Relationship between Isometric Muscle Force and Fractal Dimension of Surface Electromyogram

Abstract

The relationship between fractal dimension of the surface electromyogram (sEMG) and the intensity of muscle contraction is still controversial in simulated and experimental conditions. To support the use of fractal analysis to investigate myoelectric fatigue, it is crucial to establish the interdependence between fractal dimension and muscle contraction intensity. We analyzed the behavior of fractal dimension, conduction velocity, mean frequency, and average rectified value in twenty-eight volunteers at nine levels of isometric force. sEMG was obtained using bidimensional arrays in the biceps brachii muscle. The values of fractal dimension and mean frequency increased with force unless a plateau was reached at 30% maximal voluntary contraction. Overall, our findings suggest that, above a certain level of force, the use of fractal dimension to evaluate the myoelectric manifestations of fatigue may be considered, regardless of muscle contraction intensity.

Full text

Research Article
Relationship between Isometric Muscle Force and
Fractal Dimension of Surface Electromyogram
Matteo Beretta-Piccoli ,1,2 Gennaro Boccia,3,4 Tessa Ponti,1Ron Clijsen,5
Marco Barbero ,1and Corrado Cescon1
1Rehabilitation Research Laboratory 2rLab, Department of Business Economics, Health and Social Care,
University of Applied Sciences and Arts of Southern Switzerland, Manno, Switzerland
2Criams-Sport Medicine Centre Voghera, University of Pavia, Pavia, Italy
3NeuroMuscular Function Research Group, School of Exercise and Sport Sciences, Department of Medical Sciences,
University of Turin, Turin, Italy
4CeRiSM (Research Centre for Sport, Mountain, and Health), Rovereto, Italy
5University of Business Economics, Health and Social Care, University of Applied Sciences and Arts of Southern Switzerland,
Landquart, Switzerland
Correspondence should be addressed to Matteo Beretta-Piccoli; matteo.berettapiccoli@supsi.ch
Received 14 November 2017; Revised 1 February 2018; Accepted 11 February 2018; Published 15 March 2018
Academic Editor: Laura Guidetti
Copyright ©  Matteo Beretta-Piccoli et al. is is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
e relationship between fractal dimension of the surface electromyogram (sEMG) and the intensity of muscle contraction is still
controversial in simulated and experimental conditions. To support the use of fractal analysis to investigate myoelectric fatigue, it
is crucial to establish the interdependence between fractal dimension and muscle contraction intensity. We analyzed the behavior
of fractal dimension, conduction velocity, mean frequency, and average rectied value in twenty-eight volunteers at nine levels of
isometric force. sEMG was obtained using bidimensional arrays in the biceps brachii muscle. e values of fractal dimension and
mean frequency increased with force unless a plateau was reached at % maximal voluntary contraction. Overall, our ndings
suggest that, above a certain level of force, the use of fractal dimension to evaluate the myoelectric manifestations of fatigue may be
considered, regardless of muscle contraction intensity.
1. Introduction
e relation between electromyography (EMG) and force has
been a controversial topic for more than four decades. e
surface EMG (sEMG)/force relationship strongly depends on
motor units (MUs) control by the central nervous system
(CNS) and by the peripheral features of muscle. e CNS
modulatestheforceexpressedbythemusclebycontrolling
two parameters: the recruitment of MUs and the ring
rate of active MUs []. ese two parameters are directly
connected with the generation of electrical activity inside the
muscleandalsoinuencethesEMGsignal[].Indeed,the
sEMG signal is a result of the interferential summation of
MU action potentials (MUAPs) detected by electrodes and
thus it is of interest to understand the role played by the
neural parameters in driving the sEMG-force relationship
[]. e shape of this relationship has been explored in
experimental and simulation studies, with conicting results
ranging from linearity to nonlinearity [–]. e shape of
this relationship might also depend on the muscle investi-
gated, muscle ber composition, and muscle ber size [,
].
Inconsistent results in the literature may also reect that
muscles are not necessarily uniformly activated at increased
loads in a specic action. For this reason, sEMG varies
spatially over the muscle belly [–]. Applying multichannel
array electrode systems in sEMG recordings has been demon-
strated to improve the extraction of reliable sEMG/force
relationship increasing the representability of the measured
sEMG signal [–].
