This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 134.117.10.200

This content was downloaded on 19/06/2014 at 09:40

Please note that terms and conditions apply.

Effects of strength training on mechanomyographic amplitude

View the table of contents for this issue, or go to the journal homepage for more

2012 Physiol. Meas. 33 1353

(http://iopscience.iop.org/0967-3334/33/8/1353)

Home Search Collections Journals About Contact us My IOPscience

IOP PUBLISHING PHYSIOLOGICAL MEASUREMENT

Physiol. Meas. 33 (2012) 1353–1361 doi:10.1088/0967-3334/33/8/1353

Effects of strength training on mechanomyographicamplitude

Jason M DeFreitas1, Travis W Beck and Matt S Stock

Department of Health and Exercise Science, University of Oklahoma, Norman, OK 73019, USA

E-mail: defreitas@ou.edu

Received 11 April 2012, accepted for publication 4 July 2012Published 20 July 2012Online at stacks.iop.org/PM/33/1353

AbstractThe aim of the present study was to determine if the patterns ofmechanomyographic (MMG) amplitude across force would change withstrength training. Twenty-two healthy men completed an 8-week strengthtraining program. During three separate testing visits (pre-test, week 4, andweek 8), the MMG signal was detected from the vastus lateralis as thesubjects performed isometric step muscle actions of the leg extensors from10–100% of maximal voluntary contraction (MVC). During pre-testing, theMMG amplitude increased linearly with force to 66% MVC and then plateaued.Conversely, weeks 4 and 8 demonstrated an increase in MMG amplitude upto ∼85% of the subject’s original MVC before plateauing. Furthermore, sevenof the ten force levels (30–60% and 80–100%) showed a significant decreasein mean MMG amplitude values after training, which consequently led to adecrease in the slope of the MMG amplitude/force relationship. The decreasesin MMG amplitude at lower force levels are indicative of hypertrophy, sincefewer motor units would be required to produce the same absolute force if themotor units increased in size. However, despite the clear changes in the meanvalues, analyses of individual subjects revealed that only 55% of the subjectsdemonstrated a significant decrease in the slope of the MMG amplitude/forcerelationship.

Keywords: MMG, resistance training, neural adaptations

Introduction

When a skeletal muscle contracts, the fibers vibrate, causing pressure waves to betransmitted outward (Herroun and Yeo 1885, Gordon and Holbourn 1948, Barry 1990).Lightweight accelerometers, laser displacement sensors, piezoelectric transducers, orcondenser microphones can be used to detect the pressure waves at the surface of the skin

1 Author to whom any correspondence should be addressed

0967-3334/12/081353+09$33.00 © 2012 Institute of Physics and Engineering in Medicine Printed in the UK & the USA 1353

1354 J M DeFreitas et al

(Beck 2010). The measurement of these pressure waves from the surface of the skin is referredto as surface mechanomyography (MMG) and its amplitude is considered to be reasonablyreliable during isometric muscle actions (Watakabe et al 2003, Herda et al 2008) but slightlyless reliable during dynamic muscle actions (Stock et al 2010a, 2010b, 2010c).

Evidence suggests that as force increases, the MMG amplitude increases with therecruitment of additional motor units (Orizio et al 1989). Once maximal motor unit recruitmenthas been achieved, the muscle depends on rate coding (i.e. increases in motor unit firing rates)to continue to increase force (Kukulka and Clamann 1981). Consequently, the MMG amplitudetypically plateaus once all motor units have been recruited. In many cases, MMG amplitudemay even decrease at high force levels due to muscular stiffness limiting the pressure waves(Orizio et al 1989). The relative force level where motor unit recruitment is complete isdependent on the muscle (Kukulka and Clamann 1981). More specifically, small musclesused for fine motor control, such as the first dorsal interosseous or adductor pollicis, onlyrely on recruitment to increase force up to approximately 30–50% of maximal voluntarycontraction (MVC) (Kukulka and Clamann 1981, De Luca et al 1982). They are consideredmaximally recruited at 30–50% MVC and, therefore, rely on rate coding to produce more force.Conversely, large muscles used for gross movement, such as the deltoid or biceps brachii, willrecruit additional motor units up to force levels as high as 80–90% MVC (Kukulka andClamann 1981, De Luca et al 1982). Since it has been suggested that the two phases (i.e.linear increase and plateau) in MMG amplitude across force reflect motor unit recruitmentand then rate coding (Orizio et al 1989), differences between large and small muscles should,accordingly, be demonstrated by MMG. In support of the hypothesis by Orizio et al (1989),studies have shown that the crossover point between the linear phase and the plateau phaseoccurs at lower force levels in small muscles (Madeleine et al 2001, Akataki et al 2003) andas expected, at higher force levels in large muscles (Orizio et al 1989, Akataki et al 2003).

