Automatic analysis of the electromyographic interference pattern. Part I: Development of quantitative features

We have developed three new features of the electromyographic interference pattern (IP), based on the turns and amplitude of the signal, to quantitate some of the features of the IP that are usually assessed subjectively by an electromyographer. The activity measures the fullness of the IP. The upper centile amplitude (UCA) defines the upper limit of the maximum peak-to-peak amplitude of the motor unit action potentials (MUAPs) contained in the IP. The number of small segments (NSS) measures the complexity of the IP, which is a reflection of the polyphasicity of the component MUAPs. The activity and the logarithm of the UCA correlate strongly with the force of muscle contraction at which the IP is measured. The NSS initially increases with the force of contraction and becomes relatively constant at higher force levels. The normal values of these features and the interpretation of their relationships are described in companion papers.     

Simulation and analysis of the electromyographic interference pattern in normal muscle. Part I: Turns and amplitude measurements

The electromyographic (EMG) interference pattern (IP) was simulated by adding together motor unit action potentials (MUAPs) of different sizes that had been recorded by a concentric needle EMG electrode. The number of turns (NT) of the simulated IP increased with the number of MUAP discharges. The mean amplitude (MA) difference between successive turns in the IP increased when large amplitude MUAPs were added. Our analysis demonstrates that the MA of the IP is determined mainly by the amplitude of large MUAPs in the signal and that large amplitude spikes are more likely to be generated by single large amplitude MUAPs than by summation of several small amplitude MUAPs.     

Automatic analysis of the electromyographic interference pattern. Part II: Findings in control subjects and in some neuromuscular diseases

The electromyographic (EMG) interference pattern (IP) was measured in the biceps muscle of 16 normal male and 17 normal female subjects. The activity, upper centile amplitude (UCA), and the number of small segments (NSS) (defined in a companion paper) were measured from 500-msec epochs of the IP. The normal values of these features were defined separately for men and women by plotting the UCA and NSS values against activity for each epoch and defining an area on these plots, called a “cloud,” that contained more than 90% of the datum points from each study. The mean deviation of the individual datum points from the overall mean values was also calculated for each study. A study in one muscle is considered to be normal if more than 90% of the datum points from that muscle are within the normal clouds and the deviation values are within their normal range. In patients with neuropathy, the characteristic pattern was increased UCA with normal or decreased NSS. In patients with myopathy, NSS was increased and the UCA was normal or decreased. In all studies, the interpretations of the IP from the plots agreed with qualitative assessments of the IP made independently by an electromyographer. The use of these features to understand and quantitate the changes in the motor units produced by disease is demonstrated by serial studies performed in a patient with motor neuron disease.     

Measurement of the amplitude of the EMG envelope

When the force of muscle contraction is increased, the amplitude of the EMG envelope (ENAMP) increases. The ENAMP is usually assessed subjectively and its value in EMG analysis has been established. In this article we describe a method to make automatic measurements of the ENAMP. This method was tested on recordings of the EMG interference pattern (IP) from the biceps muscle of normal subjects. Normal values of this feature in the biceps muscle are described. There was a good concordance between the ENAMP measurements made by subjective assessment and by automatic analysis and the ENAMP values correlated strongly with a previously described feature of the IP called the upper centile amplitude. We infer that ENAMP is a robust feature of the IP that reflects the amplitude of the largest MUAP in the IP.     

EMG of reinnervated motor units: a simulation study

Using computer simulation techniques, reinnervation of motor units (MUs) was studied by increasing the number of muscle fibers in the MU without changing the MU territory. The fiber density (FD) measured by single fiber EMG electrodes, the amplitude, area and number of turns of concentric needle (CN) EMG motor unit action potentials (MUAPs) and the amplitude of macro EMG MUAPs were most affected by partial reinnervation changes. The values of these features increased during simulated advanced reinnervation, as did the number of CNEMG MUAPs that had increased numbers of phases or turns and the mean CNEMG MUAP duration. The increase in macro EMG MUAP amplitude, FD and CNEMG MUAP area were proportional to the increase in the number of muscle fibers in the MU. When loss of muscle fibers due to so-called MU fractionation was simulated, values of all EMG features fell, but were still increased compared to normal. Two patterns of change in SFEMG and macro EMG values were identified that may distinguish between recordings made from reinnervated low force threshold MUs and those from higher force threshold MUs.     

