Sample entropy analysis of surface EMG for improved muscle activity onset detection against spurious background spikes

Voluntary surface electromyogram (EMG) signal is sometimes contaminated by spurious background spikes of both physiological and extrinsic or accidental origins. A novel method of muscle activity onset detection against such spurious spikes was proposed in this study based primarily on the sample entropy (SampEn) analysis of the surface EMG. The method takes advantage of the nonlinear properties of the SampEn analysis to distinguish voluntary surface EMG signals from spurious background spikes in the complexity domain. To facilitate muscle activity onset detection, the SampEn analysis of surface EMG was first performed to highlight voluntary EMG activity while suppressing spurious background spikes. Then, a SampEn threshold was optimized for muscle activity onset detection. The performance of the proposed method was examined using both semi-synthetic and experimental surface EMG signals. The SampEn based methods effectively reduced the detection error induced by spurious background spikes and achieved improved performance over the methods relying on conventional amplitude thresholding or its extended version in the Teager Kaiser Energy domain.     

A pattern recognition technique to characterize the differential modulation of co-activating muscles at the performer/environment interface.

The purpose of this research was to develop and test an analytical tool that would recognize and classify the surface electromyographic (EMG) signal of co-activating muscles of the leg into pre-defined patterns of muscle activity: burst, tonic, and tonic-burst. Developed to study the task of landing from a jump in children, the pattern recognition technique (PRT) quantifies the full-wave rectified surface EMG signal over a short-duration sampling window by a single linear regression value. Shifting the sampling window across the data string ultimately defines the signal by a set of regression values that produce the recognizable burst, tonic and tonic-burst patterns on a least-squares surface plot. Statistical comparison of the PRT to the classical combination of threshold detection (+2 S.D. of mean baseline activity) and visual inspection proves the PRT to be more reliable on repeated measures for event detection and classification, with a Kappa statistic of 0.83 compared to 0.54 for threshold detection. Application of the PRT to motor control studies is presented for the regulation of the mechanical response of the leg during impact. Responsiveness of the PRT is tested, issues of accuracy and validity are addressed, and limitations in spatial-temporal resolution are identified.     

Backpack load affects lower limb muscle activity patterns of female hikers during prolonged load carriage

This study investigated the effect of prolonged load carriage on lower limb muscle activity displayed by female recreational hikers. Electromyography (EMG) signals from vastus lateralis (VL), biceps femoris (BF), semitendinosus (ST), tibialis anterior (TA) and gastrocnemius (GM) were recorded for fifteen female hikers carrying four loads (0%, 20%, 30% and 40% body weight (BW)) over 8 km. Muscle burst duration, muscle burst onset relative to initial contact and integrated EMG signals (iEMG) were calculated to evaluate muscle activity, whereas the shift in mean power frequency (MPF) was used to evaluate muscle fatigue. Increased walking distance significantly decreased the MPF of TA; decreased the iEMG for VL, ST and GM; and shortened VL muscle burst duration. Furthermore, carrying 20–40% BW loads significantly increased VL and GM iEMG and increased BF muscle burst duration, whereas a 40% BW load caused a later VL muscle burst onset. The differences observed in muscle activity with increased load mass seem to be adjustments aimed at maintaining balance and attenuating the increased loads placed on the lower limbs during gait. Based on the changes in muscle activity, a backpack load limit of 30% BW may reduce the risk of lower limb injury for female hikers during prolonged walking.     

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.     

Neural muscle activation detection: A deep learning approach using surface electromyography

The timing of muscles activation which is a key parameter in determining plenty of medical conditions can be greatly assessed by the surface EMG signal which inherently carries an immense amount of information. Many techniques for measuring muscle activity detection exist in the literature. However, due to the complex nature of the EMG signal as well as the interference from other muscles that is observed during the measurement of the EMG signal, the accuracy of these techniques is compromised. In this paper, we introduce the neural muscle activation detection (NMAD) framework that detects the muscle activation based on deep learning. The main motivation behind using deep learning is to allow the neural network to detect based on the appropriate signal features instead of depending on certain assumptions. Not only the presented approach significantly improves the accuracy of timing detection, but because of the training nature, it can adapt to operate under different levels of interference and signal-to-noise ratio.     

