Evolution of surface electromyography: From muscle electrophysiology towards neural recording and interfacing

Surface electromyography (EMG) comprises a recording of electrical activity from the body surface generated by muscle fibres during muscle contractions. Its characteristics depend on the fibre membrane potentials and the neural activation signal sent from the motor neurons to the muscles. EMG has been classically used as the primary investigation tool in kinesiology studies in a variety of applications. More recently, surface EMG techniques have evolved from single-channel methods to high-density systems with hundreds of electrodes. High-density EMG recordings can be deconvolved to estimate the discharge times of spinal motor neurons innervating the recorded muscles, with algorithms that have been developed and validated in the last two decades. Within limits and with some variability across muscles, these techniques provide a non-invasive method to study relatively large populations of motor neurons in humans. Surface EMG is thus evolving from a peripheral measure of muscle electrical activity towards a neural recording and neural interfacing signal. These advances in technology have had a major impact on our fundamental understanding of the neural control of movement and have exposed new perspectives in neurotechnologies. Here we provide an overview and perspective of modern EMG technology, as derived from past achievements, and its impact in neurophysiology and neural engineering.     

Review on electromyography signal acquisition and processing

Electromyography (EMG) is a technique for recording biomedical electrical signals obtained from the neuromuscular activities. These signals are used to monitor medical abnormalities and activation levels, and also to analyze the biomechanics of any animal movements. In this article, we provide a short review of EMG signal acquisition and processing techniques. The average efficiency of capture of EMG signals with current technologies is around 70%. Once the signal is captured, signal processing algorithms then determine the recognition accuracy, with which signals are decoded for their corresponding purpose (e.g., moving robotic arm, speech recognition, gait analysis). The recognition accuracy can go as high as 99.8%. The accuracy with which the EMG signal is decoded has already crossed 99%, and with improvements in deep learning technology, there is a large scope for improvement in the design hardware that can efficiently capture EMG signals.     

iEMG: Imaging electromyography

Advanced data analysis and visualization methodologies have played an important role in making surface electromyography both a valuable diagnostic methodology of neuromuscular disorders and a robust brain–machine interface, usable as a simple interface for prosthesis control, arm movement analysis, stiffness control, gait analysis, etc. But for diagnostic purposes, as well as for interfaces where the activation of single muscles is of interest, surface EMG suffers from severe crosstalk between deep and superficial muscle activation, making the reliable detection of the source of the signal, as well as reliable quantification of deeper muscle activation, prohibitively difficult. To address these issues we present a novel approach for processing surface electromyographic data. Our approach enables the reconstruction of 3D muscular activity location, making the depth of muscular activity directly visible. This is even possible when deep muscles are overlaid with superficial muscles, such as seen in the human forearm. The method, which we call imaging EMG (iEMG), is based on using the crosstalk between a sufficiently large number of surface electromyographic electrodes to reconstruct the 3D generating electrical potential distribution within a given area. Our results are validated by in vivo measurements of iEMG and ultrasound on the human forearm.     

Cross-correlation as a method for comparing dynamic electromyography signals during gait

Current clinical interpretation of dynamic electromyography (EMG) data is usually based on qualitative assessments of muscle timing. Cross-correlation may provide a method for objectively comparing the timing and shape of EMG signals. This study used cross-correlation to compare EMG signals from different walking trials, different test sessions, and different individuals in able-bodied adults. Cross-correlation results (R-values) for different walking trials within a single test session were high, averaging > or = 0.90 for all muscles tested (R = 1.0 indicates exact agreement). Cross-correlation values were also high among trials from different test sessions conducted by the same and different examiners (average R > or = 0.78 for all muscles). R-values were much more variable when comparing different subjects (average 0.40-0.81, range 0.00-0.91). R-values were lower for the medial hamstrings and rectus femoris compared with the other muscles tested. These results suggest that cross-correlation may be useful for evaluating changes in an individual patient's muscle activation patterns, such as before and after surgery, but not for comparing EMG patterns among different individuals, such as between patients and normative data. This is especially true for biarticular muscles such as the hamstrings and rectus femoris, which may have variable activation patterns and/or increased sensitivity to electrode placement. Cross-correlation may also be useful for identifying appropriate muscles for transfer, identifying "outlier" trials within a test session, and selecting representative EMG curves for a given patient. The advantages of cross-correlation are that it considers shape of the EMG signal in addition to timing and that the assessments it provides are objective, rather than subjective.     

