Introduction

In the last decades, a growing body of literature focused on the use of mechanomyography (MMG) as a means to study non-invasively skeletal muscle mechanical activity. MMG signal is detectable at the skin surface during the dimensional changes of active muscle fibres that generate pressure waves due to voluntary or evoked contractions. It has been suggested that this signal may reflect the three main physiological phenomena of the mechanical aspects of muscle contraction, including (a) the gross lateral movement of the contracting muscle at the beginning of muscle contraction, (b) the subsequent vibrations at the resonance frequency of the muscle, and (c) the dimensional changes in the active muscle fibres [1, 2]. Electromyography (EMG), as the conventional modality to monitor skeletal muscle electrical activity, is limited in providing sufficient information on muscle mechanical behaviour [3], focusing only on the electrical aspects (neural control) of muscle function [4]. EMG signal is also not sufficiently suitable to quantify electrically evoked dynamic muscle actions [5]. The exploration of complementary paradigm that is sensitive to the muscle mechanical activities and devoid of inherent electrical noise such as MMG is, therefore, needed. The classical works on muscle sound, the former term to indicate the MMG signal, revealed that its properties, i.e. vibrations and dimensional changes of the active fibres, could be used to get more insights into the mechanical counterpart of muscle contraction [6–12].

The term MMG was suggested during the CIBA Foundation Symposium in 1995 [13] to overcome the terminological confusion (phonomyogram, acoustic myogram, vibromyography, soundmyogram, muscle sound, etc.) caused also by the habit of identifying the phenomenon with other explanations than its mechanical origin [2, 6, 14].

The pressure waves transmitted from the active muscle fibres to the skin surface can be detected by specific transducers to record muscle surface oscillations related to motor units’ (MUs) mechanical activity [2, 15]. Several types of transducers can be used, among which piezoelectric contact sensors (PIZ), microphones (MIC), laser distance sensors (LDS), electret condenser microphones (ECM), and accelerometers (ACC), which is largely the most used sensor for MMG detection.

Usually, the MMG signal is analysed in time domain by calculating the root mean square (RMS), the averaged rectified value (ARV) of the signal and/or the maximum displacement during the transient phase of contraction (p–p), and in frequency domain by determining the mean frequency (MPF or MF), and/or the median frequency (MNF) of the power spectrum density distribution. Altogether, these parameters provide information about the neuromuscular strategies adopted by the contracting muscle to activate and modulate the force output during contraction (see Fig. 1).

Fig. 1
figure 1

Time- and frequency-domain analysis of MMG signal during voluntary (left-sided column) and electrically evoked contraction (right-sided column). RMS root mean square; ARV averaged rectified value, p–p peak-to-peak, MF mean frequency, MPF mean power frequency, MNF median frequency

Full size image

The application fields of MMG are numerous and increasing: they span from the assessment of muscle function during isometric or dynamic contractions under different physiological conditions, such as fatigue [16–18], muscle temperature manipulation [19, 20], stretching [21–23], ageing [24], training [25, 26], to the analysis of the effects of rehabilitation programs [27, 28], the development of prosthesis [29], and/or the use of the MMG as a triggering signal [30].

From the pioneering proposal of Hufschmidt [31], a novel and interesting application field that is taking place in the last years is the use of an EMG, MMG, and force signals combined approach as a tool to partition the electrical and the mechanical events underpinning the electromechanical delay during muscle contraction (EMD) and relaxation (R-EMD) phase [19, 32–37].

Despite the utilization and the applicability of MMG are becoming even more frequent and widespread, only four review articles related to MMG [38–41] are actually present in the literature. They all focus on particular aspects of MMG application, such as the examination of MMG amplitude and frequency responses [41], the assessment of MU recruitment strategies [38, 41], and the evaluation of muscle function [39, 40].

The present review aimed to provide a wider overview on the new and different application fields (from basic muscle physiology to sports and rehabilitation), in which the MMG signal is employed. A particular emphasis on the reliability and sensitivity of the MMG signal analysis will be also given.

Methods

Article searching

A comprehensive literature search for MMG and skeletal muscle was performed on the electronic databases PubMed, Scopus, Web of Science, Embase and Google Scholar for relevant articles published from January 2000 to October 2014. The key words used were mechanomyography, mechanomyogram, mechanomyographic, MMG, skeletal muscle, muscle function, muscle assessment and rehabilitation. The combination among key words with #AND and #OR was performed. Only papers written in English were considered during databases searching. Journal articles, conference proceedings and clinical reports were included for potentially eligible studies. Moreover, the reference list of all the articles was carefully checked.

Study selection

Article titles and abstracts identified by the search were screened for potential relevance. The full text of all potentially relevant studies was reviewed to determine if it fulfilled the eligible criteria. Articles describing a theoretical or practical use of MMG in the field of muscle contraction physiology and rehabilitation were included. Two authors (EC and SR) screened independently the results of the electronic searches to select potentially relevant citations according to the criteria defined above. The published works meeting the most relevant criteria were included. Studies without proper data presentation, with unclear or vague protocol description, and without an in-depth discussion were excluded from this review.

