Detection of synchrony in the discharge of a population of neurons. I. Development of a synchronization index

Effect of cancellation on triggered averaging used to determine synchronization between motor unit discharge in separate muscles

Synchronization between single motor unit (SMU) discharges in separate muscles has been estimated from peaks in averaged electromyographic (EMG) recordings from one muscle triggered from SMU discharge in another. This study evaluated the effect of EMG signal cancellation on this measure of synchronization. SMU activity was recorded with 8 fine-wire electrodes in vastus medialis obliquus (VMO) and vastus lateralis (VL) during gentle isometric knee extension in 7 subjects. Data from 5 VL recordings were summed then rectified, or rectified then summed, to produce multi-unit recordings with and without cancellation, respectively. Averages of summed VL data were triggered from VMO SMUs. Synchronization, defined as a peak >3 SD above the triggered average mean, occurred in 73.68% and 78.95% of recordings with and without cancellation, respectively. To further investigate the effect of cancellation on synchronization, 250 "virtual" EMG recordings were created from VL data. VL SMUs were sorted and modified with respect to discharge rate, amplitude and polarity to create a collection of possible SMU discharge patterns. Virtual recordings were added one-by-one to VL recordings that showed synchronization. Virtual channels were rectified then added or added then rectified, to create data with and without cancellation. Identification of synchronization decreased similarly in both conditions with addition of virtual data. Our data show estimation of synchronization from triggered averages is more likely to detect synchronization in recordings with fewer SMUs, but cancellation has little effect. Synchronization must be interpreted with caution if number of SMUs changes between conditions.

Amplitude cancellation reduces the size of motor unit potentials averaged from the surface EMG

The purpose of the study was to evaluate the influence of selected physiological parameters on amplitude cancellation in the simulated surface electromyogram (EMG) and the consequences for spike-triggered averages of motor unit potentials derived from the interference and rectified EMG signals. The surface EMG was simulated from prescribed recruitment and rate coding characteristics of a motor unit population. The potentials of the motor units were detected on the skin over a hand muscle with a bipolar electrode configuration. Averages derived from the EMG signal were generated using the discharge times for each of the 24 motor units with lowest recruitment thresholds from a population of 120 across three conditions: 1) excitation level; 2) motor unit conduction velocity; and 3) motor unit synchronization. The area of the surface-detected potential was compared with potentials averaged from the interference, rectified, and no-cancellation EMGs. The no-cancellation EMG comprised motor unit potentials that were rectified before they were summed, thereby preventing cancellation between the opposite phases of the potentials. The percent decrease in area of potentials extracted from the rectified EMG was linearly related to the amount of amplitude cancellation in the interference EMG signal, with the amount of cancellation influenced by variation in excitation level and motor unit conduction velocity. Motor unit synchronization increased potentials derived from both the rectified and interference EMG signals, although cancellation limited the increase in area for both potentials. These findings document the influence of amplitude cancellation on motor unit potentials averaged from the surface EMG and the consequences for using the procedure to characterize motor unit properties.

Motor unit synchronization between medial and lateral vasti muscles

OBJECTIVE

Accurate neuromuscular control of the patellofemoral joint is important in knee joint mechanics. Strategies to coordinate the vasti muscles, such as motor unit synchronization, may simplify control of patellar tracking. This study investigated motor unit synchronization between vastus medialis (VM) and lateralis (VL).

METHODS

Electromyographic (EMG) recordings of single motor unit action potentials (MUAPs) were made from VM and single- and multi-unit recordings were made from VL. Synchronization was quantified from peaks in the cross-correlogram generated from single MUAP pairs in VL and VM. The proportion of motor units in VM with synchronized firing in VL was also quantified from peaks in averages of multiunit VL EMG triggered from the VM MUAP.

RESULTS

A high degree of synchronization of motor unit firing between VM and VL was identified. Results were similar for cross-correlation ( approximately 45% of cases) and triggered averages (approximately 41% of cases).

CONCLUSIONS

The data suggest that synchronization between VM and VL is higher than expected. Agreement between traditional cross-correlation and triggered averaging methods suggest that this new technique may provide a more clinically viable method to quantify synchronization.

SIGNIFICANCE

High synchronization between VM and VL may provide a solution to simplify control of the mechanically unstable patellofemoral joint.

Limitations of the surface electromyography technique for estimating motor unit synchronization

Motor unit synchronization was estimated from the surface electromyograms (EMG) of the first dorsal interosseus muscle of human volunteers by a simplified surface-EMG technique (Milner-Brown et al. 1973, 1975). Single motor units were identified from intramuscular recordings and were used to obtain a spike-triggered average of the surface-EMG. The discharge rate of a reference motor unit was controlled at two levels (high and low), and the effect of motor unit activity on the surface-EMG estimate of synchronization was studied in 56 motor units. The surface-EMG estimate of motor unit synchronization was significantly higher when the reference motor unit discharged at the high rate than when it discharged at the low rate. A regression analysis indicated that the synchronization ratio calculated from the surface EMG was significantly correlated with the level of EMG activity in the muscle. Motor unit synchronization was also estimated from surface-EMG measurements that were derived by computer simulation. The simulation permitted manipulation of motor unit activity (discharge rate and recruitment) with a complete absence of synchrony among the units in the pool. The stimulated surface-EMG index was influenced by an artifact associated with signal rectification, and this effect changed non-monotonically with motor unit activity. Furthermore, the increase in the motor unit activity reduced the signal-to-noise ratio of the spike-triggered surface EMG average, and consequently decreased the sensitivity of the surface-EMG index as an estimate of motor unit synchronization. We conclude that the simplified surface-EMG method (Milner-Brown et al. 1973, 1975) does not provide a useful index of motor unit synchronization due to its inability to accurately distinguish the synchronization from methodological effects related to a rectification artifact and variation in the signal-to-noise ratio.