Despite the significant evidence explaining the underlying mechanisms behind changes inMMG amplitude across force, there have been no studies that have successfully demonstratedthe changes that would be expected from a strength training program. One study (Ebersoleet al 2002) measured MMG amplitude prior to and after 8 weeks of isometric training, butdid not show any changes. The aim of the present study was to see if the patterns of MMGamplitude across force would change with strength training, as would be expected by theabove mentioned hypothesis. We hypothesized that, after strength training, fewer motor unitswould be needed to produce the same absolute force levels, therefore leading to a decrease inthe slope of the MMG amplitude versus force relationship.

Methods

Study design

Each subject came to the laboratory and trained 3 days per week for 8 weeks. The trainingsessions were always 48 h apart. MMG testing was performed prior to training (PRE), duringweek 4 (W4), and after the eight weeks of training (W8). A second pre-test was performed,but solely for the purpose of calculating the test–retest reliability of the measurements. Duringthe third visit of each testing week, the subjects performed the MMG step-contraction testingprocedure described below prior to performing their training.

Subjects

Twenty-two healthy men (mean ± SD age = 21.7 ± 3.7 years) volunteered to participate inthis investigation. Each participant completed an informed consent and a pre-exercise health

Effects of training on MMG amplitude 1355

and exercise status questionnaire. The questionnaire had to indicate no current or recent (withinthe past six months) neuromuscular or musculoskeletal problems to the knees, hips, or lowerback for the subject to be considered eligible for the study. In addition, each subject had to beuntrained in resistance exercise (i.e. no participation in an organized weight training programfor at least the last six months prior to the study). The study was approved by the UniversityInstitutional Review Board for Human Subjects prior to testing.

Procedures

Resistance training and testing. During their first visit, the subjects were familiarized withthe leg press and bench press exercises. This allowed the inexperienced participants to becomeaccustomed to proper lifting technique and also allowed the investigator to find an approximateestimate of the one-repetition maximums (1-RMs) for each participant. On their next visit,the subjects performed a 1-RM test on the leg and bench presses. Each subject was allowedto warm-up with progressively heavier weights prior to the first attempt. The 1-RM wasestablished to the nearest 5 lbs within —five to six attempts to avoid fatigue and used to assessbaseline strength levels for the resistance training program. The training program consistedof the bilateral incline leg press, bilateral leg extension, and bench press exercises performed3 days per week for 8 weeks. Eighty percent of the subject’s 1-RM was used for their startingweight. Each subsequent set throughout the program was adjusted as needed. Each exercise wasperformed for three sets to failure with the goal of failure occurring between 8–12 repetitions.Therefore, the weight was adjusted set by set if the subject’s load proved to be too heavy(<8 repetitions) or too light (> 12 repetitions).

Isometric strength testing. During both pre-testing sessions, as well as during weeks 4 and8, the subjects performed maximal and submaximal isometric muscle actions of the dominantleg extensors. The strength measurements were performed with the subjects seated in acustomized chair, and with the leg attached to a load cell (LC101 Series, Omega Engineering,Inc., Stamford, CT) to measure isometric leg extension force (kg). All isometric muscleactions were performed at a joint angle of 120◦ between the thigh and the leg (180◦ = fullextension). Following multiple submaximal warm-ups, the subjects performed two maximalmuscle actions to determine their MVC. The subjects then completed nine randomly orderedsubmaximal muscle actions from 10–90% of MVC. Visual feedback was provided alongwith a target template during each submaximal muscle action to ensure that the force outputwas within ± 5% of the target force level. The isometric force was held steady at the targetforce level for at least 6 s. Each muscle action was separated by 2 min rest periods. Duringsubsequent testing sessions (W4 and W8), the subjects retested their MVC, but performed thesubmaximal muscle actions at the same absolute force levels as their pre-testing (so 10–90%of the subject’s original MVC, not their new MVC). If the subject’s MVC increased, thenthe force value corresponding to their original MVC was added as an additional submaximalmuscle action.