Data extraction

Two authors (EC and SR) analysed the articles individually using an MS-Excel structured data extraction form purposely created for this review. Extracted data were compared and discussed by these two authors before being included in the final pile of reviewed papers. Information extracted from each article included: (a) type of sensor, (b) participants characteristics, (c) experimental protocols, (d) investigated muscle/s, (e) measured and calculated parameters, and (f) main results.

Validity assessment

Three authors (EC, SR and FE) analysed the data extracted from the potentially relevant articles. After the analysis, only the information extracted from the most relevant studies (those organized with proper data presentation, clearly verified selection of protocols, and through demonstration of research methodologies) was discussed, to reduce risk of bias.

Quantitative data synthesis

All the articles were divided in three main areas: (a) reliability, (b) muscle physiology, and (c) rehabilitation. Articles in the muscle physiology area were further divided into: (a) muscle contraction under isometric (in vivo and in isolated muscle) and dynamic conditions, (b) MMG utilization in different physiological models (fatigue, stretching and different training modalities), and (c) the use of MMG in EMD partitioning. A description of these studies is presented in tables in the Results section (Tables 1, 2, 3, 4, 5, 6, 7, 8, 9).

Table 1 MMG measurements reliability
Full size table
Table 2 Evaluation of isometric contraction by MMG analysis
Full size table
Table 3 Evaluation of isometric contraction in isolated muscle by MMG analysis
Full size table
Table 4 Evaluation of dynamic contractions by MMG analysis
Full size table
Table 5 Effect of training evaluated by MMG signal analysis
Full size table
Table 6 Fatigue assessment by MMG signal analysis
Full size table
Table 7 Effect of stretching evaluated by MMG signal analysis
Full size table
Table 8 Electro-mechanical delay assessment
Full size table
Table 9 Evaluation of rehabilitation protocol by MMG signal analysis
Full size table

Results

Flow of included studies and their characteristics

The comprehensive literature search returned 111 articles (Fig. 2). Of these, 7 met the inclusion criteria for “reliability”, 88 for “muscle physiology”, and 16 for “rehabilitation”. Within “muscle physiology”, 29 articles were included in “isometric contraction” (23 in vivo and 6 concerning isolated muscle), 20 were included in “dynamic contraction”, 20 in “fatigue”, 8 in “stretching”, four in “different training” and 7 in “electromechanical delay”.

Fig. 2
figure 2

Flow chart of the process for selecting potential studies to be included in the present review

Full size image

Among the 111 relevant studies, 102 studies were peer-reviewed journal articles and 9 studies were conference papers.

Reliability of MMG signal

In accordance with Munro [42], reliability values were consider as very high if Intraclass Correlation Coefficient (ICC) was >0.90, high if between 0.70 and 0.89 and moderate if between 0.50 and 0.69. Table 1 includes the studies on the reliability of the MMG signal under isometric [33, 43–45] and dynamic contractions [46–48]. All the investigations used the ACC as sensor. With the exception of Stock et al. [48], who investigated the reliability of the MMG signal of the pectoralis major and of the triceps brachii muscles, all the other studies focused on lower limbs muscles, in particular the quadriceps and gastrocnemius muscles.

Cè et al. [33] reported a high intersession intraclass correlation coefficient (ICC) and a low standard error of measurement (SEM%) of the maximum MMG displacement during the muscle relaxation phase (R-MMG p–p) after a tetanic stimulation. Ryan et al. [43] found a lack of consistency and some degree of inter-individual variability between MMG amplitude and isometric torque. MMG mean power frequency (MPF) and isometric torque relationship were best fit with linear models for both the low-strength and high-strength groups. Al-Zahrani et al. [44] reported moderate to strong ICC levels with high small detectable differences (SDDs) in MMG RMS and MF versus torque relationship, both before and after a fatiguing protocol. The authors, given the not very high between-day reliability, suggested caution in using MMG RMS and MF and their time behaviour in assessing muscle fatigue. Similar conclusions were also supported by Herda et al. [45], who found comparable levels of reliability of the amplitude and the MPF of the MMG signal.

Under dynamic condition [46, 48], reported a lack of reliability of the relationship between MMG amplitude and force in quadriceps, pectoralis major and triceps brachii muscles. The authors concluded that when using MMG in the clinical setting, dynamic muscle actions do not appear to be appropriate for assessing training-induced changes in muscle function. Lastly, Armstrong et al. [47] showed moderate ICC values of the MMG signal detected by high-resolution triaxial ACC assessed on quadriceps muscle during repetitive single-leg balance test. The authors concluded that MMG can provide precise information on postural balance, and may have application in evaluating postural control and stability.