Hindawi
BioMed Research International
Volume 2018, Article ID 5373846, 9 pages
https://doi.org/10.1155/2018/5373846

BioMed Research International
Great interest has been given in the literature to the
nonlinear feature of the sEMG signal, such as recurrent
quantication analysis, percentage of determinism, sample
entropy, normalized mutual information, and fractal dimen-
sion (FD) [–]. Nonlinear analysis oers a powerful
approach for the investigation of physiological time series
because it provides a measure of the signal complexity.
In particular, the FD of the signal is a measure of self-
similarity over multiple time scales. Several studies [–]
have applied box-counting methods to estimate the FD of
the sEMG signal and a recent investigation showed a good
reliability of FD during isometric contractions in the biceps
brachii muscle [].
Nonlinear feature of the sEMG has been widely applied
to monitor the myoelectric manifestations of fatigue during
the course of isometric contractions []. Indeed, during
sustained submaximal contractions, the alterations in the
activity of muscles undergoing fatigue can be quantied,
using linear or nonlinear methods, prior to task failure [].
Mesin and colleagues [] computed a combination of both
linear and nonlinear analysis to synthetic and experimental
sEMG signals. ey found that FD was the most related
tothelevelofsynchronizationandtheleastrelatedtothe
changes of muscle ber conduction velocity (CV). Conse-
quently, they proposed the combination of FD and CV as
bidimensional index providing information about the central
and peripheral adjustments occurring during fatigue []. In
a more recent simulation study, Mesin and colleagues []
found that beyond synchronization level, the FD of the EMG
signals increased with the average ring rate of the active
MUs. For this reason, recently, the combined monitoring of
muscle ber conduction velocity (CV) and FD parameters
during continuous contractions was applied in the evaluation
of myoelectric manifestations of fatigue [–]. However,
tofullyunderstandtheapplicabilityofFDanalysisinthe
study of myoelectric manifestations of fatigue, it is crucial to
determineiftheFDisalsoaectedbythelevelofforceexerted
by muscles.
Some studies found that the FD of sEMG was linearly
but weakly related to the contraction level (% of maximal
voluntary contraction, MVC) in simulated and experimen-
tal conditions [, , ]. However, recent investigations
showed that FD is not related to the intensity of muscle
contraction [, ]; therefore, the relationship between force
and the FD of sEMG is still controversial. us, the aim of this
study was to evaluate the relationship between force and FD
of sEMG during isometric contractions of the biceps brachii.
2. Materials and Methods
2.1. Participants. e study was approved by the local ethics
committee of the Swiss Italian Health and Sociality Depart-
ment, Switzerland. All procedures were conducted accord-
ing to the Declaration of Helsinki. All participants signed
a written informed consent form before participation in
the experiments. Twenty-eight healthy recreationally active
volunteers ( women and  men) aged between  and 
years (25 ± 4 yrs) from a university setting were recruited to
participate in the study.
F : Electrode array position on biceps brachii muscle.
2.2. Experimental Procedure. e subjects participated in
three experimental sessions (“trials –”): the rst two trials
were conducted within the same day, with four minutes of
rest in between, without repositioning the electrodes. e
third trial was performed a week apart under the same
environmental conditions.
Subjects were seated in a height-adjustable chair with
their arm positioned on an isometric ergometer (MUC, OT
Bioelettronica, Turin, Italy), equipped with a load cell (Model
TF, CCT Transducers, Turin, Italy). In order to isolate
the action of the biceps brachii, the wrist was fastened to the
ergometer, with the elbow at ∘,asshowninFigure.
Initially, two isometric MVCs were performed, separated
by a -minute rest. During each contraction of the trial,
the force trace was displayed to participants on a computer
monitor as visual feedback. Participants were instructed to
increasetheforceuptotheirmaximumandtoholditfor-
 s. Participants were given strong verbal encouragement.
Next, aer a -minute rest, the subjects performed a
sequence of nine short contractions, from  to % of their
MVC in steps of % MVC in randomized order, lasting
 s, with  s of rest in between. Aer each contraction,
thesubjectswereaskedtoprovideavalueoftheperceived
exertion on a visual Borg scale, ranging from  to  [].
In the rst day of measurement, aer the rst session (trial
), a second sequence of contraction, constituting trial , was
performed.