MMG testing. During each muscle action, MMG signals were detected from the vastuslateralis (VL) muscle by securing an accelerometer (PCB Piezotronics, Model 352A24,bandwidth = 1.0 to 8000 Hz, dimensions = 0.19 × 0.48 × 0.28 inches, mass = 0.8 g,sensitivity = 100 mV g−1) to the surface of the skin over the VL. The sensor location onthe skin was prepared by shaving and cleansing with rubbing alcohol prior to securing theaccelerometer with double-sided sticky tape. The location was defined as 2/3 the distancefrom the anterior superior iliac spine to the lateral border of the patella. The site was marked

1356 J M DeFreitas et al

(A)

(B)

Figure 1. (A) Mean mechanomyographic (MMG) amplitudes across isometric force for threeseparate time points during an 8-week resistance training program. The error bars represent standarddeviation. There is an extra data point (> 100%) during the week 4 and week 8 testing sessions thatsignifies the new maximal voluntary contraction (MVC). The first ten data points were performedat force levels of 10–100% of the subject’s original MVC (from pre-testing). ∗ denotes the forcelevels in which the pre-testing MMG amplitude was significantly (p < 0.05) greater than at week4. † denotes the force levels in which the pre-testing MMG amplitude was significantly (p < 0.05)greater than at week 8. At no force level was week 4 MMG amplitude significantly different fromweek 8. (B) Bi-segmental linear regression was applied to reveal the point at which the positive,linear relationship transformed into a plateau. ∗ denotes that cross-over point for each time point.The vertical bars depict the force level at which that cross-over point occurred.

Effects of training on MMG amplitude 1357

with permanent marker on each visit (including training sessions) to ensure consistent sensorplacement across time.

Signal processing. The raw MMG signals were sampled at 2000 Hz and bandpass fitered(zero-lag, fourth-order Butterworth) with a pass band of 5–100 Hz. A 1 s epoch was selectedfrom the 6 s steady-force portion of each muscle action. MMG amplitude (m s−2) was calculatedas the root mean square for each 1 s epoch. All processing was performed using a custom-written program using LabVIEW programming software (version 8.2, National Instruments,Austin, TX, USA).

Statistical analyses. A two-way repeated measures (force × time) ANOVA was used toexamine MMG amplitude changes across force at each time point. Linearity of the MMGamplitude versus force relationship was determined by linear regression for each subject andtime point. Since MMG amplitude commonly plateaus at high force levels (see figure 1), onlythe linear, submaximal (10–60% MVC) portion was used for the linear regression analyses.In addition, a separate linear regression was applied to the high-force plateau region (80–100% MVC), extrapolated out to lower force levels, and overlaid on top of the original,low-force linear regression (see figure 1(B)). The intersection of these two lines quantifies thecross-over point from a linear increase to a plateau. Two separate one-way repeated measuresANOVAs were used to examine the linear slope coefficients and y-intercepts across time.Bonferroni pairwise comparisons were used for post-hoc follow-up analyses. The linear slopecoefficients and y-intercepts for the MMG amplitude versus force relationship for each subjectwere compared between the time points using the procedures described by Pedhazur (1997).This was done in addition to the group mean comparisons because the individual variabilityin MMG amplitude patterns across force can be high (Ryan et al 2007, Beck et al 2009).This tests the differences among regression coefficients by producing a common regressioncoefficient between the two conditions being tested. The test then determines if the incrementin the regression sum of squares due to the use of separate slopes is significantly different(when compared to the common regression sum of squares). If no statistical difference is foundbetween the two slope coefficients then it can be concluded that a common coefficient couldbe appropriately applied to both. If the slope comparison from two separate conditions wasfound to not be significantly different (and therefore parallel) then a follow-up test was run todetermine if the y-intercepts were significantly different. As recommended by Pedhazur (1997,p 563), an alpha of p � 0.10 was used for the individual slope and y-intercept comparisons tominimize type II error. An alpha level of p � 0.05 was used for all ANOVA analyses. In addition,data from the two pre-testing sessions allowed test–retest reliability of the MMG measurementsto be assessed using paired samples t-tests, two-way fixed-effect intra-class correlations (ICC;model 3,1), the standard error of the measurement, and the minimal difference for a change tobe considered real (Weir 2005).