MMG in muscle physiology

All the studies concerning muscle physiology are divided into sub-groups and reported in Tables 2, 3, 4, 5, 6, 7, 8. Among the 88 studies, 49 used an ACC as sensor to detect the MMG signal, 28 used a PIZ, 6 an MIC, 3 used an LDS and 2 used an ECM. As far as the investigated muscle is concerned, 33 studies over 88 were conducted on the elbow flexors muscles (biceps brachii and brachioradialis muscles), the quadriceps muscle was analysed in 25 studies, the plantar flexors muscles (the whole triceps surae or a single component) were investigated in 22 studies, 4 on the triceps brachii muscle, 2 on the upper trapezius, tibialis anterior and first digital interosseum muscles, and only 1 on the finger flexors, pectoralis major, erector spinae and biceps femoris muscles.

MMG during isometric contraction “in vivo”

Five studies observed a strong relationship between force output and MMG amplitude and frequency content up to a certain % MVC [24, 49–53]. On one hand, the MMG RMS, which provides information about the number of active MUs [2], has been reported to have a curvilinear shape with the increase in force output, reaching a plateau or even a clear decrease that differ from muscle to muscle: the more the investigated muscle is composed by slow twitch MUs, the earlier the plateau appears. Moreover, a difference in RMS and an earlier plateau in the RMS/% MVC relationship were reported in the elderly [24], by the gender of the investigated subjects [53], and at different joint angles at which the muscle is contracted [54]. On the other hand, the relationship between MMG signal frequency content and force output is mostly linear [53]. With these characteristics, MMG permitted to study the adaptations of neuromuscular activation strategies induced by several conditions, such as climbing [49] or resistance and aerobic training [55]. Also different physiological models, such as potentiation [56], muscle pain [57], hypoxia [58], hypothermia [59], muscle relaxation [60, 61], while overimposing an electrical twitch stimulation during maximum voluntary contraction [62], or during spinal anaesthesia [63] could be also investigated.

MMG during isometric contraction in isolated muscle

The articles concerning the MMG signal during isometric contraction in isolated muscle are presented in Table 3. All the six studies catalogued within this area, used muscles from a mammalian model (four studies used rat’s muscle and two cat’s muscles). Five of them analysed the gastrocnemius muscle [8, 64–67], while the other investigated the tibialis anterior muscle [68].

In their investigation, Orizio et al. [64] demonstrated that the force and MMG frequency response can be modelled by a critically damped second-order system with two real double poles and a pure time delay. This allowed the authors to conclude that MMG could be a reliable tool to investigate the muscle frequency response during stimulated isometric contraction, and that different components of the muscle mechanical model may specifically affect the force or MMG signals.

In a further study, Orizio et al. [67] observed that the rate of MMG increase during the transient phase of muscle contraction is always higher than that of force, while during the relaxation phase, the force decrease is much faster than the MMG. A counterclockwise hysteresis in the graph with the relative force exerted vs the relative MMG amplitude relationship was reported. Authors concluded that during relaxation more force is developed for the same muscle surface displacement.

Hsieh et al. [68] used MMG signal to investigate the paired-pulse transcranial magnetic stimulation inhibitory phenomenon in awaked rats, and then applied differential pharmacology to test the hypothesis that long-interval cortical inhibition is mediated by the GABAA receptor. Their results demonstrated predictable latency discrepancy between the motor-evoked potential and the evoked MMG. With pharmacological testing, time course observations showed that paired-pulse transcranial magnetic stimulation-MMG inhibition was acutely reduced following pentylenetetrazole (GABAA antagonist), acutely enhanced after pentobarbital (GABAA agonist) and then recovered to pre-treatment baseline after 1 h. Their findings support the application of the paired-pulse transcranial magnetic stimulation-MMG technique for measuring the cortical excitability in awaked rats and provide the evidence that GABAA receptor contributes to long-interval paired-pulse cortical inhibition. They concluded that paired-pulse transcranial magnetic stimulation-MMG appears to be a well-tolerated biomarker for measuring GABAA-mediated cortical inhibition in rats.

In their investigations, Bichler and coworkers [8, 65, 66] described the shape and the amplitude of the MMG signal during muscle contraction and relaxation at different stimulation frequencies. They observed that MMG onset was coincident with the beginning of the increase in tension during single twitch and tetanically evoked stimulations, followed by a weaker MMG signal accompanying the fused tetanus phase [8]. During contraction and relaxation, MMG was characterized by the reverse direction of the first extreme phase, positive and negative, respectively. MMG amplitude was correlated with both the tension increase and the velocity of tension increase during both stimulation modalities [65]. The strongest type II MUs (fast fatigable) generated the MMG signal of the highest amplitude. Lastly, MMG p–p was directly correlated with the increase in twitch tension during the potentiation phase and with the decrease of twitch tension after a fatigue test, respectively [66].