2.3.EMGandForceMeasurements. Myoelectric signals were
detected from the biceps brachii, in a monopolar congu-
ration using a bidimensional array of  electrodes ( mm
diameter,  × grid, and  mm interelectrode distance;
model ELSCHNM; OT Bioelettronica) (Figure ). is

BioMed Research International 
muscle was chosen in order to obtain high-quality sEMG
signals according to the qualitative criteria described in [].
eelectrodegridwasappliedonthemusclebelly,withits
distal edge close to the cubital fossa and the midline of the
array aligned with the midline of biceps along a line from
the cubital fossa to the acromion (see Figure ). A ground
electrode was placed on the contralateral wrist. e EMG sig-
nals were amplied (EMG-USB; OT Bioelettronica), band-
pass ltered (– Hz), sampled at  Hz, and stored on
acomputer.
e isometric ergometer was used to measure elbow
torque with a torque meter operating linearly in the range
– Nm. e torque signal was amplied (MISO II; OT
Bioelettronica) and stored on a computer with the sEMG
data. e torque signal was displayed on a screen, providing
real-time biofeedback.
2.4. Signal Processing. e number of channels used for
CV estimation was selected based on visual inspection of
single dierential signals, between the distal tendon and
the innervation zone, along one of the array columns, as
previously described [].
e number of channels chosen to estimate CV was
between  and , according to a previously published study
[]. CV values outside the physiological range (–. m/s)
were excluded from the analysis [].
For each signal, a  s lapse was identied, where the
force level was stable within the % boundaries of the
target force requested to the subjects. Signals were then
divided into epochs of  s and CV was computed using a
multichannel algorithm [] on the selected channels. e
three obtained values were then averaged. Next, each of the
threeepochsofeachsignalwasusedfortheestimationof
average rectied value (ARV), mean frequency of the power
spectrum (MNF), and FD. Estimates obtained from single
channels were averaged over the channels previously selected
by visual analysis and over the three signal epochs. erefore,
for each contraction level one value for ARV, MNF, and FD
was obtained.
In addition, ARV, MNF, CV, and FD data, as well as Borg
scale values, were normalized for each subject according to
their values at % MVC and expressed as percentages. e
force level of % was selected aer the completion of data
collection, since many of the subjects could not perform 
and % MVC contraction. e % value was the maximum
force level which all the subjects could reach.
FD was estimated using the box-counting method, as
previously reported []. Data were analyzed by custom-
written soware in MATLAB Rb (Mathworks, Natick,
USA)
2.5. Statistical Analysis. Intra- and intersession reliability
were examined using the Intraclass Correlation Coecient
(ICC(2,1)) on averaged measures [], since its use has been
recommended in reliability studies [, ]. e criteria used
for the interpretation of the ICCs were as follows: .–.:
no correlation; .–.: low correlation; .–.: mod-
erate correlation; .–.: high correlation; .–.: very
high correlation [].
To test the relationship between EMG variables and
force, only the rst session, that is, trial , was considered.
A Shapiro–Wilk test revealed that all the estimated EMG
variables were not normally distributed across subjects and,
thus, the nonparametric Kruskal–Wallis test was performed
on the sEMG variables for each contraction at dierence force
levels. Considered factors were trial and force level. When
the Kruskal–Wallis test indicated signicant variations, a
post hoc Dunn–Bonferroni test [] was applied on pairwise
comparisons; statistical signicance was accepted at the 𝑝<
0.001 level.
e epsilon-squared estimate of eect size was calculated
using []
𝐸2
𝑅=𝐻
𝑛−1,()
where 𝐻is the value obtained in the Kruskal–Wallis test (the
K-W H-test statistics) and 𝑛the total number of observations.
e 𝐸2
𝑅coecient assumes values between  (indicating no
relationship) and  (perfect relationship).
Statistical analyses were performed using SPSS version
. (SPSS Inc., Chicago, IL, USA), and signicance was set
to 𝛼 = 0.05. Results are reported as median and interquartile
range.
3. Results
3.1. Reliability Analysis. Tab l e  d o c u m e n t s t h e r e s u l t s o f
ICC(2,1) analysis for the initial values of CV, FD, MNF, and
ARV during the short isometric contractions, with force lev-
els between  and % MVC. According to the classication
of [], high to very high levels of intrasession reliability were
identied for all the parameters (ICC between . and .),
whereas the intersession reliability was considerably lower.
e most reliable parameter across experimental sessions
was indeed ARV, followed by FD and MNF. Initial values of
CV showed higher ICC values at lower contraction levels,
whereas at force levels between % and % MVC, CV
displayed a very low intersubject variability, demonstrating
dependence on days and trials larger than dependence on
subjects [, ].