Influence of amplitude cancellation on the accuracy of determining the onset of muscle activity from the surface electromyogram

The purpose of the study was to quantify the influence of amplitude cancellation on the accuracy of detecting the onset of muscle activity based on an analysis of simulated surface electromyographic (EMG) signals. EMG activity of a generic lower limb muscle was simulated during the stance phase of human gait. Surface EMG signals were generated with and without amplitude cancellation by summing simulated motor unit potentials either before (cancellation EMG) or after (no-cancellation EMG) the potentials had been rectified. The two sets of EMG signals were compared at forces of 30% and 80% of maximum voluntary contraction (MVC) and with various low-pass filter cut-off frequencies. Onset time was determined both visually and by an algorithm that identified when the mean amplitude of the signal within a sliding window exceeded a specified standard deviation (SD) above the baseline mean. Onset error was greater for the no-cancellation conditions when determined automatically and by visual inspection. However, the differences in onset error between the two cancellation conditions appear to be clinically insignificant. Therefore, amplitude cancellation does not appear to limit the ability to detect the onset of muscle activity from the surface EMG.     

Knee stability and muscle coordination in patients with anterior cruciate ligament injuries: An electromyographic approach

We present findings on the way in which to use electromyographic (EMG) measurements from muscles acting on the knee in planning rehabilitation of subjects after rupture of anterior cruciate ligament (ACL). ACL subjects demonstrated an earlier recruitment and a tendency to prolonged activity in muscles around the deficient knee as compared with a control group. Especially the hamstring lateralis and the gastrocnemius medialis (GM) muscles showed an earlier EMG onset and a longer EMG burst duration. The clinical relevance of the EMG findings was assessed by comparing the muscle coordination and relative levels of activity between a functionally excellent/good and a functionally poor ACL patient group. Significant differences between the two groups were noted in EMG onset and burst duration of the GM muscle. A rehabilitation program based on the EMG findings from the GM muscle was designed. In this program, the ACL subjects with poor stability were trained to change the EMG activity of the gastrocnemius muscles according to the recruitment pattern of the good/excellent ACL-group. We were able to train the subjects to change their muscle recruitment and to improve their knee stability. The stability of the knee joint depends on the stiffness of the muscles and ligaments around and within the knee. We discuss the importance of the gastrocnemius muscles with regard to knee joint stiffness.     

Separation of myosin subfragment 1 into two fractions, one having the burst site and the other having the non-burst site.

During Mn(II)-ATP hydrolysis by myosin, the predominant intermediate formed at the burst site of the enzyme below 10 degrees is the myosin-ADP complex formed by adding ADP to myosin, while above 10 degrees it is the myosin -ADP-P1 complex generated by ATP hydroolysis (Yazawa, Morita, & Yagi (1973) J. Biochem. 74, 1107; Hozumi & Tawada (1975) Biochim. Biophys. Acta 376, 1; Tawada & Yoshida (1975) J. Biochem. 78, 293). It is suggested that the second (non-burst) site of myosin predominantly forms the myosin-ATP complex (Hozumi & Tawada, ibid.). From these findings, it is expected that (i) myosin subfragment 1 (S1) having the burst site is bound to actin in Mn(II)-ATP solution containing ADP below 10 degrees, because it forms the S1-ADP complex even in the presence of ATP; (ii) the other S1, i.e., that having the non-burst site, is dissociated from actin, because it forms the S1-ATP complex. These two expectations were confirmed by viscosity measurements of acto-S1 solutions, giving a basis for the separation of S1 into two fractions: one having the burst site and the other having the non-burst site. S1 having the non-burst site could be extracted from partially papain [EC 3.4.22.2]-digested myofibrils of rabbit skeletal muscle with a solution containing MnCl2, ATP, and ADP at 0 degrees. S1 having the burst site was extracted from myofibrils already used for the extraction of S1 having the non-burst site, with a solution containing MgCl2 and ATP at 20 degrees. The former S1 fraction had Mg-ATPase [EC 3.6.1.3] activity, but scarcely showed any initial burst of Pi liberation. The latter S1 showed a Pi burst of more than 0.5 (M/M). The steady state ATPase activity of the former S1 was slightly higher than that of the latter. The burst size of normal S1, i.e., that extracted from papain-digested myofibrils with Mg-PPi or Mg-ATP, was 0.5 (M/M). The ultraviolet absorption spectrum of the non-burst type S1 was not changed by ADP but was changed by ATP, though the difference spectrum was distinct from that of normal S1 and the difference molar extinction coefficient at 289 nm was only 20% of that of normal S1. No significant difference was seen in the compositions of these two S1's and normal S1, as determined by SDS gel electrophoresis.     