Influence of fatigue on upper limb muscle activity and performance in tennis

The study examined the fatigue effect on tennis performance and upper limb muscle activity. Ten players were tested before and after a strenuous tennis exercise. Velocity and accuracy of serve and forehand drives, as well as corresponding surface electromyographic (EMG) activity of eight upper limb muscles were measured. EMG and force were also evaluated during isometric maximal voluntary contractions (IMVC). Significant decreases were observed after exercise in serve accuracy (−11.7%) and velocity (−4.5%), forehand accuracy (−25.6%) and consistency (−15.6%), as well as pectoralis major (PM) and flexor carpi radialis (FCR) IMVC strength (−13.0% and −8.2%, respectively). EMG amplitude decreased for PM and FCR in serve, forehand and IMVC, and for extensor carpi radialis in forehand. No modification was observed in EMG activation timing during strokes or in EMG frequency content during IMVC. Several hypotheses can be put forward to explain these results. First, muscle fatigue may induce a reduction in activation level of PM and forearm muscles, which could decrease performance. Second, conscious or subconscious strategies could lead to a redistribution of muscle activity to non-fatigued muscles in order to protect the organism and/or limit performance losses. Otherwise, the modifications of EMG activity could also illustrate the strategies adopted to manage the speed-accuracy trade-off in such a complex task.     

Reconstructing muscle activation during normal walking: a comparison of symbolic and connectionist machine learning techniques

One symbolic (rule-based inductive learning) and one connectionist (neural network) machine learning technique were used to reconstruct muscle activation patterns from kinematic data measured during normal human walking at several speeds. The activation patterns (or desired outputs) consisted of surface electromyographic (EMG) signals from the semitendinosus and vastus medialis muscles. The inputs consisted of flexion and extension angles measured at the hip and knee of the ipsilateral leg, their first and second derivatives, and bilateral foot contact information. The training set consisted of data from six trials, at two different speeds. The testing set consisted of data from two additional trials (one at each speed), which were not in the training set. It was possible to reconstruct the muscular activation at both speeds using both techniques. Timing of the reconstructed signals was accurate. The integrated value of the activation bursts was less accurate. The neural network gave a continuous output, whereas the rule-based inductive learning rule tree gave a quantised activation level. The advantage of rule-based inductive learning was that the rules used were both explicit and comprehensible, whilst the rules used by the neural network were implicit within its structure and not easily comprehended. The neural network was able to reconstruct the activation patterns of both muscles from one network, whereas two separate rule sets were needed for the rule-based technique. It is concluded that machine learning techniques, in comparison to explicit inverse muscular skeletal models, show good promise in modelling nearly cyclic movements such as locomotion at varying walking speeds. However, they do not provide insight into the biomechanics of the system, because they are not based on the biomechanical structure of the system.     

Immediate effects of co-contraction training on motor control of the trunk muscles in people with recurrent low back pain

