMF of the coherence spectra. The first column shows the frequency components themselves, defined between [0–40] Hz. The second and third columns display the adjacency matrices, averaged across subjects, during baseline and Q1 respectively. The fourth column depicts the muscle network resulting from the statistical analysis, which highlights edges that showed a significant variation in connectivity values between baseline and Q1 (p < 0.05, FDR corrected). Specifically, green edges indicate increased connectivity in Q1 compared to baseline, while red edges indicate decreased connectivity. The thickness of the network edges is proportional to the percent of variation.

No significant differences resulted from the comparison of the muscle networks obtained during Q1 and those obtained during Q4.

Imc Node-wise Results

The results of the IMC node-wise analysis are summarized in Figs. 8 and 9, for the closed eyes and open eyes trials respectively.

Figure 8 about here (all figures are at the end of the file)

Figure 8. IMC nodewise results - Closed eyes

Each row corresponds to one of the five frequency components resulting from the NNMF of the coherence spectra. The first column shows the frequency components themselves, defined between [0–40] Hz. The second, third and fourth columns show, respectively, the significant changes (p < 0.05, FDR corrected) for clustering coefficient (CC), local efficiency (LE) and strength (STR), between baseline and Q1. Specifically, green nodes denote an increased value for the corresponding muscle in Q1 compared to the baseline value. Red nodes denote a decreased value. The size of the nodes is proportional to the percent of variation.

Trials in the two experimental conditions (open and closed eyes) show similar patterns. The first two frequency components are associated with a slight but generalized decline in all the three metrics measured in Q1 compared to those measured at the baseline. The remaining three frequency components, on the other hand, see a consistent increase of CC, LE and STR. In particular, at the low frequencies (~ 2Hz), which is represented by the fourth components, a substantial increase in CC and LE can be observed for the TA and GL muscles of both legs in Q1 compared to baseline. Increased values can then be observed over all the network metrics for the GL muscles in the fifth components (with peaks at 1Hz and 10Hz). Moreover, although all significant, the changes observed for the last three components are more substantial than those observed for the first two components, especially around 2–10 Hz, for the TA and the GL.

Figure 9 about here (all figures are at the end of the file)

Figure 9. IMC nodewise results - Open eyes

Each row corresponds to one of the five frequency components resulting from the NNMF of the coherence spectra. The first column shows the frequency components themselves, defined between [0–40] Hz. The second, third and fourth columns show, respectively, the significant changes (p < 0.05, FDR corrected) for clustering coefficient (CC), local efficiency (LE) and strength (STR), between baseline and Q1. Specifically, green nodes denote an increased value for the corresponding muscle in Q1 compared to the baseline value. Red nodes denote a decreased value. The size of the nodes is proportional to the percent of variation.

No significant differences resulted from the comparison of the networks metrics obtained during Q1 and those obtained during Q4.

In line with our expectations, the analysis of the COP sway confirmed the disruptive effect that proprioceptive vibration of the calves has on the participants’ quiet stance. In fact, in both experimental conditions, the administration of the localized vibratory sequence resulted in an increased COP DIST and MV, which together reveal a decreased ability to maintain balance during Q1 compared to the resting state (BASE)³⁸. Interestingly, no significant difference was observed in the posturographic response between the initial and final phases of the trial (Q1 and Q4). With the exception of a slight but statistically significant increase of DIST during the OE task, the last quartile showed negligible postural variation when compared to Q1. These data, therefore, cannot support the presence of the phenomenon that has been termed ‘habituation’ in previous literature, and describes an adjustment in motor control strategies following the prolonged or repeated exposure to external perturbations or changing of environment ².

This seems to be further confirmed by the results of cortico-muscular and inter-muscular coherence, where no significant difference emerged in any of the considered metrics between Q1 and Q4.