Electromyographic instantaneous amplitude and instantaneous mean power frequency patterns across a range of motion during a concentric isokinetic muscle action of the biceps brachii.

The purpose of this study was to examine the electromyographic (EMG) instantaneous amplitude (IA) and instantaneous mean power frequency (IMPF) patterns for the biceps brachii muscle across a range of motion during maximal and submaximal concentric isokinetic muscle actions of the forearm flexors. Ten adults (mean +/- SD age = 22.0 +/- 3.4 years) performed a maximal and a submaximal [20% peak torque (PT)] concentric isokinetic forearm flexion muscle action at a velocity of 30 degrees s(-1). The surface EMG signal was detected from the biceps brachii muscle with a bipolar electrode arrangement, and the EMG IA and IMPF versus time relationships were examined for each subject using first- and second-order polynomial regression models. The results indicated that there were no consistent patterns between subjects for EMG IA or IMPF with increases in torque across the range of motion. Some of the potential nonphysiological factors that could influence the amplitude and/or frequency contents of the surface EMG signal during a dynamic muscle action include movement of the muscle fibers and innervation zone beneath the skin surface, as well as changes in muscle fiber length and the thickness of the tissue layer between the muscle and the recording electrodes. These factors may affect the EMG IA and IMPF patterns differently for each subject, thereby increasing the difficulty of drawing any general conclusions regarding the motor control strategies that increase torque across a range of motion.     

Relationships between muscle lactate accumulation and surface EMG activities during isokinetic contractions in man

In order to investigate the relationship between metabolic state and myoelectrical activity in working muscle during short term intense exercise, eleven healthy males performed isokinetic knee extensions at an angular velocity of 180 deg X sec-1 for 30 and 60 s. The median frequency (MF) of the surface electromyogram (EMG) recorded from vastus lateralis was decreased while the time lag of torque production after the onset of electrical activity (EMD) was increased during exercise. These changes (MF and EMD) corresponded well to muscle lactate accumulation in the same muscle. Over the exercise period, the integrated EMG/knee extension peak torque ratio (E/T ratio) was increased, which indicated a decrease in the efficiency of electrical activity. It was concluded that the changes in the frequency components of the EMG and in the contractile property of the muscle during short term intense exercise correlated with lactate accumulation in the identical muscle, and that the decrease in efficiency of the electrical activity in the muscle suggested peripheral fatigue.     

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.     

Surface myoelectric signals during ergocycle exrcises at various mechanical powers and pedalling rates

The surface myoelectric (ME) signal from the vastus lateralis muscle was studied during ergocycling at various mechanical powers (MPs) and pedalling rates (PRs). The envelope of the signal was described using a power spectral density function estimate at very low frequency. A highly smoothed burst pattern was found independently at 80 rpm on MP. The envelope presence in surface ME signals was demonstrated to affect the total band power spectrum estimate slightly. In addition, the total band power spectrum changes with MPs and PRs were described using the root mean square value and the mean power frequency. Total band power spectrum was enhanced and shifted toward the low frequencies as PR increased. It was shifted toward the high frequencies as MP increased. These changes were attributed to the progressive recruitment of fast fibers.     

Muscle tuning during running: implications of an un-tuned landing

BACKGROUND: The impact force in heel-toe running is an input signal into the body that initiates vibrations of the soft tissue compartments of the leg. These vibrations are heavily damped and the paradigm of muscle tuning suggests the body adapts to different input signals to minimize these vibrations. The objectives of the present study were to investigate the implications of not tuning a muscle properly for a landing with a frequency close to the resonance frequency of a soft tissue compartment and to look at the effect of an unexpected surface change on the subsequent step of running. METHOD: Thirteen male runners were recruited and performed heel-toe running over two surface conditions. The peak accelerations and biodynamic responses of the soft tissue compartments of the leg along with the EMG activity of related muscles were determined for expected soft, unexpected hard and expected hard landings. RESULTS AND CONCLUSIONS: For the unexpected hard landing there was a change in the input frequency of the impact force, shifting it closer to the resonance frequency of the soft tissue compartments. For the unexpected landing there was no muscle adaptation, as subjects did not know the running surface was going to change. In support of the muscle-tuning concept an increase in the soft tissue acceleration did occur. This increase was greater when the proximity of the input signal frequency was closer to the resonance frequency of the soft tissue compartment. Following the unexpected change in the input signal a change in pre-contact muscle activity to minimize soft tissue compartment vibrations was not found. This suggests if muscle tuning does occur it is not a continuous feedback response that occurs with every small change in the landing surface properties. In previous studies with significant adaptation periods to new input signals significant correlations between the changes in the input signal frequency and the EMG intensity have been shown, however, changes in soft tissue accelerations have not been found. The results of the present study showed that changes in these soft tissue accelerations can occur in response to a resonance frequency input signal when a muscle reaction has not happened.     

