ARTICLE IN PRESSG ModelNSM-4847; No. of Pages 6

Journal of Neuroscience Methods xxx (2008) xxx–xxx

Contents lists available at ScienceDirect

Journal of Neuroscience Methods

journa l homepage: www.e lsev ier .com/ locate / jneumeth

Reliability of mechanomyographic amplitude and mean powerfrequency during isometric step and ramp muscle actions

Trent J. Herda, Eric D. Ryan, Travis W. Beck, Pablo B. Costa,Jason M. DeFreitas, Jeffrey R. Stout, Joel T. Cramer ∗

-6081

reliabvastu

mp md tw5, 85ys. Tdiffentify

the stCs rato18

hat w

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

a r t i c l e i n f o

Article history:Received 24 January 2008Received in revised form 22 February 2008Accepted 25 February 2008

Keywords:MechanomyographyReliabilityStepRamp

a b s t r a c t

This study examined thequency (MMGMPF) for thesubmaximal step and raage = 24 ± 4 yrs) performe5, 15, 25, 35, 45, 55, 65, 7trials separated by 3–5 dament (SEM), and minimalANOVAs were used to qua3 trials at 5% MVC duringaction for MMGMPF. The ICfrom 14.9 to 33.9% and 6.9demonstrated reliability t

Please cite this article in press as: Herda TJ, et al., Reliability of mechanomyand ramp muscle actions, J Neurosci Methods (2008), doi:10.1016/j.jneum

1. Introduction

The reliability of a measurement refers to the consistency of atest over a certain number of trials. Often, the relative consistencyof a measurement is quantified using reliability coefficients calledintraclass correlation coefficients (ICCs) (Streiner and Norman,1995). Relative reliability refers to the consistency of the positionor rank of individuals in the group relative to others. The ICC isunitless, and can “theoretically vary between 0 and 1.0, where anICC of 0 indicates no reliability and an ICC of 1.0 indicates per-fect reliability (Weir, 2005).” Another common reliability measureis the standard error of the measurement (SEM), which providesan absolute index of reliability. Absolute reliability refers to theconsistency of scores of individuals. The SEM gives an indicationon the precision of the measurement and, therefore, can providean index of the expected trial-to-trial noise for the variable beingexamined (Weir, 2005). In addition, the minimal difference (MD)is derived from the SEM and indicates the minimum differenceneeded between the trials to be considered real (Weir, 2005). Col-

∗ Corresponding author at: Biophysics Laboratory, Department of Health andExercise Science, University of Oklahoma, 1401 Asp Avenue, Rm 12, Norman, OK73019-6081, USA. Tel.: +1 405 325 5211; fax: +1 405 325 0594.

E-mail address: jcramer@ou.edu (J.T. Cramer).

0165-0270/$ – see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.jneumeth.2008.02.017

, USA

ility of mechanomyographic amplitude (MMGRMS) and mean power fre-s lateralis during isometric maximal voluntary contractions (MVCs) anduscle actions of the leg extensors. Nineteen participants (mean ± S.D.

o isometric MVCs, 10 randomly ordered isometric step muscle actions at, and 95% of MVC, and two 6 s isometric ramp muscle actions during 3

he intraclass correlation coefficient (ICC), standard error of the measure-rence (MD) were used for test–retest reliability, while repeated measuressystematic variability. Systematic variability was only present across theep muscle action for MMGRMS and at 35% of MVC during the ramp musclenged from 0.39 to 0.89 and 0.36 to 0.80, while the SEM (% of mean) ranged.9% for MMGRMS and MMGMPF, respectively. Thus, MMGRMS and MMGMPF

as comparable to that of the surface electromyographic signal.© 2008 Elsevier B.V. All rights reserved.

lectively, these measures provide a comprehensive assessment ofreliability.