Cortico-muscular coherence is a technique traditionally used to analyse the cortical drive to the muscles, i.e., the descending neural signal that modulates muscle activation³⁹. Studies on the phase lags of CMC signals showed their consistency with the conduction time between motor cortex and muscles, reinforcing the idea that CMC reflects the drive that cortical processes exert on motor neurons ^40,41. Previous literature also established beta and low gamma bands (13–30 Hz and 30-45Hz, respectively) as the primary frequency range where neuromuscular couplings effects can be observed, showing how different kind of muscle movements modulate the speed of cortico-muscular oscillations^14,42,43.

The investigation conducted in this study focused on those muscles (GL and SOL) which are mostly involved in the postural stabilization mechanism defined as ‘ankle strategy’, the primary one used to reposition the center of gravity either during quiet stance or in response to external perturbations^44,45. Being a dorsiflexor, the TA was also included to fully characterize the muscle response to the vibratory stimulation. The most significant results were observed during the CE trial. As depicted in Fig. 3, the results display a consistent increase in CMC, spread throughout beta band, for the SOL muscles after the administration of the proprioceptive vibration, compared to resting state. A statistically significant difference in CMC, identified by the cluster-based permutation test in the frequency bins around 20Hz, is consistent with what reported in previous literature¹⁸. The same pattern can be observed for GL in gamma band (see Fig. 5)⁴². While beta-band CMC has been traditionally associated with static force output, gamma-band CMC has been reported during strong contractions and dynamic force⁴². In addition, it has been shown how superficial extensor muscles, such as GL, are characterized by higher contraction frequency than deep extensors such as SOL.⁴⁶

The results of SOL and GL support the idea that the response to a disruptive proprioceptive vibration evokes an enhanced cortico-muscular coupling, particularly observable in correspondence of more challenging tasks such as the CE trial. The lack of visual input, a key afferent in the postural control system, seems to make a greater degree of high-level control necessary, as reflected by the increased CMC which suggest a lesser extent of automaticity in the execution of the task⁴⁷.

From a statistical standpoint, the observed results are particularly significant (p < 0.01) in the GL and SOL muscles of the dominant leg (right). On the other hand, the results for the TA muscle in beta (OE and CE) present a higher p-val (p < 0.05) and a criticality in the location of the associated electrode clusters in the scalp topography. Indeed, while the electrode clusters identified in the CMC results of SOL in beta band and GL in gamma band are located in the central area of the scalp topography, which is spatially coherent with the underlying motor cortical area (primary and supplementary motor cortex) where the cortical drive is originated, the TA results show an occipital-region cluster whose electrodes are unlikely to pick up electrical activity from the motor cortex^48,49. For these reasons, it is possible that the TA results don’t actually correspond to a cortico-muscular coupling effect but represent instead some other type of not yet fully understood synchronization mechanism.

The results of the IMC analysis provide further insight on the postural mechanisms that take place to counteract the imbalance caused by the proprioceptive vibration. Here, the IMC signals computed between pairs of muscles were decomposed into 5 frequency components, and the sets of weights associated with each component were used as a metric of connectivity in determining the networks of lower leg muscles. By comparing the resting state network topography with that recorded during execution of the postural task, we evaluated how the network reconfigured itself to shape a more effective postural strategy. The network topography highlights the groups of muscles whose activity is synchronized, showing the muscles involvement in common muscle synergies¹³.

Looking at the results reported in Figs. 6 and 7 it can be noticed that the adjacency matrices associated to the first two frequency components identified by the NNMF, present high (~ 1) and most likely non physiological connectivity values in both baseline and Q1²⁴. The distribution over the frequency band of these two components (the first dominated by ~ 0Hz and the second spread all over the band) further points to their artefactual nature.