A comparison of turns analysis and motor unit analysis in electromyography.

We compared the results of turns analysis and motor unit analysis on 4056 electromyographic interference patterns (IPs) from normal subjects and patients with neuromuscular disorders. The motor unit analysis involved decomposing the IPs into their component motor unit action potentials (MUAPs) using automatic decomposition electromyography (ADEMG). We checked the accuracy of the decompositions by attempting to reconstruct some of the IPs from their identified MUAPs using computer simulations. The simulations revealed that ADEMG typically identified more than 60% (but not all) of the MUAPs in a given IP. Both turns and MUAP properties showed regular and related changes with force, age, muscle, and recording electrode type. The number of turns in each IP was highly correlated with the number of active MUAPs (r = 0.65), the mean MUAP firing rate (r = 0.72), the mean number of turns per MUAP (r = 0.34), and the product of these 3 properties (r = 0.83). The mean amplitude change per turn was highly correlated with the mean MUAP amplitude (r = 0.82), but also depended on the number of turns per MUAP. Due to the lack of a one-to-one relationship between the turns analysis properties and the MUAP properties, the turns analysis properties by themselves did not provide sufficient information to infer unambiguous physiological information about motor unit morphology or firing behavior.     

Quantitative EMG in inflammatory myopathy

Fifty-four quantitative electromyographic (EMG) studies were made in 37 patients with inflammatory myopathy (IM) at different points in their clinical course and treatment. All studies were performed in the biceps brachii which varied in clinical strength. Motor unit action potentials (MUAPs) in 45 studies and EMG interference pattern (IP) in 48 studies were recorded using a concentric needle electrode. Macroelectromyographic (Macro-EMG) MUAPs were recorded from 10 patients in 14 studies. MUAP analysis revealed a myopathic pattern (decreased duration and/or area: amplitude ratio) in 69% of studies. IP analysis was more sensitive than MUAP analysis, demonstrating a myopathic pattern in 83% of studies. Macro-EMG MUAP amplitudes were reduced in two studies, minimally increased in one study and normal in the remainder; in 6 (40%) studies, fiber density was slightly increased. Thus, reinnervation does not seem to play an important role in motor unit remodeling in IM.     

Reliability of spike triggered averaging of the surface electromyogram for motor unit action potential estimation

Introduction: The reliability of estimated motor unit parameters using spike triggered averaging (STA) of the surface electromyogram (sEMG) has not been tested thoroughly. We investigated factors that may induce amplitude bias in estimated motor unit action potentials (MUAPs) and shape variations. Methods: An sEMG record was simulated. MUAPs were then estimated from the STA of the simulated EMG. Results: Variations in MUAP duration led to under‐estimation of real MUAP amplitude, while synchronized firing led to over‐estimation of amplitude. Spurious firing resulted in over‐estimation of the amplitude of small motor units but under‐estimation of the amplitude of large ones. Variability in amplitude and high firing rates had minimal influence on amplitude estimation. High firing rates and variation in MUAP duration led to large variations in MUAP shape. Estimation errors also correlated with shape variations. Conclusions: Recommendations to enhance the accuracy of the STA estimates have been proposed. Muscle Nerve 48 : 557–570, 2013     