3.2. Relation with Force. Kruskal–Wallis test did not reveal
any statistical dependence of the variables on trials. Distri-
butions of FD, ARV, MNF, and Borg ratings were similar
for all contraction levels, as assessed by visual inspection
of boxplots (Figure ). Median scores of these parameters
were statistically dierent across the nine levels of force (𝑝<
0.0001). Only the increasing trend of CV versus force was not
statistically signicant; for this reason no post hoc analysis
was performed for CV.
To allow better visualization of the parameters trend, a
boxplot for each normalized parameter with respect to their
values at % MVC was added to Figure. Eect size analysis,
that is, the percentage of the variability of the considered
parameters which is really accounted for by the level of
force, revealed very high scores for ARV and Borg values
(epsilon-squared estimates, resp., % and %), whereas

BioMed Research International
1.650
FD
1.550
1.575
1.600
1.625
102
nFD (%)
96
98
100
Initial values
Normalized values
% MVC
10 20 30 40 50 60 70 80 90
% MVC
10 20 30 40 50 60 70 80 90
∗
∗
(a)
CV (m/s)
4.0
5.0
6.0
7.0
nCV (%)
70
80
90
100
110
Initial values Normalized values
% MVC
10 20 30 40 50 60 70 80 90
% MVC
10 20 30 40 50 60 70 80 90
(b)
0
50
100
150
200
250
300
350
ARV (microV)
0
50
100
150
200
nARV (%)
Initial values
Normalized values
% MVC
10 20 30 40 50 60 70 80 90
% MVC
10 20 30 40 50 60 70 80 90
∗
∗
∗
∗
∗
(c)
F : Continued.

BioMed Research International 
120
MNF (Hz)
60
70
80
90
100
110
Initial values
Normalized values
% MVC
10 20 30 40 50 60 70 80 90
% MVC
10 20 30 40 50 60 70 80 90
70
80
90
100
110
nMNF (%)
∗
∗
(d)
F : Box-and-whisker plots of initial and normalized values (with respect to their values at % MVC) of fractal dimension (FD)
conduction velocity (CV), average rectied value (ARV), and mean frequency (MNF) during short isometric –% maximal voluntary
contractions (MVCs) of the biceps brachii. Statistically signicant results of the Dunn–Bonferroni post hoc test are indicated (∗𝑝 < 0.001).
T : Results of the reliability analysis of initial values of CV, FD, MNF, and ARV at  to % MVC. Intra- and intersession ICC scores
are reported.
MVC ICC ICC ICC ICC
Intra Inter Intra Inter Intra Inter Intra Inter
CV FD MNF ARV
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
% . . . . . . . .
Note. MVC: maximal voluntary contraction; ICC: intraclass correlation coecient intra- and intersession.
smaller eect size was found for FD and MNF (epsilon-
squared estimates, resp., % and %). e post hoc analysis
revealed statistically signicant dierences in the considered
parameters obtained at low force levels (resp., –% MVC
for ARV and Borg ratings and –% MVC for FD and
MNF) with respect to high force levels (–% MVC)
(Figures  and ).
4. Discussion
4.1. Intra- and Intersession Reliability. FD, MNF, and ARV
showed high intra- and intersession reliability, in accordance
with previously published studies [, , –]; the inters-
ession reliability of CV at contraction levels higher than %
MVC was very low. is result might be explained by the fact
that the variability of CV between subjects decreases as the
level of contraction increases over % MVC [].
4.2. Relation between EMG Parameters, Borg Ratings, and
Force. Inthepresentstudy,FDandMNFwerethevariables
least inuenced by the level of exerted force (Figures (a)
and (d)). In fact, both variables showed a trend, increasing
from%to%MVC,butthereaerreachingaplateau
beyond % of MVC (conrmed by the results of the post hoc
analysis, as well). e little or even independence of FD and
MNFonthelevelofmuscleforcewasreportedalsointwo
previous investigations in other muscles and with dierent
methods [, ]. In particular, in [] the upper trapezius
muscle was investigated, which was compared to the biceps
brachii, and presents a much more complex architecture and
an heterogeneous distribution of the muscle activity [].