Consistently with the previous results, significant differences in connectivity values and network metrics emerged among baseline and Q1. In this case, though, significant changes were observed during the OE trial as well as during the CE one (Figs. 6 and 7). In particular, higher connectivity values were observed during the execution of the postural task, as well as increased network metrics. The increased connectivity observed in the third frequency component supports the previously reported CMC-based results by confirming, even from a muscular network perspective, a stronger cortical drive expressed in terms of enhanced muscular synergies, IMC in the 16–40 Hz range is in fact thought to have a cortical origin⁵⁰. Because IMC does not reflect only cortical drive to the muscles, but also common afferent inputs¹³, the increased connectivity observed in the 6–15 Hz range (fifth component) and in the 0–5 Hz one (fourth component) could reflect an increased inflow of subcortical inputs too⁵⁰. The fact that not only cortical, but also afferent inputs are enhanced during the task execution, thus contributing to shape the postural response, is in line with the renowned fact that vibrations applied to the muscle bellies induce a stimulation of the muscle spindles⁹. Therefore, these proprioceptive structures are likely to be more sensitive immediately after the stimulation, increase the sensory input inflow and provide a larger contribution to the modulation of muscle activation, as suggested by our results.

In addition, the nodewise analysis highlights a strong increase in the clustering coefficient and local efficiency of the tibialis and gastrocnemius muscles in the low frequencies, around 2Hz (third frequency component, see Figs. 8 and 9). CC and LE are related network metrics which characterize the density of the network around a certain node by quantifying, respectively, the tendency of a node to form a cluster and the density of connections within the cluster (inverse of the shortest path length). A strengthening of the network at this low frequency (< 5Hz) is consistent with results reported in previous IMC studies during balance tasks ^24,47,51.

A note of caution must be introduced concerning a number of factors which may influence the ability to provide a univocal interpretation of the presented results. From a methodological point of view, both CMC and IMC techniques present some limitations, specifically the arbitrariness in the choice of the number of frequency components to use in the IMC analysis, as well as the intrinsic limitations in NNMF procedures themselves¹³. These techniques have been traditionally used in motor tasks involving voluntary movements and steady-state contractions such as grip, while not as much in balance or postural control tasks^52–54. The literature on the topic also lacks an extensive review of the factors affecting CMC and IMC, which may explain the variability observed across experimental designs, individuals etc. Finally, the current data are based on a limited cohort of young, healthy individuals. It will be of great interest to conduct further investigation on different subpopulations (such as ageing individuals) to compare the cortico-muscular recalibration strategies in the wake of the same postural challenge.

Our results show how proprioceptive mechanical vibratory stimulation of the calves causes a disruption of the quiet upright stance, which is met with the execution of a postural responsive mechanism to prevent falling. Cortico-muscular coherence analyses revealed how the stimulation of the proprioceptive system increased the higher-level modulation required to maintain balance. This is true especially during more challenging postural tasks (e.g., eyes closed), where increased CMC levels are likely to reflect the presence of more cortical modulation and more direct cortical control of muscular activation. This is observable in the GL and SOL muscles (particularly those of the dominant leg), which are known to shape the primary mechanical postural responses of the ankle strategy.

The stimulation of the proprioceptive system via means of focal vibrations led to a clear change in the control strategies adopted to shape the postural response. Not only the cortical drive modulating the recruitment of muscle synergies increased, but also the subcortical -or spinal- control of muscle groups increased, which is likely to be explained by the increased sensitivity of muscle spindle induced by the stimulation itself.

More interestingly, no changes were observed in the recruiting mechanism over the course of the task. The previously proposed concept of habituation is therefore not supported by the conducted analyses, where the same muscle networks were observed across frequency components over Q1 and Q4.

It will be beneficial to further this work by performing a comparative assessment on groups of subjects with impaired postural control, with the aim to identify how the postural response differs in these subjects in terms of cortico-muscular synergies.

Author’s contributions

Conception and design: FB, IR, PG, and AF. Data acquisition and analysis: FB and IR. Interpretation: FB, IR and AF. Drafting manuscript: FB, IR and AF. FB and IR are first co-authors.

Competing interests All authors declare that they have no conflicts of interest.

Availability of data and material Since sharing data in an open-access repository was not included in our participant’s consent and therefore compromises our ethical standards, data are only available on request from the corresponding author.