EMG monitoring in functional electrostimulation]

When using functional electrical stimulation (FES), correct adjustment of stimulation parameters, and monitoring of the stimulated muscle is mandatory if tissue damage is to be avoided. Although several FES systems are already in regular use, a method for direct muscle monitoring is still lacking. This paper investigates the suitability of the electromyogram (EMG) for such a purpose. In six sheep, the right latissimus dorsi muscle (LDM) and the associated thoracodorsal nerve were exposed. Stimulation was effected via electrodes placed on the nerve. Three electrodes were placed in the LDM for EMG recording, and the tendon was connected to a force transducer for isometric force measurement. Stimulation was applied for one second (burst), followed by a three-second pause. The stimulation current was increased in 0.2 mA steps, starting at 0 mA and ending at 4 mA. Throughout the investigation, the EMG signal was monitored with an oscilloscope. In addition, the EMG signal and the force transducer signal were recorded for subsequent analysis. An analysis of the data of all six sheep revealed an almost linear relationship between muscle force and m-wave amplitude (magnitude of r = 0.95, p < 0.001). M-wave monitoring during EMG recording with three intramuscular electrodes is a reliable method of monitoring FES-induced muscle activity, but the absolute force cannot be measured.     

Red muscle recruitment during steady swimming correlates with rostral-caudal patterns of power production in trout

Rainbow trout (Oncorhynchus mykiss) and brook trout (or charr, Salvelinus fontinalis) display different rostral-caudal patterns of power production by the red or aerobic muscle during steady swimming. The anterior muscle of rainbow trout produces much less power for swimming than the posterior, while in brook trout there is no variation in power output. To determine if red muscle recruitment is associated with anterior-posterior patterns of power production, electromyography (EMG) was used to record red muscle activity at three body positions across a range of swimming speeds in fish of each species. The initial recruitment of the anterior red muscle in swimming rainbow trout was predicted to lag behind, i.e. occur at higher speeds, that of the posterior due to the variation in power production, but no variation in recruitment was expected for brook trout. Burst of red muscle EMG activity occurring with each tailbeat was analyzed for frequency (tailbeat frequency), duty cycle (DC) (duration of burst relative to the period of the tailbeat) and burst intensity (BI) (magnitude of the measured EMG activity). Brook trout swam with higher tailbeat frequencies and longer values of DC than rainbow trout. Both species showed a pattern of longitudinal variation in DC, with longer DC values in the anterior red muscle. BI also differed significantly along the length of rainbow trout but not brook trout. In the former, BI of anterior muscle was significantly less than the posterior at lower steady swimming speeds. The EMG data suggest that power production and muscle recruitment are related. In rainbow trout, where there is longitudinal variation in muscle power output, there are also significant rostral-caudal differences in red muscle recruitment.     

Cross-talk from adjacent muscle has a negligible effect on surface electromyographic activity of vastus intermedius muscle during isometric contraction

The purposes of this study were to attempt to record surface electromyography (EMG) from the superficial region of vastus intermedius (VI) and to investigate the influence of adjacent muscle activity on surface EMG of VI. First, serial axial magnetic resonance imaging of the thigh was performed for 45 healthy young men to determine morphological characteristics of the VI. Second, surface EMG activity of the VI and other quadriceps femoris (QF) muscle group components were recorded in maximum voluntary contraction during isometric knee extension from 11 healthy young men. To test cross-talk of EMG signals between VI and the nearest adjacent muscle, vastus lateralis (VL), we applied cooling for 20-min on VL to selectively alter activity. Cooling the skin above a muscle is known to decrease median frequency (MF) of EMG signal of the muscle. All subjects displayed a superficial region in VI sufficiently large (14 cm²) to record surface EMG. Surface EMG of VI could be detected in the same scale as other QF muscle group components. Cooling induced a significant MF decrease only in VL (from 92.5 to 44.2 Hz, p < 0.001), but no significant change was observed in VI (from 63.8 to 61.7 Hz). From this result, we concluded the muscle activity of VL is negligible on surface EMG detected from VI during isometric contraction.     