Results

The mean MMG changes with force across the training program are shown in figure 1(A).Included in the figure are the pairwise comparisons. The results showed that the MMGamplitude values during week 4 were significantly less than those from pre-testing at 30–60%MVC. Additionally, the MMG amplitude values during week 8 were significantly less thanthose from pre-testing at 30–50% and 80–100% MVC. At no force level was week 4 MMGamplitude significantly different from week 8. The linear regression results of the MMG

1358 J M DeFreitas et al

Table 1. The mean linear regression coefficients of the mechanomyographic (MMG) amplitudeversus force relationship for each time point.

Pre Week 4 Week 8

Slope (m s−2 kg−1) 0.0074 0.0049a 0.0048a

y-intercept (m s−2) 0.0224 0.0387 0.0409aA significant (p < 0.05) difference from pre-testing.

Table 2. Test retest reliability results for mechanomyographic (MMG) amplitude, maximumvoluntary contractions (MVC), and the linear regression coefficients for the MMG amplitudeversus force relationship from 10–60% MVC. P1 = first pre-testing session; P2 = second pre-testing session; ICC = intraclass correlation coefficient; SEM = standard error of the measurement;MD = minimal difference for a change to be considered real.

P1 Mean P2 Mean ICC3,1 SEM MD p-value

MMG Amplitude (m s−2)At 10% MVC 0.091 0.092 0.818 0.0216 0.0423 0.739At 20% MVC 0.122 0.113 0.768 0.0331 0.0647 0.245At 30% MVC 0.167 0.154 0.827 0.0514 0.1004 0.268At 40% MVC 0.231 0.239 0.877 0.0623 0.1220 0.547At 50% MVC 0.297 0.277 0.895 0.0626 0.1223 0.150At 60% MVC 0.352 0.363 0.747 0.1002 0.1962 0.599At 70% MVC 0.362 0.380 0.709 0.0949 0.1855 0.395At 80% MVC 0.383 0.369 0.772 0.0664 0.1298 0.339At 90% MVC 0.379 0.399 0.747 0.0716 0.1396 0.248At 100% MVC 0.384 0.404 0.263 0.1329 0.2593 0.488

Slopes (m s−2 kg-1) 0.0074 0.0077 0.756 0.0025 0.0048 0.568y-intercepts (m s−2) 0.0224 0.0120 0.615 0.0408 0.0794 0.241MVC (kg) 73.7 74.7 0.951 4.3092 8.4390 0.246

amplitude versus force relationship are shown for each time point in table 1. Figure 1(B)shows the results of the bi-segmental linear regression, intended to demonstrate the cross-overpoint to the plateau region. One of 22 subjects showed a significant difference in linear slopecoefficients between the two pre-testing sessions and was excluded from further analyses.Of the remaining 21 subjects that demonstrated reliable slopes during pre-testing (i.e. nosignificant difference between pre-tests), 12 demonstrated significant changes in their slopesafter training. Therefore, of those 21 subjects that showed reliable pre-tests, 9 did not show asignificant change in slope with training. Of those remaining nine subjects that did not show asignificant change in slope, four showed a significant change in their y-intercept after training.Results of the test–retest reliability measures are shown in table 2.

Discussion

The results from the present study indicated that resistance training altered both MMGamplitude and its pattern of response across force in previously untrained individuals. Sevenof the ten force levels (30–60% and 80–100%) showed a significant decrease in mean MMGamplitude values after training. It has been suggested that the linearly increasing portion ofthe MMG amplitude/force relationship represents increases in motor unit recruitment, whilethe plateau or decrease at higher force levels represents rate coding becoming the primarymodulator of force output (Orizio et al 1989). Therefore, the decreases in MMG amplitudesshown in the present study at lower force levels (30–60%) likely represented a decrease inmotor unit recruitment. Since the same absolute force levels were used throughout the study,