Code availability The code used for the analyses of the data will be shared upon request to the corresponding author

Ethics approval The study was carried out according to the Declaration of Helsinki (2013) and was approved by the University Research Ethics Committee at Aston University (reference number: 1432).

Consent to participate All participants provided informed consent before participating.

Consent for publication All co-authors were aware of the publication of this study

Magnus, R. The Croonian Lecture: animal posture. Proc. R. Soc. B 98, 339–353 (1925).
Fransson, P.-A. Analysis of adaptation in human postural control. Lund University Sweden (Lund University, 2005).
Patel, M., Fransson, P. A. & Magnusson, M. Effects of ageing on adaptation during vibratory stimulation of the calf and neck muscles. Gerontology 55, 82–91 (2009).
Forbes, P. A., Chen, A. & Blouin, J. S. Sensorimotor control of standing balance. Handb. Clin. Neurol. 159, 61–83 (2018).
Horak, F. B. & Macpherson, J. M. 7. Postural orientation and equilibrium. in Handbook of Physiology, Exercise: Regulation and Integration of Multiple Systems (ed. Rowell LB, S. J.) 255–92 (Oxford University Press, 1996).
Horak, F. B. & Nashner, L. M. Central programming of postural movements: Adaptation to altered support-surface configurations. J. Neurophysiol. 55, 1369–1381 (1986).
Ledin, T., Hafström, A., Fransson, P. & Magnusson, M. Influence of neck proprioception on vibration-induced postural sway. Acta Otolaryngol. 123, 594–599 (2003).
Malmström, E., Fransson, P., Jaxmar Bruinen, T., Facic, S. & Tjernström, F. Disturbed cervical proprioception affects perception of spatial orientation while in motion. Exp. brain Res. 235, 2755–2766 (2017).
Burke, D., Hagbarth, K. E., Lofstedt, L. & Wallin, B. G. The responses of human muscle spindle endings to vibration during isometric contraction. J. Physiol. 261, 695–711 (1976).
Hagbarth, K. E., Hellsing, G. & Löfstedt, L. TVR and vibration-induced timing of motor impulses in the human jaw elevator muscles. J. Neurol. Neurosurg. Psychiatry 39, 719–728 (1976).
Homma, S., Kanda, K. & Watanabe, S. Preferred spike intervals in the vibration reflex. Jpn. J. Physiol. 22, 421–432 (1972).
Čapičíková, N., Rocchi, L., Hlavačka, F., Chiari, L. & Cappello, A. Human Postural Response to Lower Leg Muscle Vibration of Different Duration. Physiol. Res 55, 129–134 (2006).
Boonstra, T. W. The potential of corticomuscular and intermuscular coherence for research on human motor control. Front. Hum. Neurosci. 7, 1–2 (2013).
Liu, J., Sheng, Y. & Liu, H. Corticomuscular coherence and its applications: A review. Frontiers in Human Neuroscience 13, 100 (2019).
Witham, C. L., Riddle, C. N., Baker, M. R. & Baker, S. N. Contributions of descending and ascending pathways to corticomuscular coherence in humans. J. Physiol. 589, 3789–3800 (2011).
Witham, C. L., Wang, M. & Baker, S. N. Cells in somatosensory areas show synchrony with beta oscillations in monkey motor cortex. Eur. J. Neurosci. 26, 2677–2686 (2007).
Witham, C. L. & Baker, S. N. Network oscillations and intrinsic spiking rhythmicity do not covary in monkey sensorimotor areas. J. Physiol. 580, 801–814 (2007).
Baker, S. N. Oscillatory interactions between sensorimotor cortex and the periphery. Current Opinion in Neurobiology 17, 649–655 (2007).
Engel, A. K. & Fries, P. Beta-band oscillations-signalling the status quo? Current Opinion in Neurobiology 20, 156–165 (2010).