Surface mechanomyography (MMG) has become a fairly com-mon technique for examining muscle function in the movement

ographic amplitude and mean power frequency during isometric stepeth.2008.02.017

sciences. Despite the growth in MMG research in recent years, how-ever, there is relatively little known regarding the reliability of theMMG signal. For example, previous investigations have reportedreliability coefficients ranging from 0.94 to 0.98 and 0.72 to 0.82 forMMG amplitude (MMGRMS) and mean power frequency (MMGMPF),respectively (Smith et al., 1997; Cramer et al., 2002; Ebersole et al.,2002). In addition, Akataki et al. (1999) examined the reliability ofMMGRMS for the biceps brachii during submaximal isometric stepmuscle actions of the forearm flexors at 10–90% at of the maxi-mal voluntary contraction (MVC). However, the authors (Akatakiet al., 1999) reported only the average ICC across all percents ofthe MVC (ICC = 0.83) for MMGRMS, and the frequency componentof the MMG signal was not examined. Given that the literatureon the reliability of the MMG signal has focused mainly on ICCs,without assessing the SEM and MD, additional studies are neededto examine the reliability of the MMG signal in a more com-prehensive manner. Therefore, the purpose of this investigationwas to use the ICC, SEM and MD to examine the reliability ofMMGRMS and MMGMPF from the vastus lateralis during isometricMVCs and submaximal step and ramp muscle actions of the legextensors.

IN PRESSoscience Methods xxx (2008) xxx–xxx

ARTICLEG ModelNSM-4847; No. of Pages 6

2 T.J. Herda et al. / Journal of Neur

2. Methods

2.1. Subjects

Nineteen healthy men (mean ± S.D. age = 24 ± 4 yrs; stature =173 ± 9 cm; mass = 75 ± 13 kg) volunteered for this investigation.None of them reported any current or ongoing neuromuscular dis-eases or musculoskeletal injuries that involved the ankle, knee,or hip joints. This study was approved by the University of Okla-homa Institutional Review Board. All participants read and signedan informed consent form and completed a Pre-Exercise TestingHealth & Exercise Status Questionnaire.

2.2. Research design

Isometric torque for the right leg extensor muscles was mea-sured using the analog torque signal extracted from a BiodexSystem 3 isokinetic dynamometer (Biodex Medical Systems Inc.,Shirley, NY). The participants were seated with restraining strapsover the pelvis, trunk, and contralateral thigh, and the lateralcondyle of the femur was aligned with the input axis of thedynamometer in accordance with the Biodex User’s Guide (BiodexPro Manual, Applications/Operations. Biodex Medical Systems Inc.,Shirley, NY. 1998). All isometric torque assessments were per-formed at a leg flexion angle of 120◦, with full extension of theleg at the knee joint considered to be 180◦. A standard hand-held goniometer was used to determine the 120◦ leg flexionangle.

Each participant visited the laboratory on four occasions, oncefor a familiarization trial and three separate experimental tri-als. During the familiarization trial, participants performed two4-s isometric maximal voluntary contractions (MVCs), severalsubmaximal isometric step muscle actions at randomly orderedpercentages of the MVC (%MVC = 5, 15, 25, 35, 45, 55, 65, 75, 85,and 95%), and several isometric ramp muscle actions (describedlater) until they successfully completed the protocol. The purpose ofthe familiarization trial was for the participants to gain experiencewith the testing equipment and to practice the MVCs as well as thesubmaximal step and ramp muscle actions. Within 2–5 days afterthe familiarization trial, participants reported to the laboratory fortheir first experimental trial. Each subsequent experimental trialwas also separated by 2–5 days and occurred at the same time ofday (±2 h). During each experimental trial, participants performedthe following procedures: (a) two pre-testing isometric MVCs, (b)ten submaximal isometric step muscle actions, and (c) two 6-s

Please cite this article in press as: Herda TJ, et al., Reliability of mechanomyand ramp muscle actions, J Neurosci Methods (2008), doi:10.1016/j.jneum

isometric ramp muscle actions. In addition, the order of the sub-maximal isometric step and ramp muscle actions was randomlyadministered during each experimental trial.

During the isometric step and ramp muscle actions, partici-pants were required to track their torque production on a computermonitor placed in front of them that displayed the real-time, digi-tized torque signal overlaid onto a programmed template. For theisometric step contraction, horizontal lines were programmed astemplates on the computer monitor that served as the target torquelines for each submaximal step muscle action. The ramp templateconsisted of a 5-s horizontal baseline at 5% MVC followed by alinearly increasing ramp line lasting 6 s and ending with a 2-s hor-izontal plateau at 100% MVC (Bilodeau et al., 2003). The isometricstep and ramp muscle action templates and real-time torque over-lay was programmed and displayed using LabVIEW 7.1 software(National Instruments, Austin, TX).