Effects of training on MMG amplitude 1359

this decrease would be indicative of skeletal muscle hypertrophy. Hypothetically, increasedcontractile protein content would cause each fiber, and therefore each motor unit, to be able toproduce more force. So after the occurrence of skeletal muscle hypertrophy, fewer motor unitswould be required to produce the same amount of force when compared to before training. Thishypothesis can be confirmed since muscle cross-sectional area (CSA) was measured with thesame subjects in conjunction with the present study, but as a separate experiment (DeFreitaset al 2011). In those findings, both muscle CSA and MVC were significantly greater at week4 than at pre-testing. As would be expected, the decreases in mean MMG amplitude valuesat multiple force levels in the present study led to a significant decrease in the slope of theMMG amplitude/force relationship. If the point at which that relationship begins to plateauindeed represents a crossover point from maximal motor unit recruitment to rate coding, then arightward shift of that point should be expected as a result of strength training (i.e. the point ofmaximal recruitment should now occur at a higher force level). This was shown in figure 1(B)with the plateau beginning at approximately 66% MVC prior to training and shifting to about83–87% MVC after training. It should be noted that these results contradict previous findingsin which no changes were seen in MMG amplitude for the biceps brachii after 8 weeks oftraining (Ebersole et al 2002). However, it is possible that the disparity may be accounted forby differences in the training protocol; the subjects in Ebersole et al (2002) performed —threeto five sets of isometric training, while the subjects in the present study performed three setswith a dynamic constant external resistance (e.g. leg press) and three sets with a dynamicvariable resistance (e.g. leg extension).

Previous findings have suggested the importance of subject-by-subject comparisons whenexamining MMG amplitude and mean frequency patterns (Ryan et al 2007). This suggestion isdue to the high degree of inter-individual variability in the patterns across force. As mentionedpreviously, tests for differences among regression coefficients were applied to each individualsubject’s data. This assessment allowed us to determine how many of the subjects showedchanges that were similar to those of the group. Twenty-one of the 22 subjects demonstrated areliable MMG amplitude versus force slope coefficient (i.e. no significant difference betweenthe two pre-testing sessions). Sixteen of those 21 showed a significant change in a regressioncoefficient with training. However, four of those were changes in the y-intercept, which didnot occur for the group mean data (table 1). Overall, 12 of the subjects demonstrated the sametraining-induced change in the linear slope coefficient that was demonstrated in the group data(table 1). All 22 subjects showed an increase in muscle CSA (DeFreitas et al 2011) and allbut one (i.e. 21) improved their MVC. Therefore, the reason for this dichotomy (the 12 thatchanged in slope versus the 9 that did not) is unknown and cannot be accounted for by lack ofchanges in muscle size or strength.

Another interesting finding is the decrease in maximal MMG amplitude. As mentionedpreviously, the 100% MVC value shown in figure 1(A) reflects the same absolute force levelat all three trials, based on the pre-testing MVC. Therefore, the decrease in MMG amplitudeat later time points is consistent with the increases in strength since that force level wouldsignify a lower percent of MVC after training. However, the values in figure 1(A) depicted as‘> 100% MVC’ represent the new, stronger MVCs after training, and those maximal MMGamplitude values are also lower than the pre-test MVC value. One possible explanation isthat the skeletal muscle hypertrophy was accompanied by an increase in musculotendinousstiffness. An increase in stiffness can lead to both decreases in MMG amplitude (Orizio et al1989) and increases in force (due to better transmission of force from the fascicles to thetendon). Theoretically, an increase in tissue CSA results in a stiffer material (Butler et al1978, see figure 10), although it should be noted that this work was performed in tendons andligaments. Additionally, Klinge et al (1997) showed an increase in musculotendinous stiffness

1360 J M DeFreitas et al

after 13 weeks of isometric strength training. Although this is a convenient and temptingexplanation, further research needs to be performed which simultaneously tracks changes inMMG amplitude, muscle size and musculotendinous stiffness with strength training beforeconclusions can be drawn. It should also be noted that the 100% MVC is the only force trialthat did not produce an acceptable reliability coefficient (ICC = 0.26) for MMG amplitude(table 2), which introduces further reservations about drawing any conclusions about themaximal MMG values.

In summary, the mean MMG amplitude values and their pattern across force were alteredby the eight weeks of resistance training. There was a decrease in the slope of the linearlyincreasing portion of the MMG amplitude/force relationship (10–60%), as well as a shiftto higher absolute force levels for the plateau in MMG amplitude. Additionally, there was adecrease in maximal MMG amplitude after training. These changes were likely due to skeletalmuscle hypertrophy, and possibly even an increase in musculotendinous stiffness. However,for unknown reasons, analyses of individual subjects revealed that only 55% of the subjectsdemonstrated a significant decrease in the slope of the MMG amplitude/force relationship.