As already reported in literature, FD is sensitive to
thepresenceoflargeactiveMUAPsthatusuallyappear
in the signal due to synchronization at high force levels,
during fatiguing contractions []. Nevertheless, a similar

BioMed Research International
Borg value
6
8
10
12
14
16
18
20
Initial values
Normalized values
% MVC
10 20 30 40 50 60 70 80 90
% MVC
10 20 30 40 50 60 70 80 90
25
45
65
85
105
125
145
nborg value (%)
∗
∗
∗
∗
∗
F : Box-and-whisker plots of the initial and normalized values (with respect to their values at % MVC) of Borg ratings during short
isometric –% maximal voluntary contractions (MVCs) of the biceps brachii. Statistically signicant results of the Dunn–Bonferroni post
hoc test are indicated (∗𝑝 < 0.001).
phenomenon happens also at low force levels, whenever
larger MUs, with low ring frequency, are recruited according
to the Henneman’s size principle. Moreover, in simulated
EMG signals, FD was positively correlated to the ring rate
oftheactiveMUsandnegativelycorrelatedtothelevelof
MU synchronization []. Since the level of synchronization
is not expected to change in nonfatiguing contractions, it was
reasonabletohypothesizethatFDcouldsomehowincrease
with increasing force levels. us, it is possible to speculate
thatFDmightbeareliableindicatorofMUsynchronization,
less dependent from the ring rate.
Muscle ber CV seems to be the most aordable vari-
able for relating EMG signals modications and MUs pool
recruitment []. Since CV increases gradually when larger
MUs are recruited [], it was expected to increase with
contraction intensity []. Contrary to the expectation, the
average CV did not increase signicantly with increasing
forcelevelsalthoughwecouldobserveatrendinthat
direction in our dataset (see Figure (b)). ere are two main
confounding factors that could have aected CV estimates:
(1) the subcutaneous tissue and (2) the alignment of the
electrode grid along the direction of muscle bers. Indeed,
a high thickness of subcutaneous tissue and malalignment
of electrode grids might both produce an overestimation of
CV and consequently aect the trend of CV across force
levels.SincetheCVvalueswererelativelyhigh(>. m/s)
even at the lowest force levels (i.e., % of MVC), this
explanation seems to be plausible. Anyway, an overestima-
tionofCV,ifpresent,wouldbevisibleatallcontraction
levels; thus normalized values would not be aected by this
bias.
e amplitude of the EMG signal (ARV) was the variable
most dependent on the level of force exerted (Figure (c)).
is was an expected result, since many previous studies
demonstrated a direct relationship between EMG and force
[–]. In particular, ARV values obtained at the highest force,
that is, the % of the MVC, were greater than those lower or
equaltothe%ofMVC.Whereas,between%and%of
MVC, no increase in ARV was found. us, EMG amplitude
seemed to be sensitive to the increase of force only from low
(% of MVC) to medium (% of MVC) force levels, but not
from % to %. Even this was an expected result because
Troiano and colleagues previously reported the same pattern
[]. e recruitment of motor units and the ring rate of
active motor units progressively increase at increasing force
exertion [], and this leads to increasing electrical activity
inside the muscle []. Consequently, increasing amplitude
of EMG signal would be expected throughout the whole
range of forces. However, our results showed that the EMG
amplitude was not consistently aected by the increase in
forceaer%ofMVC.iscanbeexplainedbythefactthat
the amplitude cancellation inuenced the measures of EMG
amplitudemostlyathighforcelevels.Indeed,theamplitude
cancellation has been proven to increase with increasing
number of active motor units [].
Finally, the present study found a relation between ratings
of perceived exertion (Borg ratings) and force levels (Fig-
ure)in-linewithpreviouspublishedstudies,wherealinear
relationship, during isometric contractions, was found [,
, ]. Interestingly, as occurred with ARV, no statistically
signicant increase in perceived exertion was found between
% and % of MVC. Together, these results furthermore
support previous ndings indicating the relationship between
muscle activation and perceived exertion [].
e limitations of this study are mainly related to tech-
nical constraints. Firstly, we investigated only one muscle,

BioMed Research International 
which, of course, does not represent the behavior of all the
muscles. Secondly, to our knowledge, literature is currently
lacking studies on validity of FD in estimating MU synchro-
nization. If future studies will overcome this gap, FD will
provideavalidandrobustmeasureofMUsynchronization
during fatiguing contractions.
5. Conclusions
e present study showed that FD is a reliable EMG param-
eter at all contraction levels and has little dependency from
muscle force, in the biceps brachii muscle above % MVC.
In such conditions, FD can be applied in experimental
studies focusing on fatigue or on motor unit synchronization,
independently from the force exerted.
Conflicts of Interest
e authors declare that there are no conicts of interest.
Acknowledgments
e authors thank Emiliano Soldini (LABStat, University
of Applied Sciences and Arts of Southern Switzerland) for
his contributions to the statistical analysis. e study was
approved by the local ethics committee of the Swiss Italian
Health and Sociality Department, Bellinzona, Switzerland
(no. CE-). is study was supported by the im van
der Laan Foundation, Switzerland (to Matteo Beretta-Piccoli,
Corrado Cescon, Ron Clijsen, and Marco Barbero).