2.3. Isometric step muscle actions

During each experimental trial, participants performed a seriesof randomly ordered submaximal isometric step muscle actions at

Fig. 1. An example of the mechanomyographic (MMG) sensor placement.

5, 15, 25, 35, 45, 55, 65, 75, 85, and 95% of the highest pre-testingMVC value. Each isometric step muscle action was held steady forapproximately 4 s. A 2 min rest was allowed between each muscleaction.

2.4. Isometric ramp muscle actions

During each experimental trial, participants were also askedto perform 2–3 ramp muscle actions (5–100% MVC) and verbalencouragement was provided. A 2-min rest period was allowedbetween each ramp muscle action. The ramp trial that best satis-fied the following criteria were used for later analysis: (a) maximalforce reaching at least 95% of MVC and (b) a maximum trackingerror of ± 3% MVC around the template.

2.5. Mechanomyography

The MMG signal was detected with an active miniatureaccelerometer (EGAS-FS-10-/VO5, Measurement Specialties Inc.,Hampton, VA) that was pre-amplified with a gain of 200, frequencyresponse of 0–200 Hz, sensitivity of 70 mV/m s−2, and range of±98.1 m s−2. The accelerometer was placed over the vastus later-alis (VL) on the lateral/anterior portion of the muscle at 50% of thedistance between the greater trochanter and lateral condyle of the

ographic amplitude and mean power frequency during isometric stepeth.2008.02.017

femur and was fixed to the skin with 3 M double-sided foam tape.Fig. 1 shows the position for the MMG sensor placement.

2.6. Signal processing

The MMG and torque signals were recorded simultaneouslywith a data acquisition system (MP150WSW, Biopac systems Inc.,Santa Barbara, CA) during each isometric MVC, step, and ramp mus-cle action. The torque (Nm) signal from the dynamometer and MMG(m s−2) signal recorded from the VL were sampled at 1 kHz. All sig-nals were stored on a personal computer (Dell Inspiron 8200, DellInc., Round Rock, TX), and processing was completed off-line usingcustom written software (LabVIEW V 7.1, National Instruments,Austin, TX). The MMG signal was digitally filtered (zero-phase 4th-order Butterworth filter) with a pass band of 5–100 Hz, while thetorque signal was low-pass filtered with a 10 Hz cutoff (zero-phase4th-order Butterworth filter). All subsequent analyses were per-formed on the filtered signals.

Since isometric peak torque (PT, Nm) was determined as thehighest 1-s average torque value that occurred during the 4-s MVCplateau, the same 1-s epoch was selected from the MMG signal of

INoscien

ARTICLEG ModelNSM-4847; No. of Pages 6

T.J. Herda et al. / Journal of Neur

the VL muscle to calculate the time and frequency domain esti-mates during the MVCs. For the submaximal step muscle actions,the most stable 1 s epoch of the 4 s torque plateau was selected tocalculate the submaximal torque level, MMG amplitude (MMGRMS),and MMG mean power frequency (MMGMPF) values. For the isomet-ric ramp muscle actions, 0.6 s epochs were extracted every 0.1 s forthe torque and MMG signals (Akataki et al., 2001; Ryan et al., 2007,2008a,b). The amplitude was represented as the root-mean-square(RMS) value. For the MMGMPF analyses, each MMG signal epoch wasprocessed with a Hamming window and a discrete Fourier trans-form (DFT). The MPF was selected to represent the power spectrumon the basis of the recommendations of Diemont et al. (1988) andwas calculated as described by Kwatny et al. (1970).

Please cite this article in press as: Herda TJ, et al., Reliability of mechanomyand ramp muscle actions, J Neurosci Methods (2008), doi:10.1016/j.jneum

2.7. Statistical analyses

For submaximal isometric torque, MMGRMS, and MMGMPF at 5,

15, 25, 35, 45, 55, 65, 75, 85, 95, and 100% MVC for the step, ramp,and MVC muscle actions, one-way repeated measures analyses ofvariance (ANOVAs) were used to examine the means among thethree trials. Consequently, the repeated measures ANOVAs assessedthe level of systematic variability among the trials. In addition, theintraclass correlation coefficient (ICC) represented the relative con-sistency (test–retest reliability), the standard error of measurement(SEM) represented absolute consistency, and an estimate of theminimal difference (MD) needed to be considered real was calcu-lated across trials for each percentage of MVC. Both the SEM andMD values were expressed as a percentage of the PT, MMGRMS, andMMGMPF values (%MVC) recorded during the MVC trial that yieldedthe highest isometric PT.