MMG during dynamic contraction

Table 4 presents the analysed articles concerning the MMG signal during dynamic contractions. Ten over twenty studies used an ACC to detect MMG signal [46, 69–77], nine used a PIZ [78–85], and one used an MIC [86]. Ten articles focused on the quadriceps muscle, seven investigated the biceps brachii muscle, two were centred on the triceps brachii muscle, while pectoralis major, first digital interosseum and forearm flexors muscles were analysed in just one study.

The investigations mainly focused on the relationship between MMG signal and force output during concentric isotonic [48, 69, 74, 76] and isokinetic contraction [78, 81–83], as well as eccentric isokinetic contraction [76, 80, 84, 85]. Altogether, authors observed that MMG amplitude increases linearly in both isotonic and isokinetic contraction until a certain muscle-dependent threshold that could be identified to be about the 80 % 1-RM and 240°/s for isotonic and isokinetic contraction, respectively. Then, MMG amplitude reaches a plateau or even decreases. MMG amplitude was reported to be higher during concentric than eccentric contractions [71].

Two investigations evaluated the relationship between MMG amplitude and excess post-exercise oxygen consumption after incremental cycle ergometer exercise [70], and between the metabolic request and MMG responses during continuous exercise at critical power determined from the 3-min all-out test [75]. Malek et al. [70] evidenced significant time-constant values for excess post-exercise oxygen consumption and MMG amplitude for quadriceps muscles during the 60-min post-exercise recovery period. These results suggested that excess post-exercise oxygen consumption after exercise could not be exclusively attributed to elevated activity of the working muscles. Bergstrom et al. [75] found no changes in MMG amplitude or MPF over time against an increase in oxygen uptake and heart rate.

Three studies demonstrated no changes in MMG signal properties during dynamic contractions as a consequence of muscle compression [72], change in muscle oxygen tension [77], and muscle dehydration [79].

MMG analysis and the effects of training

Four studies evaluated the effect of different training modalities on muscle performance utilizing MMG signal analysis (Table 5). All works used a PIZ to detect MMG signal. Two studies analysed the quadriceps muscle [26, 87], while the other two analysed the biceps brachii [88] and the plantar flexors muscles [25], respectively.

Evetovich et al. [87] examined the effects of concentric isokinetic leg extension training on MMG MPF in a group of 21 young men. Training consisted in six sets of leg extensions, 3 days per week, for 12 weeks at an angular velocity of 90°/s. The authors found an increase in peak torque but no changes in the MMG MPF from the vastus lateralis muscle over the 12-week training period. They explained these results with training-induced changes in muscle stiffness that could affect the MMG signal, and/or adaptations in other quadriceps muscles than the investigated vastus lateralis.

Esposito et al. [26] verified the hypothesis that isokinetic training could induce even in the elderly changes in EMG and MMG parameters, compatible with functional changes in the MU pool. In a group of ten sedentary males, aged from 62 to 78 years., the surface EMG and MMG were recorded from the vastus lateralis muscle during isometric contractions at 20, 40, 60, 80 and 100 % MVC, before and after 12 weeks of isokinetic actions. Training sessions consisted of six series of ten repetitions, each at an angular velocity of 2.09 and 4.19 rad/s, twice a week). The authors found increases in MVC and cross-sectional area determined by magnetic resonance imaging (CSA), MVC/CSA ratio, EMG RMS and MMG MF at the highest workloads. Altogether, these findings indicated that (a) isokinetic training can improve muscle size and performance even in the elderly, and (b) EMG and MMG changes may be compatible with a retrieval of the fast twitch fibre MUs.

Ebersole et al. [88] evaluated the effects of unilateral, isometric training of the forearm flexors on strength, MMG and EMG responses of the biceps brachii in the trained and untrained limb at three joint angles. A group of seventeen young women was involved in a unilateral training program consisting in 3–5 sets of 8/6-s repetitions at 80 % MVC, 3 times per week, for 8 weeks. The authors found an increase in arm circumference as well as maximum isometric strength in the trained limb at all three joint angles. There were, however, no changes in MMG or EMG amplitude in the trained or untrained limb. These findings indicated that the increased strength might have been due to factors associated with hypertrophy, independent of neural adaptations in the biceps brachii. Furthermore, hypertrophy may have had counteractive effects on the MMG signal that could be responsible for the lack of training-induced changes in the MMG amplitude.

Cramer et al. [25] assessed the effects of 3 days of isokinetic resistance training combined with 8 days of creatine monohydrate supplementation on peak torque, mean power output, acceleration time, and EMG and MMG signals of the vastus lateralis muscle during maximal concentric isokinetic leg extension actions. A group of 25 young men were randomly assigned to either the creatine or the placebo group. Training consisted of three sets of 10 repetitions at 150°/s, performed on day 3, 5 and 7. The authors reported an increase in EMG and MMG MDF and in peak torque, with a decrease in acceleration time in both groups.

MMG and fatigue

The investigations on the assessment of the fatigue phenomenon through MMG signal analysis are reported in Table 6.