References
[] Z.Erim,C.J.DeLuca,K.Mineo,andT.Aoki,“Rank-ordered
regulation of motor units,” Muscle & Nerve,vol.,no.,pp.
–, .
[]A.J.Fuglevand,D.A.Winter,andA.E.Patla,“Modelsof
recruitment and rate coding organization in motor unit pools,”
Journal of Neurophysiolog y, vol. , no. , pp. –, .
[] K. G. Keenan and F. J. Valero-Cuevas, “Experimentally valid
predictions of muscle force and EMG in models of motor-unit
function are most sensitive to neural properties,” Journal of
Neurophysiology, vol. , no. , pp. –, .
[] H. S. Milner-Brown and R. B. Stein, “e relation between the
surface electromyogram and muscular force.,” e Journal of
Physiology,vol.,no.,pp.–,.
[] T. Moritani and H. A. DeVries, “Reexamination of the relation-
ship between the surface integrated electromyogram (IEMG)
and force of isometric contraction,” American Journal of Physi-
cal Medicine & Rehabilitation,vol.,no.,pp.–,.
[]J.V.Basmajian,Muscles Alive, eir Functions Revealed by
Electromyography, Williams & Wilkins, Baltimore, Md, USA,
th edition, .
[] B. Bigland-Ritchie, “EMG/force relations and fatigue of human
voluntary contractions,” Exercise and Sport Sciences Reviews,
vol.,no.,pp.–,.
[] J. H. Lawrence and C. J. De Luca, “Myoelectric signal versus
force relationship in dierent human muscles,” Journal of
Applied Physiology,vol.,no.,pp.–,.
[] P. Zhou and W. Z. Rymer, “Factors governing the form of
the relation between muscle force and the EMG: A simulation
study,” Journal of Neurophysiology,vol.,no.,pp.–,
.
[]A.Botter,H.R.Marateb,B.Afsharipour,andR.Merletti,
“Solving EMG-force relationship using Particle Swarm Opti-
mization,” in Proceedings of the 2011 33rd Annual International
Conference of the IEEE Engineering in Medicine and Biology
Society, pp. –, Boston, MA, USA, August .
[] M. Al Harrach, V. C arriou,S. B oudaoud, J. Laforet,and F. Marin,
“Analysis of the sEMG/force relationship using HD-sEMG
technique and data fusion: A simulation study,” Computers in
Biology and Medicine,vol.,pp.–,.
[] B. A. Alkner, P. A. Tesch, and H. E. Berg, “Quadriceps
EMG/force relationship in knee extension and leg press,”
Medicine & Science in Sports & Exercise,vol.,no.,pp.–
, .
[] D. Staudenmann, J. H. van Die¨
en,D.F.Stegeman,andR.M.
Enoka, “Increase in heterogeneity of biceps brachii activation
during isometric submaximal fatiguing contractions: A multi-
channel surface EMG study,” Journal of Neurophysiology, vol. ,
no.,pp.–,.
[] D. Staudenmann, I. Kingma, A. Daertshofer, D. F. Stegeman,
andJ.H.vanDie
¨
en, “Heterogeneity of muscle activation
in relation to force direction: A multi-channel surface elec-
tromyography study on the triceps surae muscle,” Journal of
Electromyography & Kinesiology,vol.,no.,pp.–,
.
[] A. Holtermann, K. Roeleveld, and J. S. Karlsson, “Inhomo-
geneities in muscle activation reveal motor unit recruitment,”
Journal of Electromyography & Kinesiology,vol.,no.,pp.–
, .
[] T. Rantalainen, A. Kłodowski, and H. Piitulainen, “Eect of
innervation zones in estimating biceps brachii force-EMG
relationship during isometric contraction,” Journal of Elec-
tromyography & Kinesiology,vol.,no.,pp.–,.
[] D. Staudenmann, I. Kingma, A. Daertshofer, D. F. Stegeman,
and J. H. Van Die¨
en, “Improving EMG-based muscle force
estimation by using a high-density EMG grid and principal
component analysis,” IEEE Transactions on Biomedical Engi-
neering,vol.,no.,pp.–,.