Model “2,1” from Shrout and Fleiss (1979) was used to calculatethe ICC. Model 2,1 is a 2-way random factor model that utilizes ran-dom and systematic error in the denominator of the ICC equation,

Table 1Reliability and measurement variability statistics for isometric torque, mechanomyograpramp muscle actions at each submaximal percentage of the maximal voluntary contractio

P-value = type I error rate for the one-way repeated measures ANOVA across trials 1, 2, aSEM = standard error of measurement, expressed as a percentage of the mean. MD = minReflects a significant (P ≤ 0.05) difference across trials.

PRESSce Methods xxx (2008) xxx–xxx 3

thus, the ICCs generated with this model can be generalized to otherlaboratories and testers (Weir, 2005). The ICC (2,1) was calculatedwith the following equation (Shrout and Fleiss, 1979):

ICC = MSB − MSE

MSS + (k − 1)MSE + (k(MST − MSE)/n)

All components of the ICC (2,1) equation were derived from theoutput of the one-way repeated measures ANOVAs (trial 1 vs. trial2 vs. trial 3). MSB represents the mean square between-subjects,MSE is the mean square error, MSS is the mean square for subjects,MST is the mean square total, k is the number of trials, and n is thesample size. SEM for model 2,1 was calculated with the followingequation (Hopkins, 2000):

ographic amplitude and mean power frequency during isometric stepeth.2008.02.017

SEM =√

MSE

MD for model 2,1 was calculated with the following equation (Weir,2005):

MD = SEM × 1.96 ×√

2

Most stasticial analyses were performed using SPSS v. 12.0 (SPSSInc., Chicago, IL), however, the calculations of ICC, SEM, and MDwere performed with custom-written software (LabVIEW v 7.1,National Instruments, Austin, TX) based on the article by Weir(2005). A type I error rate of 5% was used to determine statisticalsignificance.

3. Results

The reliability and measurement variability statistics for isomet-ric torque, MMGRMS, and MMGMPF during the step and ramp muscleactions are presented in Table 1. Table 2 contains the same statis-tics for the MVCs, which were common to both the step and rampmuscle actions.

hic amplitude (MMGRMS) and mean power frequency (MMGMPF) for the step andn (MVC)

nd 3. ICC2,1 = intraclass correlation coefficient, model 2,1 (Shrout and Fleiss, 1979).imum difference to be considered real, expressed as a percentage of the mean. (*)

INoscien

yograp

ICC

0.910.810.62

1, 2, amini

ARTICLEG ModelNSM-4847; No. of Pages 6

4 T.J. Herda et al. / Journal of Neur

Table 2Reliability and measurement variability statistics for peak torque (PT), mechanomvoluntary contractions

P-value

PT 0.123MMGRMS 0.593MMGMPF 0.695

P-value = type I error rate for the one-way repeated measures ANOVA across trialsSEM = standard error of measurement, expressed as a percentage of the mean. MD =

Please cite this article in press as: Herda TJ, et al., Reliability of mechanomyand ramp muscle actions, J Neurosci Methods (2008), doi:10.1016/j.jneum

4. Discussion

In the present investigation, the ICCs ranged from 0.39 to 0.89and 0.36 to 0.80 for MMGRMS and MMGMPF, respectively (Fig. 2).Systematic variability was present at 5% of MVC during the stepmuscle action for MMGRMS (P = 0.012) and at 35% of MVC during theramp muscle action for MMGMPF (P = 0.044). For MMGRMS, the high-est ICCs were recorded during muscle actions equal to or greaterthan 35% of MVC (ICC > 0.70), except for the ramp muscle action at55% of MVC (ICC = 0.67). The lowest ICCs occurred at muscle actionsequal to or less than 25% of MVC (ICC ≤ 0.65), except for the stepmuscle action at 5% of MVC (ICC = 0.76). For MMGMPF, there was arelative consistency through out all muscle actions and there wasnot a clear difference between ICCs at different percents of MVC.For the most part, ICCs for MMGRMS increased as force increased;however, this association with force was unseen with MMGMPF. Apossible reason why the ICCs for MMGRMS increased with force isbecause the magnitude of an ICC reflects the between-subjects vari-ability (Weir, 2005). It is known that a greater between-subjectsvariability will result in a higher ICC (Weir, 2005) when compared