Simulation of concentric needle EMG motor unit action potentials

Computer simulations of motor unit action potentials (MUAPs) as measured by a concentric needle (CN) electromyography (EMG) electrode in normal motor units (MUs) indicated that the MUAP amplitude is determined mainly by the proximity of the electrode to the closest muscle fiber. The area and duration of the simulated MUAPs were affected by all muscle fibers in front of the active recording surface but mainly by those that were less than 2 and 2.5 mm, respectively, from the active recording surface. The MUAP area was also affected by the proximity of the electrode to the closest muscle fiber. The number of phases of the simulated MUAPs increased when the dispersion of the arrival times of individual muscle fiber APs at the electrode was increased. Increased temporal dispersion of APs decreased the MUAP amplitude and area slightly but did not affect the MUAP duration. It is inferred that different features of the CN MUAP are determined by the distribution of muscle fibers within different portions of the MU territory and thus provide complementary information about the MU architecture.     

Recording characteristics of monopolar EMG electrodes

Motor unit action potentials (MUAPs) and the electromyographic (EMG) interference pattern (IP) were recorded from the biceps muscle of 5 normal subjects using both a concentric needle (CN) and a disposable monopolar needle (MN) electrode. The MUAPs recorded by the MN electrode had higher amplitude and area and were more frequently complex than those recorded with the CN electrode. The MUAP duration and area: amplitude ratio were similar for both electrodes. Although the MN electrode had a larger recording surface, its dimensions (maximum diameter and length of the cone shaped tip) were similar to those of the CN electrode (minor and major axes of the elliptical recording tip). Based on these observations, we infer that the MN electrode may be more selective than the CN electrode, ie, the AP amplitude recorded by the MN electrode decreases faster than the AP amplitude recorded by a CN electrode when the distance of the muscle fiber from the recording electrode increases. Photomicrographs of the MN electrode after use demonstrated no evidence that the insulating material had peeled off. There was also no evidence that MUAP measurement values changed during the recordings as would be expected if the recording surface changed due to peeling of the insulating material.     

Can standard surface EMG processing parameters be used to estimate motor unit global firing rate?

The relations between motor unit global firing rates and established quantitative measures for processing the surface electromyogram (EMG) signals were explored using a simulation approach. Surface EMG signals were simulated using the reported properties of the first dorsal interosseous muscle in man, and the models were varied systematically, using several hypothetical relations between motor unit electrical and force output, and also using different motor unit firing rate strategies. The utility of using different EMG processing parameters to help estimate global motor unit firing rate was evaluated based on their relations to the number of motor unit action potentials (MUAPs) in the simulated surface EMG signals. Our results indicate that the relation between motor unit electrical and mechanical properties, and the motor unit firing rate scheme are all important factors determining the form of the relation between surface EMG amplitude and motor unit global firing rate. Conversely, these factors have less impact on the relations between turn or zero-crossing point counts and the number of MUAPs in surface EMG. We observed that the number of turn or zero-crossing points tends to saturate with the increase in the MUAP number in surface EMG, limiting the utility of these measures as estimates of MUAP number. The simulation results also indicate that the mean or median frequency of the surface EMG power spectrum is a poor indicator of the global motor unit firing rate.     

Magnetic transcranial stimulation in healthy humans: influence on the behavior of upper limb motor units