Although deficits in the activation of abdominal muscles are present in people with low back pain (LBP), this can be modified with motor training. Training of deep abdominal muscles in isolation from the other trunk muscles, as an initial phase of training, has been shown to improve the timing of activation of the trained muscles, and reduce symptoms and recurrence of LBP. The aim of this study was to determine if training of the trunk muscles in a non-isolated manner can restore motor control of these muscles in people with LBP. Ten subjects with non-specific LBP performed a single session of training that involved three tasks: “abdominal curl up”, “side bridge” and “birdog”. Electromyographic activity (EMG) of trunk and deltoid muscles was recorded with fine-wire and surface electrodes during rapid arm movements and walking, before and immediately following the intervention. Onset of trunk muscle EMG relative to that of the prime mover (deltoid) during arm movements and the mean, standard deviation (SD) and coefficient of variation of abdominal muscle EMG during walking were calculated. There was no significant change in the times of onset of trunk muscle EMG during arm movements nor was there any change in the variability of EMG of the abdominal muscles during walking. However, the mean amplitude and SD of abdominal EMG was reduced during walking after training. The results of this study suggest that unlike isolated voluntary training, co-contraction training of the trunk muscles does not restore the motor control of the deep abdominal muscles in people with LBP after a single session of training.     

Linear relationship between electromyography and shear wave elastography measurements persists in deep muscles of the upper extremity

Recent works have demonstrated a linear relationship between muscle activation and shear modulus in various superficial muscles. As such, it may be possible to overcome limitations of traditional electromyography (EMG) methods by assessing activation using shear wave elastography. However, the relationship has not been wholly validated in deep muscles. This study measured the association between squared shear wave velocity, which is related to shear modulus, and activation within superficial and deep muscles. This relationship was also compared between surface and intramuscular EMG electrodes. We simultaneously recorded EMG and shear wave velocity in one deep (brachialis) and one superficial (brachioradialis) muscle in ten healthy individuals during isometric elbow flexion across a wide range of contraction intensities. Muscle activation and squared shear wave velocity demonstrated good reliability (ICC > 0.75) and showed a linear relationship (P < 0.05) for all muscle/EMG electrode type combinations (study conditions) after down-sampling. Study condition was not a significant within-subject factor to the slope or intercept of the relationship (P > 0.05). This work demonstrates that activation of both superficial and deep muscles can be assessed noninvasively using ultrasound shear wave elastography and is a critical step toward demonstrating elastography’s utility as an alternative to EMG.     

The extraction of neural strategies from the surface EMG.

This brief review examines some of the methods used to infer central control strategies from surface electromyogram (EMG) recordings. Among the many uses of the surface EMG in studying the neural control of movement, the review critically evaluates only some of the applications. The focus is on the relations between global features of the surface EMG and the underlying physiological processes. Because direct measurements of motor unit activation are not available and many factors can influence the signal, these relations are frequently misinterpreted. These errors are compounded by the counterintuitive effects that some system parameters can have on the EMG signal. The phenomenon of crosstalk is used as an example of these problems. The review describes the limitations of techniques used to infer the level of muscle activation, the type of motor unit recruited, the upper limit of motor unit recruitment, the average discharge rate, and the degree of synchronization between motor units. Although the global surface EMG is a useful measure of muscle activation and assessment, there are limits to the information that can be extracted from this signal.     

Application of singular spectrum-based change-point analysis to EMG-onset detection

While many approaches have been proposed to identify the signal onset in EMG recordings, there is no standardized method for performing this task. Here, we propose to use a change-point detection procedure based on singular spectrum analysis to determine the onset of EMG signals. This method is suitable for automated real-time implementation, can be applied directly to the raw signal, and does not require any prior knowledge of the EMG signal’s properties. The algorithm proposed by Moskvina and Zhigljavsky (2003) was applied to EMG segments recorded from wrist and trunk muscles. Wrist EMG data was collected from 9 Parkinson’s disease patients with and without tremor, while trunk EMG data was collected from 13 healthy able-bodied individuals. Along with the change-point detection analysis, two threshold-based onset detection methods were applied, as well as visual estimates of the EMG onset by trained practitioners. In the case of wrist EMG data without tremor, the change-point analysis showed comparable or superior frequency and quality of detection results, as compared to other automatic detection methods. In the case of wrist EMG data with tremor and trunk EMG data, performance suffered because other changes occurring in these signals caused larger changes in the detection statistic than the changes caused by the initial muscle activation, suggesting that additional criteria are needed to identify the onset from the detection statistic other than its magnitude alone. Once this issue is resolved, change-point detection should provide an effective EMG-onset detection method suitable for automated real-time implementation.     