Tresch, M. & Jarc, A. The case for and against muscle synergies. Curr. Opin. Neurobiol. 19, 601–607 (2009).
Todorov, E. Optimality principles in sensorimotor control. Nat. Neurosci. 2004 79 7, 907–915 (2004).
Hassan, M. & Wendling, F. Electroencephalography Source Connectivity: Aiming for High Resolution of Brain Networks in Time and Space. IEEE Signal Process. Mag. 35, 81–96 (2018).
Rubinov, M. & Sporns, O. Complex network measures of brain connectivity: Uses and interpretations. Neuroimage 52, 1059–1069 (2010).
Boonstra, T. W. et al. Muscle networks: Connectivity analysis of EMG activity during postural control. Sci. Rep. 5, 1–14 (2015).
Kerkman, J. N., Bekius, A., Boonstra, T. W., Daffertshofer, A. & Dominici, N. Muscle Synergies and Coherence Networks Reflect Different Modes of Coordination During Walking. Front. Physiol. 0, 751 (2020).
Stegeman, D. F. Standards for suface electromyography: The European project Surface EMG for non-invasive assessment of muscles (SENIAM). (2007).
Brunet, D., Murray, M. M. & Michel, C. M. Spatiotemporal analysis of multichannel EEG: CARTOOL. Computational Intelligence and Neuroscience 2011, (2011).
Parra, L. C., Spence, C. D., Gerson, A. D. & Sajda, P. Recipes for the linear analysis of EEG. Neuroimage 28, 326–341 (2005).
Roeder, L., Boonstra, T. W. & Kerr, G. K. Corticomuscular control of walking in older people and people with Parkinson’s disease. Sci. Rep. 10, (2020).
Boonstra, T. W. & Breakspear, M. Neural mechanisms of intermuscular coherence: implications for the rectification of surface electromyography. J. Neurophysiol. 107, 796–807 (2012).
Zhou, J., Lipsitz, L., Habtemariam, D. & Manor, B. Sub-sensory vibratory noise augments the physiologic complexity of postural control in older adults. J. Neuroeng. Rehabil. 13, 44 (2016).
Donker, S. F., Roerdink, M., Greven, A. J. & Beek, P. J. Regularity of center-of-pressure trajectories depends on the amount of attention invested in postural control. Exp. Brain Res. (2007). doi:10.1007/s00221-007-0905-4
Prieto, T. E., Myklebust, J. B., Hoffmann, R. G., Lovett, E. G. & Myklebust, B. M. Measures of postural steadiness: Differences between healthy young and elderly adults. IEEE Trans. Biomed. Eng. 43, 956–966 (1996).
Welch, P. D. The Use of Fast Fourier Transform for the Estimation of Power Spectra: A Method Based on Time Averaging Over Short, Modified Periodograms. IEEE Trans. Audio Electroacoust. 15, 70–73 (1967).
Jacobs, J. V., Wu, G. & Kelly, K. M. Evidence for beta corticomuscular coherence during human standing balance: Effects of stance width, vision, and support surface. Neuroscience (2015). doi:10.1016/j.neuroscience.2015.04.009
Maris, E. & Oostenveld, R. Nonparametric statistical testing of EEG- and MEG-data. J. Neurosci. Methods 164, 177–190 (2007).
Oostenveld, R., Fries, P., Maris, E. & Schoffelen, J. M. FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput. Intell. Neurosci. 2011, (2011).
Palmieri-Smith, R. M., Ingersoll, C. D., Stone, M. B. & Krause, B. A. Center-of-Pressure Parameters Used in the Assessment of Postural Control. J. Sport Rehabil. 11, 51–66 (2002).
Halliday, D. M. et al. A framework for the analysis of mixed time series/point process data-theory and application to the study of physiological tremor, single motor unit discharges and electromyograms. Prog. Biophys. Mol. Biol. 64, 237–278 (1995).
Gross, J. et al. Cortico-muscular synchronization during isometric muscle contraction in humans as revealed by magnetoencephalography. J. Physiol. 527 Pt 3, 623–631 (2000).