Fig. 2. Intraclass correlation coefficients for (a) mechanomyographic amplitude (MMGRMS

trials 1, 2, and 3 for each submaximal percentage of the maximal voluntary contraction (M

PRESSce Methods xxx (2008) xxx–xxx

hic amplitude (MMGRMS) and mean power frequency (MMGMPF) for the maximal

SEM (%) MD (%)

6.8 18.918.1 50.715.8 43.7

nd 3. ICC2,1 = intraclass correlation coefficient, model 2,1 (Shrout and Fleiss, 1979).mum difference to be considered real, expressed as a percentage of the mean.

ographic amplitude and mean power frequency during isometric stepeth.2008.02.017

to a lower between-subjects variability. For example, the range indata for MMGRMS during the step muscle action at 5% of MVC was0.117 m s−2rms and at 95% of MVC was 0.515 m s−2rms, thus, greaterbetween-subject variability at 95% of MVC likely contributed to thehigher ICC compared to 5% of MVC. For this reason, the ICC is con-sidered to be context specific for a test (Weir, 2005). As noted byStreiner and Norman (1995), “There is literally no such thing asthe reliability of a test, unqualified; the coefficient has meaningonly when applied to specific populations.” Although the ICC mea-surement is considered to be context specific, the present study’sresults align with previous investigations that have examined thereliability of electromyographic (EMG) amplitude and frequencydata during submaximal isometric step muscle actions (EMGRMSICC = 0.83 averaged across 10–90% of MVC) (Akataki et al., 1999),isometric ramp muscle actions (EMGRMS ICC = 0.76 averaged across10–90% of MVC; EMG median frequency ICC = 0.71 averaged across10–90% of MVC) (Kellis and Katis, in press), and isometric MVCs(EMGRMS ICC = 0.76) (Yang and Winter, 1983), which suggests thatthe ICCs for the MMG data in the present study may be at an accept-able level.

) and (b) mechanomyographic mean power frequency (MMGMPF) calculated acrossVC).

ARTICLE IN PRESSG ModelNSM-4847; No. of Pages 6

T.J. Herda et al. / Journal of Neuroscience Methods xxx (2008) xxx–xxx 5

ue forperce

Fig. 3. Standard errors of measurement expressed as a percentage of the mean valpower frequency (MMGMPF) calculated across trials 1, 2, and 3 for each submaximal

In general, the magnitudes of the ICCs were slightly lower, whilethe SEMs were consistently higher for the ramp than the step mus-cle actions for MMGRMS and MMGMPF (Fig. 3). There are a couple ofpossible explanations for differences in the ICCs and SEMs betweenthe ramp and step muscle actions: (1) shorter epoch durationswere used during the ramp when compared to the step muscleactions; (2) nonstationarity of the MMG signal during the rampmuscle action. A 1 s epoch was selected during the 4 s submaxi-mal step muscle actions, while during the ramp muscle action, a

Please cite this article in press as: Herda TJ, et al., Reliability of mechanomyand ramp muscle actions, J Neurosci Methods (2008), doi:10.1016/j.jneum

0.6 s epoch was recorded every 0.1 s. It is possible that the smallerepoch duration could have had an affect on the repeatability ofthe signal at a given percentage of the MVC during the ramp mus-cle action. Previous investigations (Inbar and Noujaim, 1984; Inbaret al., 1986; Paiss and Inbar, 1987) have indicated that during iso-metric muscle actions, the EMG signal is stationary for epochs of0.5–1.0 s. Therefore, MMG and EMG signals are considered to be atleast weakly stationary (Beck et al., 2005) during isometric mus-cle actions, which allows them to be processed with algorithmsthat require the assumption of signal stationarity (i.e., the discreteFourier transform) (Beck et al., 2005). However, it has been reportedthat during dynamic muscle actions, the EMG signal is nonstation-ary due to movement of the muscle fibers beneath the recordingelectrode(s), alterations in the number of active motor units, and/orchanges in muscle fiber lengths (Bonato et al., 1996; Karlsson etal., 2001). Although an isometric ramp muscle action is not con-sidered a dynamic task, it does require changes in the number ofactive motor units during the 6 s ramp from 0 to 95% of MVC. There-fore, an isometric ramp muscle action does not meet one of theunderlying assumptions of signal stationarity according to Bonatoet al. (1996) and Karlsson et al. (2001). Although the signal anal-