Eleven studies over twenty used an ACC to detect MMG signal [2, 16–18, 89–94], seven used a PIZ [82, 95–100], and two utilized an MIC [101, 102].

These articles focused mainly on the modifications of MMG signal (in particular by calculating MMG RMS and MF or MPF) after isometric voluntary [2, 90, 92–95, 97, 100, 102] and electrically evoked contractions [16, 17], or dynamic contractions [82, 89, 91, 96, 98]. With the only exception of Camic et al. [89] and Orizio et al. [2], increases in MMG amplitude accompanied by decreases in MMG signal frequency content were reported in all the other investigations under both isometric and dynamic conditions. Interestingly, Gobbo et al. [16] and Cè et al. [17] highlighted strong correlations between the peak torque and the p–p of the MMG signal during the transient phase of muscle contraction and relaxation. These correlations suggested that MMG amplitude of these two phases might be strictly correlated with the transient of Ca2+ release and re-uptake occurring during muscle contraction and relaxation, respectively [16, 17].

Two studies [101, 103] aimed to evaluate the use of MMG as a biomarker to evidence the occurrence of the anaerobic threshold during incremental cycling exercise until maximum effort, with different results: Kimura et al. [101] found an abrupt change in MMG signal coinciding with the ventilatory threshold, suggesting that the MMG could be used practically to retrieve the fatigue-related changes in muscle mechanical properties during cycle exercise. On the other hand, Zuniga et al. [103] retrieved no correlations between MMG signal and ventilatory threshold, indicating that gas exchange and neuromuscular fatigue thresholds may demarcate different exercise intensity domains.

Lastly, Hendrix et al. [18], aiming to compare the mean torque levels derived from the critical torque and MMG MPF fatigue threshold (MPFFT) during isometric forearm flexion muscle actions, found no correlations between the two parameters.

MMG and stretching effects

The articles aimed to evaluate the acute effects of stretching on muscle force output are summarized in Table 7. Six investigations over eight used an ACC to detect MMG signal, whereas the other two studies used a PIZ. Plantar flexors muscles were analysed in five investigations [21, 23, 104–106], while the others focused on quadriceps muscle [107], biceps femoris [108], and biceps brachii [22].

In their investigations, Esposito et al. [21, 105], Ce et al. [104] and Longo et al. [23] found a decrease in MMG p–p after passive stretching coupled to a decrement of evoked twitch and tetanic force. The authors stated that such decrements were ascribable to mechanical modifications of the muscle–tendon unit (stretching-induced decrease in stiffness), which altered the transmission of force from muscle to the tendon insertion point. These modifications have been shown to persist for at least 2 h [21]. Interestingly, MMG RMS during the force plateau phase of tetanic stimulation, increased after stretching, recovering its pre-stretch value within 15 min. The authors explained this finding suggesting that during the plateau phase both series elastic components and contractile elements reach their isometric condition, during which MMG is mainly influenced by contractile and parallel viscoelastic elements. Thus, the stretching-induced decrease in MTU stiffness should lead to an increase in MMG amplitude. Evidence of this mechanism was provided by Longo et al. [23], who reported an inverse correlation between MMG RMS (calculated during torque plateau phase) and stiffness (calculated at muscle–tendon unit and tendon level).

Similar decreases in muscle force output evaluated under concentric isokinetic condition with a concomitant increase in MMG amplitude were also reported by other investigations [22, 106–108]. They all found a stretching-induced reduction in MVC and single twitch after passive stretching of plantar flexor and quadriceps muscles, without significant increases in MMG amplitude. It was suggested that other mechanisms than muscle–tendon unit stiffness, such as force reduction, might have played a major role in determining MMG amplitude.

MMG and EMD components

The investigations that used MMG signal to assess the EMD during the contraction and/or the relaxation phase of muscle activation are summarized in Table 8. To this purpose, MMG signal was detected by ACC from the gastrocnemius medialis muscle [32, 35, 36], and biceps brachii muscle [19, 34, 37]. Four over seven articles analysed the contraction phase and three focused on the muscle relaxation phase [32–34].

Esposito et al. [35] assessed the effects of a bout of passive stretching on both the electrochemical and mechanical contributors to EMD, and their relative recovery time course. They found a stretching-induced increase in both EMD components, with different recovery times. Indeed, while the “electrochemical” component recovered within 15 min, the mechanical contribution (as well as the total EMD duration) remained lengthened for the entire recovery period, suggesting that stretching had effects of short duration at the electrochemical level, but more persisting effects on muscle–tendon unit viscoelastic characteristics.

Sasaki et al. [37] quantified the contribution of the electrochemical and mechanical processes of EMD in the human biceps brachii muscle over a wide range of elbow joint angles, finding that the time for electrochemical process was independent of joint angle, while the time for mechanical process and the total EMD duration were significantly greater at the most extended joint positions.