References

Akataki K, Mita K, Watakabe M and Itoh K 2003 Mechanomyographic responses during voluntary ramp contractionsof the human first dorsal interosseous muscle Eur. J. Appl. Physiol. 89 520–5

Barry D 1990 Acoustic signals from skeletal muscle Physiology 5 17 (available at http://physiologyonline.physiology.org/content/5/1/17.short)

Beck T W 2010 The mechanisms underlying the surface mechanomyogram Application of Mechanomyography forExamining Muscle Function ed T W Beck (Kerala: Transworld Research Network chapter 1) pp 1–16

Beck T W, DeFreitas J M and Stock M S 2009 The linearity and reliability of the mechanomyographic amplitudeversus submaximal isometric force relationship Physiol. Meas. 30 1009–16

Butler D L, Grood E S, Noyes F R and Zernicke R F 1978 Biomechanics of ligaments and tendons Exerc. Sport Sci.Rev. 6 125–82

DeFreitas J M, Beck T W, Stock M S, Dillon M A and Kasishke P R II 2011 An examination of the time course oftraining-induced skeletal muscle hypertrophy Eur. J. Appl. Physiol. 111 2785–90

De Luca C J, LeFever R S, McCue M P and Xenakis A P 1982 Behaviour of human motor units in different musclesduring linearly varying contractions J. Physiol. 329 113–28

Ebersole K T, Housh T J, Johnson G Q, Perry S R, Bull A J and Cramer J T 2002 Mechanomyographic andelectromyographic responses to unilateral isometric training J. Strength Cond. Res. 16 192–201

Gordon G and Holbourn A 1948 The sounds from single motor units in a contracting muscle J. Physiol. 107 456–64Herda T J, Ryan E D, Beck T W, Costa P B, DeFreitas J M, Stout J R and Cramer J T 2008 Reliability of

mechanomyographic amplitude and mean power frequency during isometric step and ramp muscle actionsJ. Neurosci. Methods 171 104–9

Herroun E and Yeo G F 1885 Note on the sound accompanying the single contraction of skeletal muscle J. Physiol.6 287–92

Klinge K, Magnusson S P, Simonsen E B, Aagaard P, Klausen K and Kjaer M 1997 The effect of strength andflexibility training on skeletal muscle electromyographic activity, stiffness, and viscoelastic stress relaxationresponse Am. J. Sports Med. 25 710–6

Kukulka C G and Clamann H P 1981 Comparison of the recruitment and discharge properties of motor units in humanbrachial biceps and adductor pollicis during isometric contractions Brain Res. 219 45–55

Madeleine P, Bajaj P, Søgaard K and Arendt-Nielsen L 2001 Mechanomyography and electromyography forcerelationships during concentric, isometric and eccentric contractions J. Electromyogr. Kinesiol. 11 113–21

Orizio C, Perini R and Veicsteinas A 1989 Muscular sound and force relationship during isometric contraction in manEur. J. Appl. Physiol. 58 528–33

Pedhazur E J 1997 Continuous and categorical independent variables: I. Attribute-treatment interaction; comparingregression equations Multiple Regression in Behavorial Research: Explanation and Prediction (Fort Worth, TX:Harcourt Brace College Publishers) pp 560–71

Ryan E D, Cramer J T, Housh T J, Beck T W, Herda T J and Hartman M J 2007 Inter-individual variability in thetorque-related patterns of responses for mechanomyographic amplitude and mean power frequency J. Neurosci.Methods 161 212–9

Effects of training on MMG amplitude 1361

Stock M S, Beck T W, DeFreitas J M and Dillon M A 2010a Linearity and reliability of the mechanomyographicamplitude versus concentric dynamic constant external resistance relationships for the bench press exerciseJ. Strength Cond. Res. 24 785–95

Stock M S, Beck T W, DeFreitas J M and Dillon M A 2010b Linearity and reliability of the mechanomyographicamplitude versus dynamic constant external resistance relationships for the biceps brachii Physiol.Meas. 31 1487–98

Stock M S, Beck T W, DeFreitas J M and Dillon M A 2010c Linearity and reliability of the mechanomyographicamplitude versus dynamic torque relationships for the superficial quadriceps femoris muscles MuscleNerve 41 342–9

Watakabe M, Mita K, Akataki K and Ito K 2003 Reliability of the mechanomyogram detected with an accelerometerduring voluntary contractions Med. Biol. Eng. Comput. 41 198–202

Weir J P 2005 Quantifying test–retest reliability using the intraclass correlation coefficient and the SEM J. StrengthCond. Res. 19 231–40