Relating neuromuscular control to functional anatomy of limb muscles in extant archosaurs

Electromyography (EMG) is used to understand muscle activity patterns in animals. Understanding how much variation exists in muscle activity patterns in homologous muscles across animal clades during similar behaviours is important for evaluating the evolution of muscle functions and neuromuscular control. We compared muscle activity across a range of archosaurian species and appendicular muscles, including how these EMG patterns varied across ontogeny and phylogeny, to reconstruct the evolutionary history of archosaurian muscle activation during locomotion. EMG electrodes were implanted into the muscles of turkeys, pheasants, quail, guineafowl, emus (three age classes), tinamous and juvenile Nile crocodiles across 13 different appendicular muscles. Subjects walked and ran at a range of speeds both overground and on treadmills during EMG recordings. Anatomically similar muscles such as the lateral gastrocnemius exhibited similar EMG patterns at similar relative speeds across all birds. In the crocodiles, the EMG signals closely matched previously published data for alligators. The timing of lateral gastrocnemius activation was relatively later within a stride cycle for crocodiles compared to birds. This difference may relate to the coordinated knee extension and ankle plantarflexion timing across the swing-stance transition in Crocodylia, unlike in birds where there is knee flexion and ankle dorsiflexion across swing-stance. No significant effects were found across the species for ontogeny, or between treadmill and overground locomotion. Our findings strengthen the inference that some muscle EMG patterns remained conservative throughout Archosauria: for example, digital flexors retained similar stance phase activity and M. pectoralis remained an ‘anti-gravity’ muscle. However, some avian hindlimb muscles evolved divergent activations in tandem with functional changes such as bipedalism and more crouched postures, especially M. iliotrochantericus caudalis switching from swing to stance phase activity and M. iliofibularis adding a novel stance phase burst of activity.     

Comparing electro- and mechano-myographic muscle activation patterns in self-paced pediatric gait

Electromyography (EMG) is the standard modality for measuring muscle activity. However, the convenience and availability of low-cost accelerometer-based wearables makes mechanomyography (MMG) an increasingly attractive alternative modality for clinical applications. Literature to date has demonstrated a strong association between EMG and MMG temporal alignment in isometric and isokinetic contractions. However, the EMG-MMG relationship has not been studied in gait. In this study, the concurrence of EMG- and MMG-detected contractions in the tibialis anterior, lateral gastrocnemius, vastus lateralis, and biceps femoris muscles were investigated in children during self-paced gait. Furthermore, the distribution of signal power over the gait cycle was statistically compared between EMG-MMG modalities. With EMG as the reference, muscular contractions were detected based on MMG with balanced accuracies between 88 and 94% for all muscles except the gastrocnemius. MMG signal power differed from that of EMG during certain phases of the gait cycle in all muscles except the biceps femoris. These timing and power distribution differences between the two modalities may in part be related to muscle fascicle length changes that are unique to muscle motion during gait. Our findings suggest that the relationship between EMG and MMG appears to be more complex during gait than in isometric and isokinetic contractions.     

Modular neuromuscular control of human locomotion by central pattern generator

The central pattern generators (CPG) in the spinal cord are thought to be responsible for producing the rhythmic motor patterns during rhythmic activities. For locomotor tasks, this involves much complexity, due to a redundant system of muscle actuators with a large number of highly nonlinear muscles. This study proposes a reduced neural control strategy for the CPG, based on modular organization of the co-active muscles, i.e., muscle synergies. Four synergies were extracted from the EMG data of the major leg muscles of two subjects, during two gait trials each, using non-negative matrix factorization algorithm. A Matsuoka׳s four-neuron CPG model with mutual inhibition, was utilized to generate the rhythmic activation patterns of the muscle synergies, using the hip flexion angle and foot contact force information from the sensory afferents as inputs. The model parameters were tuned using the experimental data of one gait trial, which resulted in a good fitting accuracy (RMSEs between 0.0491 and 0.1399) between the simulation and experimental synergy activations. The model׳s performance was then assessed by comparing its predictions for the activation patterns of the individual leg muscles during locomotion with the relevant EMG data. Results indicated that the characteristic features of the complex activation patterns of the muscles were well reproduced by the model for different gait trials and subjects. In general, the CPG- and muscle synergy-based model was promising in view of its simple architecture, yet extensive potentials for neuromuscular control, e.g., resolving redundancies, distributed and fast control, and modulation of locomotion by simple control signals.     