[] D. Staudenmann, I. Kingma, D. F. Stegeman, and J. H. Van
Die¨
en, “Towards optimal multi-channel EMG electrode cong-
urations in muscle force estimation: A high density EMG study,”
J
ournalofElectromyography&Kinesiology,vol.,no.,pp.–,
.
[] J.-P. Eckmann, S. Olison Kamphorst, and D. Ruelle, “Recur-
rence plots of dynamical systems,” EPL (Europhysics Letters),
vol. , no. , pp. –, .
[] F. Felici, A. Rosponi, P. Sbriccoli, G. C. Filligoi, L. Fattorini,
and M. Marchetti, “Linear and non-linear analysis of surface
electromyograms in weightliers,” European Journal of Applied
Physiology,vol.,no.,pp.–,.
[] J. S. Richman and J. R. Moorman, “Physiological time-series
analysis using approximate entropy and sample entropy,” Amer-
ican Journal of Physiology-Heart and Circulatory Physiology,vol.
, no. , pp. H–H, .
[] A. Bingham, S. P. Arjunan, B. Jelfs, and D. K. Kumar,
“Normalised mutual information of high-density surface elec-
tromyography during muscle fatigue,” Entropy,vol.,p.,
.

BioMed Research International
[] J. A. Gitter and M. J. Czerniecki, “Fractal analysis of the elec-
tromyographic interference pattern,” Journal of Neuroscience
Methods,vol.,no.-,pp.–,.
[] Z. Xu and S. Xiao, “Fractal dimension of surface EMG and its
determinants,” in Proceedings of the 19th Annual International
Conference of the IEEE Engineering in Medicine and Biology
Society,vol.,pp.–,.
[] C.J.Anmuth,G.Goldberg,andN.H.Mayer,“Fractaldimen-
sion of electromyographic signals recorded with surface elec-
trodes during isometric contractions is linearly correlated with
muscle activation,” Muscle & Nerve,vol.,no.,pp.-,
.
[] V. Gupta, S. Suryanarayanan, and N. P. Reddy, “Fractal analysis
of surface EMG signals from the biceps,” International Journal
of Medical Informatics, vol. , no. , pp. –, .
[] R. Shields, “P. Fractal dimension of the EMG interference
pattern: preliminary observations and comparisons with other
measures of interference pattern analysis,” Clinical Neurophysi-
ology,vol.,p.,.
[] M. Beretta-Piccoli, G. D’Antona, C. Zampella, M. Barbero, R.
Clijsen, and C. Cescon, “Test-retest reliability of muscle ber
conduction velocity and fractal dimension of surface EMG
during isometric contractions,” Physiological Measurement,vol.
,no.,pp.–,.
[] G. Marco, B. Alberto, and T. M. Vieira, “Surface EMG and
muscle fatigue: Multi-channel approaches to the study of
myoelectric manifestations of muscle fatigue,” Physiological
Measurement,vol.,no.,pp.R–R,.
[] L. Mesin, C. Cescon, M. Gazzoni, R. Merletti, and A. Rain-
oldi, “A bi-dimensional index for the selective assessment of
myoelectric manifestations of peripheral and central muscle
fatigue,” Journal of Electromyography & Kinesiology,vol.,no.
, pp. –, .
[] L. Mesin, D. Dardanello, A. Rainoldi, and G. Boccia, “Motor
unit ring rates and synchronisation aect the fractal dimen-
sion of simulated surface electromyogram during isomet-
ric/isotonic contraction of vastus lateralis muscle,” Medical
Engineering & Physics, vol. , no. , pp. –, .
[] G. Boccia, D. Dardanello, M. Beretta-Piccoli et al., “Muscle
ber conduction velocity and fractal dimension of EMG during
fatiguing contraction of young and elderly active men,” Physi-
ological Measurement,vol.,no.,articleno.,pp.–,
.
[] M. Beretta-Piccoli, G. D’Antona, M. Barbero et al., “Evaluation
of central and peripheral fatigue in the quadriceps using fractal
dimension and conduction velocity in young females,” PLoS
ONE,vol.,no.,ArticleIDe,.
[] F. Meduri, M. Beretta-Piccoli, L. Calanni et al., “Inter-Gender
sEMG evaluation of central and peripheral fatigue in biceps
brachii of young healthy subjects,” PLoS ONE, vol. , no. ,
Article ID e, .
[] S. Poosapadi Arjunan and D. K. Kumar, “Computation of fractal
features based on the fractal analysis of surface Electromyogram
to estimate force of contraction of dierent muscles,” Computer
Methods in Biomechanics and Biomedical Engineering,vol.,no.