Cè et al. [19] evaluated the effects of muscle temperature and fatigue on the electrochemical and mechanical EMD components. Fatigue lengthened both of them while muscle cooling affected only the electrochemical processes of EMD.

Similarly, Rampichini et al. [36] found a fatigue-induced increase in all EMD contributors under electrically evoked, tetanic conditions. Reliability of the measurements and calculations were found to be from high to very high.

Lastly, Cè et al. [32–34] assessed for the first time during muscle relaxation the duration of the electrochemical and the mechanical contributors to R-EMD. The effects of fatigue on single R-EMD components were also determined. They found a lengthening of all R-EMD contributors after fatigue, with a main effect mainly on the mechanical component.

MMG in the rehabilitation field

The investigations aimed to assess the use of MMG for the evaluation of muscle activation in patients with different neuromuscular disorders, or the effects of particular rehabilitation programs, are reported in Table 9. Fourteen over sixteen studies detected the MMG signal by ACC and the other two investigations used an MIC. Six investigations analysed the quadriceps muscle, four the biceps brachii muscle, two the diaphragm muscle, and one on the masticatory muscles, soleus, pronator teres, wrist extensors and flexors, and rectus abdominis muscles, respectively. Five studies focused on patients with spinal cord injuries [109–113], two on patients with Parkinson’s disease [27, 114] and with chronic obstructive pulmonary disease [115, 116], one study on patients with myopathy [117]. The other six studies were conducted on healthy, able-body participants [20, 28, 30, 118–120].

Huang et al. [111] reported that patients presenting spasticity (patients with either spinal cord injuries or stroke) exhibited a larger H reflex/M-wave ratio and MMG amplitude than able-bodies controls. Moreover, MMG amplitude correlated with functional impairments. The authors concluded that spastic hypertonia involved an atypical increase in motoneuronal excitability and muscle mechanical properties, while impairment of functional performance and daily activity was attributable primarily to altered mechanical properties of a spastic muscle.

Some authors [109, 110, 112, 113] tested the best profile for functional electrical stimulation of the quadriceps muscle allowing raising the knee from 90° to 40° of flexion in a group of patients with spinal cord injuries. They found that a functional electrical stimulation protocol set to 1 kHz pulse frequency, with a 200 μs active pulse duration and a burst frequency of 50 Hz was the most effective. Higher MMG signal amplitude in patients than controls with no differences between rectus femoris and vastus lateralis muscles was also reported.

Marusiak et al. [27, 114] found no differences in MMG activity of agonist and antagonist muscles and in peak torque in patients with Parkinson’s disease during the “on” phase of their medication. On the other side, without pharmacological support, patients with Parkinson’s disease exhibited higher MMG amplitude and a lower median frequency of the MMG signal in the biceps brachii muscle. The authors concluded that, since MMG was not affected by physiological postural tremor and can depict differences between parkinsonians and controls, it might be a valuable tool for neuromuscular assessment in this scenario.

Sarlabous et al. [115] found a high positive correlation between the maximum inspiratory pressure developed in a respiratory cycle and MMG amplitude of the left and right hemi-diaphragm, and a negative correlation between the maximum inspiratory pressure and the maximum frequency of the MMG signal spectrum. These correlations were more evident in patients with severe chronic obstructive pulmonary disease. Torres et al. [116] found that the slope of the evolution of the MMG amplitude parameters, as the load increases during the respiratory test, has positive correlations with the Tiffenau index (% forced expiratory volume/forced vital capacity). Altogether, these authors suggested that the information provided by MMG signal could be used to evaluate the respiratory effort, muscular function and efficiency in patients with chronic obstructive pulmonary disease.

Lamraoui et al. [118] reported that, although detection of cough effort by MMG presents lower performances than EMG detection, mostly in terms of cough anticipation, MMG might be a useful tool in detecting cough event because of its better portability.

Ng et al. [117] reported a significant positive correlation between the post-activation potentiation evaluated with force signal and that assessed by MMG, with lower values in patients with non-dystrophic myopathies. They concluded that post-activation potentiation assessed by MMG could be used as an index of muscle contractility.

Kawakami et al. [119] reported correlations between the needle EMG signal of the lateral pterygoid muscle and the MMG signal during different mandibular positions, suggesting that the activity of this muscle could be evaluated by MMG signal recorded in the external ear canal.

Alves, Chau [120] investigated the discriminability of multiple hand motions using multichannel forearm MMG. In a group of nine able-bodied participants, MMG signals from six sites were differentiated among eight classes of forearm muscle activity with high level of accuracy. These results suggest that, with additional research, MMG may become a usable control signal for multifunction access devices.

Nolan, dePaor [30] described a system to provide a communication and control tool for disabled people, triggered by MMG signal, which can be detected and processed by the computer. A decision is made as to whether the muscle is contracted or relaxed based on the amplitude of the processed signal. If the computer decides that the muscle is contracted, a software switching action is performed. This switching action is used to control a software alphabet board that the disabled person can use to spell out messages.