Electromyogram recordings from freely moving animals

Electromyography can be used to record activity from sets of muscles in awake, freely moving animals using implanted intramuscular electrodes. As a tool, EMG has a wide range of applications ranging from inferring neural processes to analyzing movement. The amplitude of the rectified and filtered electromyogram (EMG) can be used as an indirect measure of muscle activity. Although it is often tempting to correlate the EMG with muscle force, the fact that force varies more with different activation strategies than with EMG estimates must be taken into account. The purpose of this article is to provide the researcher wishing to introduce the technique of recording EMGs from conscious animals using intramuscular electrodes with a step-by-step guide. It includes details on the manufacture of electromyograph electrodes, recording, and analysis considerations along with a section on solving common problems. For the sake of clarity, this article focuses on using the cat as a model and on the implantation of hindlimb muscles with intramuscular wire electrodes. However, the procedures can be adapted for use on other striated muscles and species.     

Electromyography variables during the golf swing: A literature review

The aim of the study was to review systematically the literature available on electromyographic (EMG) variables of the golf swing. From the 19 studies found, a high variety of EMG methodologies were reported. With respect to EMG intensity, the right erector spinae seems to be highly activated, especially during the acceleration phase, whereas the oblique abdominal muscles showed moderate to low levels of activation. The pectoralis major, subscapularis and latissimus dorsi muscles of both sides showed their peak activity during the acceleration phase. High muscle activity was found in the forearm muscles, especially in the wrist flexor muscles demonstrating activity levels above the maximal voluntary contraction. In the lower limb higher muscle activity of the trail side was found. There is no consensus on the influence of the golf club used on the neuromuscular patterns described. Furthermore, there is a lack of studies on average golf players, since most studies were executed on professional or low handicap golfers.Further EMG studies are needed, especially on lower limb muscles, to describe golf swing muscle activation patterns and to evaluate timing parameters to characterize neuromuscular patterns responsible for an efficient movement with lowest risk for injury.     

Trunk muscle onset detection technique for EMG signals with ECG artefact.

The timing of trunk muscle activation has become an important element in the understanding of human movement in normal and chronic low back pain populations. The detection of anticipatory postural adjustment via trunk muscle onsets from electromyographic (EMG) signals can be problematic due to baseline noise or electro-cardiac (ECG) artefact. Shewhart protocols or whole signal analyses may show different degrees of sensitivity under different conditions.Muscle activity onsets were determined from surface EMG of seven muscles for five trials before and after fatigue were examined in four subjects (n=280). The objective of this study was to examine two detection methods (Shewhart and integrated protocol (IP)) in determining the onsets of trunk muscles. The variability of the baseline amplitude and the impact of added Gaussian noise on the detected onsets were used to test for robustness.The results of this study demonstrate that before and after fatigue there is a large degree of baseline variance in the trunk muscles (coefficients of variation between 40-65%) between trials. This could be normal response to body sway. The IP method was less susceptible to false onsets (detecting onsets in the baseline window) 3 vs. 51%. The findings suggest the IP method is robust with large variance in the baseline if the signal to noise ratio is greater than six.In spite of the robustness of the algorithm, the findings would suggest that statistical assessments should be used to target trials for selective visual inspection for subtle trunk muscle onsets.     