Aim of the study was to analyze the characteristics of motor action potentials recruitment during magnetic trans-cranial stimulation (TCS) of the brain. Coaxial needle recordings from hand and upper limb musculature, as well as surface electrodes were employed in 20 healthy controls during magnetic TCS with regular and figure-of-8 coil in different experimental protocols including: (a) simple reaction time paradigm during which TCS at subthreshold intensity for eliciting MEPs in relaxation was delivered at various intervals between the signal to move and the onset of the voluntary EMG burst; (b) suprathreshold TCS was randomly delivered while the subject was voluntarily firing at a regular rate one ‘low’ and/or ‘high threshold’ motor unit action potential (MUAP). The pre- and post-TCS MUAPs recruitment as well as their firing rates were compared; (c) recordings with two separate needles picking up individual MUAPs from the same or from two different muscles were obtained in order to test ‘synchrony’ of MUAP's discharge before and after TCS; (d) the influence of the time-interval separating the last discharged MUAP from ICS was evaluated. (e) differences between simultaneous surface and depth recordings were examined. The following results were obtained. (a) The same low-amplitude MUAP which is first voluntarily recruited at the onset of the EMG burst is the one initially fired by TCS in the pre-movement period. Latency shortenings and amplitude enlargement of surface MEPs were observed with faster reaction times. Such changes were coupled to the recruitment of high-threshold MUAPs being larger in amplitude and briefer in latency than the initial one. (b) When using suprathreshold TCS, MEPs followed by silent periods were found. The SP was followed by a rebound acceleration of the MUAPs firing rate compared with pre-TCS levels. Besides rebound acceleration, new MUAPs of larger amplitude than the original (=pre-stimulus) ones were recruited beyond the voluntary control. This phenomenon-together with longer SPs- was progressively more pronounced with stronger stimuli. (c) TCS was affecting the ‘synchrony’ of MUAPs. (d) If the latency difference between the last pre-stimulus spike and the TCS was exceeding the half-cycle of the MUAP ‘natural’ firing, the SP was longer in duration. (e) SPs not preceded by MEPs were clearly present in depth recordings. Surface recordings mainly reflected the behavior of high-threshold and large MUAPs.     

An evaluation of the utility and limitations of counting motor unit action potentials in the surface electromyogram

The number of motor unit action potentials (MUAPs) appearing in the surface electromyogram (EMG) signal is directly related to motor unit recruitment and firing rates and therefore offers potentially valuable information about the level of activation of the motoneuron pool. In this paper, based on morphological features of the surface MUAPs, we try to estimate the number of MUAPs present in the surface EMG by counting the negative peaks in the signal. Several signal processing procedures are applied to the surface EMG to facilitate this peak counting process. The MUAP number estimation performance by this approach is first illustrated using the surface EMG simulations. Then, by evaluating the peak counting results from the EMG records detected by a very selective surface electrode, at different contraction levels of the first dorsal interosseous (FDI) muscles, the utility and limitations of such direct peak counts for MUAP number estimation in surface EMG are further explored.     

A comparison of two commercial quantitative electromyographic algorithms with manual analysis.

Quantitative EMG (QEMG) is a useful technique in the evaluation of neuromuscular disease. Manual waveform measurements have been replaced by automated computer-based measurements, but there is no uniformity in computer algorithms used to make waveform measurements. We compared QEMG measurements made by algorithms in two commercially available EMG machines with manual measurements. Motor unit action potentials (MUAPs) were simultaneously fed into the two machines and analyzed using QEMG default settings and automatic waveform marking. The averaged MUAPs were also manually marked. The two algorithms and manual marking did not differ significantly for MUAP amplitude. There were significant differences between algorithms for duration and number of phases. Our study indicates that, although automated algorithms make QEMG more practical, visual inspection, and remarking of each MUAP if needed, is necessary before making clinical judgments from the data.     

Needle electromyography of the rectus abdominis in patients with amyotrophic lateral sclerosis

We examined the role of needle electromyography (EMG) of the rectus abdominis (RA) in assessing thoracic involvement in amyotrophic lateral sclerosis (ALS). Needle EMG of the RA was performed in 67 patients with sporadic ALS and 110 healthy controls. The presence of abnormal spontaneous activity, configuration of motor unit action potentials (MUAPs), and recruitment pattern of motor unit potentials were examined. In ALS patients, MUAPs in the RA were of prolonged duration, large amplitude, and showed increased prevalence of polyphasic waveforms compared to controls. Significant differences in MUAP parameters, presence of abnormal spontaneous potentials, and interference patterns were noted between ALS patients and controls. Additionally, we found that active denervation was more frequent in the RA of ALS patients with dyspnea than those without dyspnea. Thus, conventional needle EMG of the RA is a valuable electrophysiological method to assess clinical and subclinical involvement of thoracic lower motor neurons in patients with suspected ALS.     