(a) mechanomyographic amplitude (MMGRMS) and (b) mechanomyographic meanntage of the maximal voluntary contraction (MVC).

ysis techniques used for the isometric ramp muscle actions in thepresent study (i.e., 0.6 s epochs that overlap in 0.1 s intervals) weredesigned to compensate for signal nonstationarity, it is possiblethat the epoch duration was too long, resulting in a violation of thestationarity assumption and higher SEM values for the ramp thanthe step muscle actions. Furthermore, it should be noted that theorder of submaximal isometric step muscle actions was random-ized, whereas the order was not randomized for the ramp muscleactions. Therefore, future research should examine the effects of

ographic amplitude and mean power frequency during isometric stepeth.2008.02.017

different epoch durations on signal stationarity during ramp muscleactions as well as the potential order effect of ramp muscle actions.

The results from the present study also indicated that at lowerisometric force levels (≤25% of MVC) during both the ramp and stepmuscle actions, there were lower ICCs for MMGRMS and MMGMPFwhen compared to those at higher force levels. The cause for thesedifferences is unclear; however, it is possible that the lower ICCswere due to lower between-subjects variability, which in turn, mayresult from the inherently lower signal-to-noise ratios at low forcelevels (≤25% of MVC). In theory, at lower force levels fewer musclefibers were being activated. Since the MMG signal may reflect motorunit recruitment up to 60–80% MVC (Ryan et al., 2008b), lowerMMG signal amplitude values would have resulted in lower signal-to-noise ratios. Consequently, lower signal-to-noise ratios mayhave reduced the between-subject variability. As stated previously,low between-subjects variability may cause a decrease in the ICC(Weir, 2005). The signal-to-noise ratio compares the magnitude ofthe background noise to that of the MMG signal; thereby providinga measure of the quality of the signal. Therefore, increases in muscleactivity at greater force levels should result in a greater signal-to-noise ratio (assuming similar background noise at all levels) when

INoscien

tendinosus and biceps femoris muscles during ramp isometric contractions. J

ARTICLEG ModelNSM-4847; No. of Pages 6

6 T.J. Herda et al. / Journal of Neur

compared to lower force levels. Theoretically, a smaller signal-to-noise ratio may cause a greater variability between-subjects anda decrease in the ICC. Future research is needed to examine theeffects of the signal-to-noise ratio on the repeatability of the MMGsignal.

In summary, the results from the present investigation indicatedthat the ICCs from surface MMG measurements of the vastus lat-eralis during the isometric step and ramp muscle actions rangedfrom 0.39 to 0.89 and 0.36 to 0.80 for MMGRMS and MMGMPF,respectively. Reliability was slightly higher during the step mus-cle actions compared to the ramp muscle actions and, overall, wasgreater at high force levels than at low force levels. It is concludedthat MMGRMS and MMGMPF measured across days using isometricstep and ramp muscle actions of the vastus lateralis demonstrated

Please cite this article in press as: Herda TJ, et al., Reliability of mechanomyand ramp muscle actions, J Neurosci Methods (2008), doi:10.1016/j.jneum

acceptable reliability. However, future research is needed to exam-ine the effects of different MMG epoch durations during rampmuscle actions and the effects of the signal-to-noise ratio at lowforce levels on the repeatability of the MMG signal. In addition,future studies should investigate the repeatability of the patternsof responses for MMGRMS and MMGMPF from isometric step andramp muscle actions.

References

Akataki K, Mita K, Itoh Y. Repeatability study of mechanomyography in submaximalisometric contractions using coefficient of variation and intraclass correlationcoefficient. Electromyogr Clin Neurophysiol 1999;39:161–6.

Akataki K, Mita K, Watakabe M, Itoh K. Mechanomyogram and force relationshipduring voluntary isometric ramp contractions of the biceps brachii muscle. EurJ Appl Physiol 2001;84:19–25.