Tian et al. [28] investigated the effects of age-related sarcopenia on the time- and frequency-domain properties of lower extremity muscles’ MMG activities. Compared to the young subjects, the elderly had significantly less lean thigh volume, absolute and relative maximal force and power. While the MMG RMS of the young subjects increased with testing intensities, in the elderly it increased only from 45 to 60 %, but levelled off from 60 to 75 % of 1-RM. These observations could be explained by the fact that neuromuscular performance and changes in MU activation pattern may result from age-related sarcopenia. The authors suggested that MMG could be used as an important means to study muscle contraction in age-related sarcopenia.

Mitchell et al. [20] examined the effects of pulsed shortwave diathermy on intramuscular temperature and MMG of the vastus lateralis muscle. They found that intramuscular temperature, as well MMG amplitude and MF increased during the MVC with the greatest increases observed for the diathermy group. During ramp contractions, MMG amplitude and MF increased at all percentages of MVC for the diathermy group only. The authors concluded that diathermy treatments may decrease musculotendinous stiffness, but not absolute strength or motor control strategies that influence force production.

Discussion

The present review aimed to provide a wider overview on the new and different application fields (from basic muscle physiology to sports and rehabilitation), in which the MMG signal can be employed, giving also particular emphasis on the reliability of the measurements.

The analysis of the literature showed that MMG, besides its well-established use to investigate muscle activation and MU recruitment strategies under different types of exercise paradigms, (a) has a high level of reliability, in particular during isometric contraction; (b) is influenced by the mechanical characteristics of cross-bridges and SEC behaviour during contraction; and (c) in the rehabilitation field, could be a useful biomarker for triggering orthosis or multifunction access devices, and for the assessment of patients with alterations in muscle function.

Reliability of MMG signal

Contradictory findings on MMG signal reliability emerged from the analysis of the literature: while moderate to high reliability values were reported for both time- and frequency-domain MMG parameters under voluntary isometric contraction [33, 43–45], during dynamic contraction the measurements reliability appears less evident [46, 48]. This aspect should be taken into consideration when planning experimental protocols in which repeated MMG measurements on different days or sessions are required.

The use of MMG to assess muscle activation and training effects

MMG signal has been widely used to assess MU recruitment strategies under both static and dynamic conditions. Recently, MMG signal was also acquired during incremental aerobic exercise, with the aim to correlate MMG variables with some cardiopulmonary and metabolic parameters related to the anaerobic threshold [70, 75]. Whereas correlations between MMG signal amplitude and frequency content and the level of force output were reported under both isometric [24, 49–53, 64] and dynamic contractions [48, 69, 74, 76], the use of MMG signal to evaluate muscle activation during aerobic tasks has been less investigated [70, 75].

Irrespective of muscle contraction regime (static or dynamic), MMG amplitude increases with force until a certain level, which for isometric contraction depends strictly from muscle dimension and fibres composition, spanning from 60 to 100 % MVC. In dynamic contractions, it occurs near the 80 % 1-RM for concentric and 240°/s for isokinetic contractions. After this “threshold”, MMG amplitude reaches a plateau or even decreases [48, 78, 121]. On the other side, MMG frequency content increases linearly with force output [41, 53]. The typical behaviour of MMG amplitude and frequency content with respect to force output, allowed investigators to apply this approach to evaluate the effects of different interventions and/or training protocols on muscle activation and MU recruitment strategies [24, 26, 48–51, 53, 56, 64, 69, 74, 76, 89, 122].

As far as aerobic exercise is concerned, only the relationship between MMG and excess post-exercise oxygen consumption (EPOC) after incremental cycle ergometer exercise [70], and between the metabolic request and MMG responses during continuous exercise at critical power determined from the 3-min all-out test [75] was evaluated. The first study reported a significant correlation between EPOC and MMG amplitude. The second investigation found no changes in MMG amplitude and MPF during a 3-min all-out cycling exercise performed at critical power. The use of MMG signal as a biomarker to evaluate muscle activation during aerobic tasks is still to be fully examined in depth.

MMG signal for fatigue assessment

Studies concerning the use of MMG as a biomarker for fatigue assessment, in particular at peripheral level, were mainly performed during sustained contractions under both static and dynamic conditions. Generally, a decrease in MMG amplitude, coupled with a downshift of its frequency content, was reported. Under a physiological point of view, these findings indicate that the fatigue-induced modifications in MU recruitment and firing rate (two events typically involved in fatigue) can be investigated by MMG. In particular, some authors found the MMG signal to be more sensitive to fatigue than EMG [82, 96]. Indeed, MMG amplitude and MPF were reported to change because of fatigue during isokinetic concentric contractions more noticeably than EMG parameters [82, 96]. Interestingly, MMG p–p, which represents the maximum displacement of the muscle belly during the on- (muscle contraction) and off-phase (muscle relaxation), has been shown to be strongly correlated with the force signal and its rate of development, both before and after fatigue [16, 17]. These correlations, in particular between MMG p–p and the second derivative of force development, and between R-MMG p–p and the second derivative of force relaxation, led the authors to suggest that these two MMG parameters may reflect the transient of Ca2+ uptake and re-uptake.