Quantitative EMG and motor unit recruitment threshold using a concentric needle with quadrifilar electrode

According to Henneman's size principle, small motor units are recruited before large ones. We used the electromyographic (EMG) signal decomposition technique to determine the quantitative relationships between five motor unit action potential (MUAP) parameters (amplitude, duration, area, thickness, and size index) and the recruitment threshold of the motor units recruited up to 50% of the maximum voluntary contraction in the first dorsal interosseous, biceps brachii, rectus femoris, and tibialis anterior muscles of 5 healthy young men. In each muscle, the amplitude, duration, area, and size index had significant, positive high correlations with the motor unit recruitment thresholds. We conclude that the size principle applies to recordings made with concentric needle EMG electrodes under special recording conditions, and therefore that more importance should be attached to the patient's contraction force during EMG examinations in order to evaluate MUAPs for electrodiagnostic purposes.     

Multi-motor unit action potential analysis (MMA)

We have developed an algorithm, called multi–motor unit action potential analysis (MMA), to aid quantification in routine needle EMG examination. In only 5–8 min, it was possible to extract, analyze, and validate 20 motor unit action potentials (MUAPs). In the biceps muscle of normal subjects, the MUAP measurements are compared with measurements using manual, triggered averaging, automatic decomposition, and other MMA algorithms described in the literature. The mean values of MUAP features by MMA fell between the extreme limits obtained from the amplitude-triggered method and the manual method described in the literature. In patients, the results are consistent with routine EMG and similar to those results described in the literature. Fast acquisition and analysis of MUAPs as well as interference pattern (IP) analysis make this technique useful to document EMG abnormalities in routine needle examination.© 1995 John Wiley & Sons, Inc.     

Electrophysiological evidence of doubly innervated branched muscle fibers in the human brachioradialis muscle

OBJECTIVE: Motor-unit action potentials (MUAPs) with unstable satellite (late-latency) components are found in EMG signals from the brachioradialis muscles of normal subjects. We analyzed the morphology and blocking behavior of these MUAPs to determine their anatomical origin. METHODS: EMG signals were recorded from the brachioradialis muscles of 5 normal subjects during moderate-level isometric contractions. MUAP waveforms, discharge patterns, and blocking were determined using computer-aided EMG decomposition. RESULTS: Twelve MUAPs with unstable satellite potentials were detected, always two together in the same signal. Each MUAP also had a second unstable component associated with its main spike. The blocking behavior of the unstable components depended on how close together the two MUAPs were when they discharged. CONCLUSIONS: The latencies and blocking behavior indicate that the unstable components came from branched muscle fibers innervated by two different motoneurons. The satellite potentials were due to action potentials that traveled to the branching point along one branch and back along the other. The blockings were due to action-potential collisions when both motoneurons discharged close together in time. SIGNIFICANCE: Animal studies suggest that branched muscle fibers may be a normal characteristic of series-fibered muscles. This study adds to our understanding of these muscles in humans.     

Effects of high‐pass filtering on MUAP metrics

Previous observations suggest that elevated high‐pass filter settings (1,600–3,200 HZ ) can reveal greater motor unit action potential (MUAP) complexity (turns). We assessed the effect of high‐pass filter settings (500, 1,000, 2,000 HZ ) on MUAP metrics. MUAPs were recorded with a concentric needle and initially extracted by a decomposition software algorithm at 10 HZ –10 kHZ and further filtered offline at 500, 1,000, and 2,000 HZ . When reanalyzed by the decomposition software there were marked reductions in peak–peak amplitude, area, area‐to‐amplitude ratio, and duration at the 500 HZ filter with lesser subsequent reductions at higher filter settings. In contrast, turn and phase counts did not change significantly. Individual MUAPs tracked across filter settings showed rare increases in turn count at the 500 HZ setting but a subsequent decrease in counts with higher filter settings. We conclude that the routine use of elevated high‐pass filters, as in quantitative EMG analysis, does not enhance turn count. Muscle Nerve, 2009