Artifactual averaged 'twitch tension' waveforms resulting from synchronized activity: recording from feline diaphragmatic motor units

Averaging techniques have been used to measure contractile properties of spontaneously active motor units (MUs). This study examined the potential for artifactual results due to synchronization between the triggering, single-MU action potentials, and activity of other MUs within the muscle. A muscle strip was formed in situ from feline diaphragm. Single MUs were recorded from the strip and from the contralateral diaphragm. The diaphragm including the muscle strip continue to contract rhythmically in this preparation and a high-gain, AC-coupled recording of force was averaged using MUs recorded in either hemidiaphragm to trigger the averager. Twitch-tension waveforms occurred in 42 of 49 cases triggering from spikes of MUs contained within the strip and in 13 of 19 averages triggered from contralateral MUs. The waveforms generated using contralateral MUs as triggers could only arise from synchronization with MUs contained within the diaphragmatic strip. Although twitch waveforms that were generated from external and internal triggers could appear similar qualitatively, contraction times were significantly (P less than 0.05) longer for averages using contralateral MUs. This study demonstrates that synchronization of triggering events is a major source for error in determining mechanical properties of MUs.

Coupling between human muscle spindle endings and motor units assessed using spike-triggered averaging

The electromyographic activity (EMG) generated by voluntary contraction of a muscle was averaged using the potentials from 18 identified muscle spindle afferents as a trigger. In post-spike averages of 1000-10,000 sweeps, no evidence of reflex excitation of the homonymous motoneurone pool was detected. In pre-spike averages there was no evidence of a motor-unit EMG potential that was closely correlated to the trigger spike. A single spindle afferent has only a weak reflex effect on an active motoneurone pool and must be part of a synchronized volley to affect motoneurone discharge significantly. No evidence was found for spindle activation via beta-motoneurones in weak voluntary contractions.

What do GABA neurons really do? They make possible variability generation in relation to demand

It is proposed that GABA neurons play key roles in maintaining meaningful communications within and among neural units by making possible variability generation in relation to demand. Activities of GABAergic inhibitory projection neurons from command centers and local circuit GABAergic inhibitory interneurons allow adaptive nervous system function to take place in a manner characterized by freedom without license. Through their multiple activities and connections these neurons make possible smooth transitions between modes of nervous system function over a range of increasing demands (neural pressures), enabling organisms to explore full ranges of their options.

Synchronous afferent discharge from a passive muscle of the cat: significance for interpreting spike-triggered averages

Evidence is presented for the existence of synchrony between the spike trains of muscle afferents of the passive cat medial gastrocnemius muscle held at fixed length. Synchrony between the spike trains of a population of muscle afferents was quantified by means of a synchronization index (Is), derived from spike-triggered averages of the muscle-nerve neurogram and the rectified neurogram. A previously used test based solely upon the neurogram average (Watt, Stauffer, Taylor, Reinking & Stuart, 1976) is shown to be invalid. The differences between experimentally derived Is values and theoretical Is values derived for the condition of asynchrony were compared to estimated confidence limits for those differences. This comparison revealed that twenty-two of fifty-three muscle-afferent spike trains whose rectified averages satisfied certain conditions for interpreting the Is were synchronized with the discharge of other afferents. The form of the rectified averages of another eight afferents suggested that these afferents might also have been synchronized. Synchrony in the discharge of muscle afferents was found in three experiments in which the neurogram was recorded from a single nerve branch to medial gastrocnemius, as well as in the data of experiments in which the whole muscle nerve was used. The degree of synchrony was similar for Ia, spindle group II and Ib afferents. The magnitude of the synchrony found in these experiments was judged by comparison to the results of analog simulations and the increase in Is values resulting from the application of small, quick stretches to the medial gastrocnemius muscle. The degree of synchrony found on average was approximately equivalent to that of a single spike occurring once for every four discharges of the reference spike train. Simulations were performed to determine the distortion of monosynaptic excitatory post-synaptic potentials (e.p.s.p.s) obtained by spike-triggered averaging which would be produced by synchrony between the spike trains of Ia and spindle group II afferents of the magnitude found in this study. These simulations indicate that the apparent amplitude would be increased by approximately 4 microV on average. Both the 10-90% rise time and the half-width would increase, the effects being greater for smaller e.p.s.p.s. Consequently, the synchrony found in this study is of most concern in the study of small post-synaptic potentials, such as those due to spindle group II afferents.

Detection of synchrony in the discharge of a population of neurons. II. Implementation and sensitivity of a synchronization index

This report describes the use of a synchronization index (Is; Hamm et al., 1985a) and its sensitivity to various forms and degrees of synchrony between spike trains. The dependence of the Is on signal-to-noise ratio, the number of synchronized spike trains and their degree of synchrony is shown in analog and digital simulations. These simulations and a comparison with peristimulus time histograms under conditions of induced synchrony reveal that the Is is a sensitive measure of synchronization in a population of spike trains.