As a novelty, two investigations [101, 103] evaluated the use of MMG to mark the occurrence of the anaerobic threshold during incremental cycle exercise. In the light of the contradictory results, this peculiar MMG approach in detecting anaerobic threshold still needs to be fully explored.

MMG signal for stretching effects assessment

A correlation between MMG amplitude and MTU stiffness has been often hypothesized [1, 12, 21, 22, 123–126]. The “axiom” underpinning this hypothesis was that a reduction in MTU stiffness should yield to a greater slack when the series elastic components lengthen due to the shortening of contractile elements. Therefore, a less stiff MTU would lead to wider oscillations of muscle surface and, in turn, to an increase in MMG amplitude during the on-phase of contraction and during the subsequent force plateau.

Passive stretching is largely used in sport and rehabilitation practice to improve joint range of motion (ROM) [127]. Acute passive stretching has been shown to reduce MTU stiffness [128, 129]. Recently, Longo et al. [23], through an ultrasound and MMG approach, provided experimental evidence that, after the administration of a bout of passive stretching, the reduction in MTU stiffness was significantly correlated with the increase in MMG amplitude. Explanation was given that in this scenario contractile and elastic elements are allowed to oscillate to a higher extent, thus increasing MMG amplitude. Under a practical point of view, the correlation between MTU stiffness and MMG amplitude indicates that MMG may represent a reliable and effective means to evaluate MTU stiffness in all those muscles and under those circumstances where a direct measure of this parameter is not possible or difficult to obtain.

MMG signal for EMD and R-EMD assessment

Traditionally, EMD and R-EMD have been calculated only through the EMG and force signals. Recently, a combined EMG–MMG and force signals approach permitted to partition EMD and R-EMD in different sub-components that may reflect the duration of the mainly electrochemical and mechanical events underpinning the two latencies [17, 19, 32, 35–37]. However, it should be taken into account that, due to temporal/spatial limitations inherent in recording electrical phenomena and lateral accelerations from the surface of the skin, and to elasticity and viscosity of non-contractile tissues, a precise endpoint for the electrochemical mechanisms and a starting point for the mechanical events cannot be precisely determined [34].

With this approach, the effects of different joint angles [37], muscle temperature manipulation [19], peripheral fatigue [19, 32, 36], and passive stretching [35] were assessed on the different components included in EMD and R-EMD. Moreover, the analysis of MMG signal, in particular during the relaxation phase provides more insights into the mechanical behaviour of cross-bridges and the series elastic components toward their return to a pre-contraction condition, giving rise to a new overall delay during relaxation, namely R-DelayTOT [32].

MMG signal in the rehabilitation field

Given the information provided by MMG about neuromuscular activation and MTU behaviour, its application in the rehabilitation area is growing. From this review, two main application fields emerged: (a) the use of MMG signal as an alternative to EMG in evaluating muscle functionality in different physio-pathological models (chronic obstructive pulmonary disease [115, 116]; spinal cord injury [109–113]; Parkinson’s disease [27, 114], and myopathy [117]); and (b) its employment as a biomarker to trigger orthosis or devices devoted to ameliorate patients self-sufficiency [30, 120]. As far as the first application is concerned, MMG signal permits to discriminate differences in muscle activation between patients and controls. Although the sensitivity of MMG is not significantly higher than EMG, in the light of the lack of influence of the change in the skin impedance due to sweating and its better portability [29, 118, 130], some authors consider the MMG as a reliable alternative method in assessing neuromuscular activation and MTU behaviour in those conditions where the EMG cannot be easily detected. In consideration of the second point, the use of MMG as a trigger to activate or deactivate some devices represents to date a promising area of investigation. Actually, only few studies used MMG to this purpose, but with interesting results [30, 120].

Nolan, dePaor [30] investigated the possibility to use MMG as a trigger signal to activate a software alphabet board that the disabled person can use to spell out messages. Alves, Chau [120] investigated the discriminability of multiple hand motions using multichannel forearm MMG. Further investigations aiming at evaluating the use of MMG as a control signal for multifunction access devices, may be strongly recommended.

Conclusions

MMG parameters have a high level of reliability when they are calculated during isometric contractions. They can be used to examine the level of muscle activation and MU recruitment strategies, and provide information on the mechanical behaviour of cross-bridges and SEC during contraction. An MMG approach can be useful in different types of exercise paradigms and it may provide deeper insights into muscle mechanical behaviour under several physiological models. Moreover, MMG could be a useful biomarker in the rehabilitation field for triggering orthosis or multifunction access devices, and for the evaluation of patients with alterations in muscle function.