Negro, F. & Farina, D. Linear transmission of cortical oscillations to the neural drive to muscles is mediated by common projections to populations of motoneurons in humans. J. Physiol. 589, 629 (2011).
Gwin, J. T. & Ferris, D. P. Beta- and gamma-range human lower limb corticomuscular coherence. Front. Hum. Neurosci. 0, 258 (2012).
Reyes, A., Laine, C. M., Kutch, J. J. & Valero-Cuevas, F. J. Beta Band Corticomuscular Drive Reflects Muscle Coordination Strategies. Front. Comput. Neurosci. 0, 17 (2017).
Gatev, P., Thomas, S., Kepple, T. & Hallett, M. Feedforward ankle strategy of balance during quiet stance in adults. J. Physiol. 514, 915–928 (1999).
Winter, D. A.. Human balance and posture control during standing and walking. Gait Posture 3, 193–214 (1995).
Honma, S. & Seki, Y. Muscle spindles in phasic and tonic muscle. Tohoku J. Exp. Med. 83, 391–397 (1964).
Peterson, S. M. & Ferris, D. P. Group-level cortical and muscular connectivity during perturbations to walking and standing balance. Neuroimage 198, 93–103 (2019).
Melnik, A., Hairston, W. D., Ferris, D. P. & König, P. EEG correlates of sensorimotor processing: independent components involved in sensory and motor processing. Sci. Rep. 7, (2017).
Sreeja, S. R., Rabha, J., Samanta, D., Mitra, P. & Sarma, M. Classification of motor imagery based EEG signals using sparsity approach. Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics) 10688 LNCS, 47–59 (2017).
Nandi, T., Hortobágyi, T., van Keeken, H. G., Salem, G. J. & Lamoth, C. J. C. Standing task difficulty related increase in agonist-agonist and agonist-antagonist common inputs are driven by corticospinal and subcortical inputs respectively. Sci. Rep. 9, 1–12 (2019).
Boonstra, T. W. et al. Low-alcohol doses reduce common 10- to 15-Hz input to bilateral leg muscles during quiet standing. J. Neurophysiol. 100, 2158–2164 (2008).
Watanabe, T., Nojima, I., Mima, T., Sugiura, H. & Kirimoto, H. Magnification of visual feedback modulates corticomuscular and intermuscular coherences differently in young and elderly adults. Neuroimage (2020). doi:10.1016/j.neuroimage.2020.117089
Ushiyama, J., Yamada, J., Liu, M. & Ushiba, J. Individual difference in β-band corticomuscular coherence and its relation to force steadiness during isometric voluntary ankle dorsiflexion in healthy humans. Clin. Neurophysiol. 128, 303–311 (2017).
Omlor, W., Patino, L., Hepp-Reymond, M. C. & Kristeva, R. Gamma-range corticomuscular coherence during dynamic force output. Neuroimage 34, 1191–1198 (2007).

Table 1. Posturography results. The average values of COP mean distance (DIST) and mean velocity (MV) are reported in the table, with their standard deviation interval, for baseline, Q1 and Q4, in the two experimental conditions (CE and OE).

	CE			OE
	Base	Q1	Q4	Base	Q1	Q4
DIST	0.735 ± 0.317	1.546±0.432	1.443±0.132	0.882±0.429	1.310±0.257	1.499±0.292
MV	4.132±2.3414	8.236±3.807	7.537 ±1.873	6.415±7.429	9.183±7.383	10.036±7.746

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Proprioceptive vibratory stimulation modulates cortico-muscular postural response

Status:

Version 1

Abstract

Figures

Introduction

Materials And Methods

Participants and experimental design

Data pre-processing

Posturography metrics

CMC

IMC

Results

Posturography results

CMC results

IMC edge-wise results

Imc Node-wise Results

Discussions

Conclusions

Declarations

References

Table

Additional Declarations

Status:

Version 1