Robust Muscle Activity Onset Detection Using an Unsupervised Electromyogram Learning Framework

Accurate muscle activity onset detection is an essential prerequisite for many applications of surface electromyogram (EMG). This study presents an unsupervised EMG learning framework based on a sequential Gaussian mixture model (GMM) to detect muscle activity onsets. The distribution of the logarithmic power of EMG signal was characterized by a two-component GMM in each frequency band, in which the two components respectively correspond to the posterior distribution of EMG burst and non-burst logarithmic powers. The parameter set of the GMM was sequentially estimated based on maximum likelihood, subject to constraints derived from the relationship between EMG burst and non-burst distributions. An optimal threshold for EMG burst/non-burst classification was determined using the GMM at each frequency band, and the final decision was obtained by a voting procedure. The proposed novel framework was applied to simulated and experimental surface EMG signals for muscle activity onset detection. Compared with conventional approaches, it demonstrated robust performance for low and changing signal to noise ratios in a dynamic environment. The framework is applicable for real-time implementation, and does not require the assumption of non EMG burst in the initial stage. Such features facilitate its practical application.     

The role of the cerebellum in modulating voluntary limb movement commands

We recorded the activity of cerebellar Purkinje cells (PCs), primary motor cortical (M1) neurons, and limb EMG signals while monkeys executed a sequential reaching and button pressing task. PC simple spike discharge generally correlated well with the activity of one or more forelimb muscles. Surprisingly, given the inhibitory projection of PCs, only about one quarter of the correlations were negative. The largest group of neurons burst during movement and were positively correlated with EMG signals, while another significant group burst and were negatively correlated. Among the PCs that paused during movement most were negatively correlated with EMG. The strength of these various correlations was somewhat weaker, on average, than equivalent correlations between M1 neurons and EMG signals. On the other hand, there were no significant differences in the timing of the onset of movement related discharge among these groups of PCs, or between the PCs and M1 neurons. PC discharge was modulated largely in phase, or directly out of phase, with muscle activity. The nearly synchronous activation of PCs and muscles yielded positive correlations, despite the fact that the synaptic effect of the PC discharge is inhibitory. The apparent function of this inhibition is to restrain activity in the limb premotor network, shaping it into a spatiotemporal pattern that is appropriate for controlling the many muscles that participate in this task. The observed timing suggests that the cerebellar cortex learns to modulate PC discharge predictively. Through the cerebellar nucleus, this PC signal is combined with an underlying cerebral cortical signal. In this manner the cerebellum refines the descending command as compared with the relatively crude version generated when the cerebellum is damaged.     

Estimation of monopolar signals from sphincter muscles and removal of common mode interference

The detection of surface electromyogram (EMG) by multi-electrode systems is applied in many research studies. The signal is usually recorded by means of spatial filters (linear combination of the potential under at least two electrodes) with vanishing sum of weights. Nevertheless, more information could be extracted from monopolar signals measured with respect to a reference electrode away from the muscle. Under certain conditions, surface EMG signal along a curve parallel to the fibre path has zero mean (property approximately satisfied when EMG is sampled by an array of electrodes that covers the entire support of the signal in space). This property allows estimating monopolar from single differential (SD) signals by pseudoinversion of the matrix relating monopolar to SD signals. The method applies to EMG signals from the external anal sphincter muscle, recorded using a specific cylindrical probe with an array of electrodes located along the circular path of the fibres. The performance of the algorithm for the estimation of monopolar from SD signals is tested on simulated signals. The estimation error of monopolar signals decreases by increasing the number of channels. Using at least 12 electrodes, the estimation error is negligible. The method applies to single fibre action potentials, single motor unit action potentials, and interference signals.The same method can also be applied to reduce common mode interference from SD signals from muscles with rectilinear fibres. In this case, the last SD channel defined as the difference between the potentials of the last and the first electrodes must be recorded, so that the sum of all the SD signals vanishes. The SD signals estimated from the double differential signals by pseudoinvertion are free of common mode.