Critical band in auditory lateralization.

A new and powerful procedure for determining frequency analysis in the auditory system, as evidence by the critical band, is described. The onset time difference, delta T, needed to lateralize 30-msec tone bursts toward the leading ear was measured as a function of the frequency difference, delta F, between the brust in one ear and the burst in the other ear. When delta F was less than the critical band, threshold delta T was constant at 100 mu sec or less, depending on center frequency; beyond the critical band, delta T increased with delta F. These dichotically measured critical bandwidths increased from 110 Hz at a center frequency of 500 Hz to 1100 Hz at a center frequency of 6000 Hz. They were unaffected by varying signal level from 25 to 80 dB or signal duration from 10 to 300 msec. The sam e critical-band values have been measured with monaural stimuli in loudness summation, maskin, detection, phase perception, consonance, and so forth.     

Detection of motor unit action potentials with surface electrodes: influence of electrode size and spacing

A model of the motor unit action potential was developed to investigate the amplitude and frequency spectrum contributions of motor units, located at various depths within muscle, to the surface detected electromyographic (EMG) signal. A dipole representation of the transmembrane current in a three-dimensional muscle volume was used to estimate detected individual muscle fiber action potentials. The effects of anisotropic muscle conductance, innervation zone location, propagation velocity, fiber length, electrode area, and electrode configuration were included in the fiber action potential model. A motor unit action potential was assumed to be the sum of the individual muscle fiber action potentials. A computational procedure, based on the notion of isopotential layers, was developed which substantially reduced the calculation time required to estimate motor unit action potentials. The simulations indicated that: 1) only those motor units with muscle fibers located within 10–12 mm of the electrodes would contribute significant signal energy to the surface EMG, 2) variation in surface area of electrodes has little effect on the detection depth of motor unit action potentials, 3) increased interelectrode spacing moderately increases detection depth, and 4) the frequency content of action potentials decreases steeply with increased electrode-motor unit territory distance.     

Time-frequency analysis of surface electromyographic signals during fatiguing isokinetic muscle actions

The purpose of this study was to use a wavelet analysis designed specifically for electromyography (EMG) signals in combination with a trend plot to examine changes in EMG intensity patterns during maximal, fatiguing isokinetic muscle actions. Eleven men (mean ± SD age = 22.4 ± 1.1 years) and 7 women (mean ± SD age = 22.7 ± 2.1 years) performed 50 consecutive maximal concentric isokinetic muscle actions of the dominant leg extensors at a velocity of 180°·s(-1). During each muscle action, a bipolar surface EMG signal was detected from the vastus lateralis. All signals were then processed with a wavelet analysis designed specifically for EMG signals, which resulted in EMG intensity patterns. The patterns for each subject were then analyzed with a trend plot, which provided information regarding the changes that occurred because of fatigue. The results indicated that for all the 18 subjects, the EMG intensity patterns moved in a predictable manner in pattern space, but the changes to the patterns were different for each subject. These findings reflect the complex changes that occur in the EMG signal during fatigue. These changes cannot be characterized fully with a single amplitude and center frequency parameter and can be useful for athletes and coaches who need to track the fatigue status of individual muscles.     

Muscle activity in the leg is tuned in response to impact force characteristics

Based on results from quasi-static experiments, it has been suggested that the lower extremity muscle activity is adjusted in reaction to impact forces with the goal of minimizing soft-tissue vibrations. It is not known whether a similar muscle tuning occurs during dynamic activities. Thus, the purpose of this study was to determine the effect of changes in the input signal on (a) vibrations of lower extremity soft-tissue packages and (b) EMG activity of related muscles during heel-toe running. Subjects performed heel-toe running in five different shoe conditions. Ground reaction forces were measured with a KISTLER force platform, soft-tissue vibrations were measured with tri-axial accelerometers and muscle activity was measured using surface EMG from the quadriceps, hamstrings, tibialis anterior and triceps surae groups from 10 subjects. By changing both the speed of running and the shoe midsole material the impact force characteristics were changed. There was no effect of changes in the input signal on the soft-tissue peak acceleration following impact. A significant correlation (R2=0.819) between the EMG pre-activation intensity and the impact loading rate changes was found for the quadriceps. In addition, the input frequency was shown to approach the vibration frequency of the quadriceps. This evidence supports the proposed paradigm that muscle activity is tuned to impact force characteristics to control the soft-tissue vibrations.