Beck TW, Housh TJ, Johnson GO, Weir JP, Cramer JT, Coburn JW, Malek MH.Comparison of Fourier and wavelet transform procedures for examiningmechanomyographic and electromyographic frequency versus isokinetic torquerelationships. Electromyogr Clin Neurophysiol 2005;45:93–103.

Bilodeau M, Schindler-Ivens S, Williams DM, Chandran R, Sharma SS. EMG frequencycontent changes with increasing force and during fatigue in the quadricepsfemoris muscle of men and women. J Electromyogr Kinesiol 2003;13:83–92.

Bonato P, Gagliati G, Knaflitz M. Analysis of myoelectric signals recorded duringdynamic contractions. IEEE Eng Med Biol Mag 1996;15:102–11.

PRESSce Methods xxx (2008) xxx–xxx

Cramer JT, Housh TJ, Weir JP, Johnson GO, Ebersole KT, Perry SR, Bull AJ. Poweroutput, mechanomyographic, and electromyographic responses to maximal,concentric, isokinetic muscle actions in men and women. J Strength Cond Res2002;16:399–408.

Diemont B, Figini MM, Orizio C, Perini R, Veicsteinas A. Spectral analysis of muscularsound at low and high contraction level. Int J Biomed Comput 1988;23:161–75.

Ebersole KT, Housh TJ, Johnson GO, Perry SR, Bull AJ, Cramer JT. Mechanomyographicand electromyographic responses to unilateral isometric training. J StrengthCond Res 2002;16:192–201.

Hopkins WG. Measures of reliability in sports medicine and science. Sports Med2000;30:1–15.

Inbar GF, Allin J, Paiss O, Kranz H. Monitoring surface EMG spectral changes by thezero crossing rate. Med Biol Eng Comput 1986;24:10–8.

Inbar GF, Noujaim AE. On surface EMG spectral characterization and its applicationto diagnostic classification. IEEE Trans Biomed Eng 1984;31:597–604.

Karlsson JS, Gerdle B, Akay M. Analyzing surface myoelectric signals recorded duringisokinetic contractions. IEEE Eng Med Biol Mag 2001;20:97–105.

Kellis, E, Katis A. Reliability of EMG power-spectrum and amplitude of the semi-

ographic amplitude and mean power frequency during isometric stepeth.2008.02.017

Electromyogr Kinesiol; in press.Kwatny E, Thomas DH, Kwatny HG. An application of signal processing techniques

to the study of myoelectric signals. IEEE Trans Biomed Eng 1970;17:303–13.Paiss O, Inbar GF. Autoregressive modeling of surface EMG and its spectrum with

application to fatigue. IEEE Trans Biomed Eng 1987;34:761–70.Ryan ED, Beck TW, Herda TJ, Hartman MJ, Stout JR, Housh TJ, Cramer JT.

Mechanomyographic amplitude and mean power frequency responses duringisometric ramp vs. step muscle actions. J Neurosci Methods 2008a;168:293–305.

Ryan ED, Cramer JT, Egan AD, Hartman MJ, Herda TJ. Time and frequency domainresponses of the mechanomyogram and electromyogram during isometric rampcontractions: a comparison of the short-time Fourier and continuous wavelettransforms. J Electromyogr Kinesiol 2008b;18:54–67.

Ryan ED, Cramer JT, Housh TJ, Beck TW, Herda TJ, Hartman MJ, Stout JR. Inter-individual variability among the mechanomyographic and electromyographicamplitude and mean power frequency responses during isometric ramp muscleactions. Electromyogr Clin Neurophysiol 2007;47:161–73.

Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. PsycholBull 1979;36:420–8.

Smith DB, Housh TJ, Stout JR, Johnson GO, Evetovich TK, Ebersole KT. Mechanomyo-graphic responses to maximal eccentric isokinetic muscle actions. J Appl Physiol1997;82:1003–7.

Streiner DL, Norman GR. Measurement scales: a practical guide to their developmentand use. 2nd ed. Oxford: Oxford University Press; 1995. pp. 104–127.

Weir JP. Quantifying test–retest reliability using the intraclass correlation coefficientand the SEM. J Strength Cond Res 2005;19:231–40.

Yang JF, Winter DA. Electromyography reliability in maximal and submaximal iso-metric contractions. Arch Phys Med Rehabil 1983;64:417–20.