, pp. –, .
[] A. Troiano, F. Naddeo, E. Sosso, G. Camarota, R. Merletti,
and L. Mesin, “Assessment of force and fatigue in isometric
contractions of the upper trapezius muscle by surface EMG
signal and perceived exertion scale,” Gait & Posture,vol.,no.
, pp. –, .
[] G. A. Borg, “Psychophysical bases of perceived exertion,”
Medicine & Science in Sports & Exercise,vol.,no.,pp.–
, .
[] M. Beretta Piccoli, A. Rainoldi, C. Heitz et al., “Innervation zone
locations in  supercial muscles: Toward a standardization of
electrode positioning,” Muscle & Nerve,vol.,no.,pp.–
, .
[] D. Farina, D. Zagari, M. Gazzoni, and R. Merletti, “Repro-
ducibility of muscle-ber conduction velocity estimates using
multichannel surface EMG techniques,” Muscle & Ner ve,vol.,
no. , pp. –, .
[] S. Andreassen and L. Arendt-Nielsen, “Muscle bre conduction
velocity in motor units of the human anterior tibial muscle: a
new size principle parameter.,” e Journal of Physiology,vol.
, no. , pp. –, .
[] D. Farina and R. Merletti, “A Novel Approach for Estimating
Muscle Fiber Conduction Velocity, by Spatial and Tempo-
ral Filtering of Surface EMG Signals,” IEEE Transactions on
Biomedical Engineering,vol.,no.,pp.–,.
[] J. P. Weir, “Quantifying test-retest reliability using the intraclass
correlation coecient and the SEM,” eJournalofStrengthand
Conditioning Research,vol.,no.,pp.–,.
[] G. Rankin and M. Stokes, “Reliability of assessment tools in
rehabilitation: an illustration of appropriate statistical analyses,”
Clinical Rehabilitation,vol.,no.,pp.–,.
[] A.Bruton,J.H.Conway,andS.T.Holgate,“Reliability:what
is it, and how is it measured?” Physiotherapy,vol.,no.,pp.
–, .
[] B. H. Munro, StatisticalMethodsforHealthCareResearch,
Williams & Wilkins, Philadelphia, Pa, USA, th edition, .
[] O. J. Dunn, “Multiple comparisons using rank sums,” Te c hno-
metrics,vol.,no.,pp.–,.
[] M. Tomczak and E. Tomczak, “e need to report eect
size estimates revisited. An over view of some recommended
measures of eect size,” Trends in Sport Science,vol.,pp.–
, .
[] A. Rainoldi, J. E. Bullock-Saxton, F. Cavarretta, and N. Hogan,
“Repeatability of maximal voluntary force and of surface EMG
variables during voluntary isometric contraction of quadriceps
muscles in healthy subjects,” Journal of Electromyography &
Kinesiolog y,vol.,no.,pp.–,.
[]F.A.Arnall,G.A.Koumantakis,J.A.Oldham,andR.G.
Cooper, “Between-days reliability of electromyographic mea-
suresofparaspinalmusclefatigueat,and%levels
of maximal voluntary contractile force,” Clinical Rehabilitation,
vol.,no.,pp.–,.
[] D. Falla, P. Dall’Alba, A. Rainoldi, R. Merletti, and G. Jull,
“Repeatability of surface EMG variables in the sternocleidomas-
toid and anterior scalene muscles,” European Journal of Applied
Physiology,vol.,no.,pp.–,.
[] J. Lee, M.-Y. Jung, and S.-H. Kim, “Reliability of spike and turn
variables of surface EMG during isometric voluntary contrac-
tionsofthebicepsbrachiimuscle,”Journal of Electromyography
&Kinesiology, vol. , no. , pp. –, .
[]A.Rainoldi,G.Galardi,L.Maderna,G.Comi,L.LoConte,
and R. Merletti, “Repeatability of surface EMG variables during
voluntary isometric contractions of the biceps brachii muscle,”
Journal of Electromyography & Kinesiology,vol.,no.,pp.–
, .
[] M. Bilodeau, S. Schindler-Ivens, D. M. Williams, R. Chandran,
andS.S.Sharma,“EMGfrequencycontentchangeswith

BioMed Research International 
increasing force and during fatigue in the quadriceps femoris
muscle of men and women,” Journal of Electromyography &
Kinesiolog y,vol.,no.,pp.–,.
[] A. Gallina, R. Merletti, and M. Gazzoni, “Uneven spatial
distribution of surface EMG: What does it mean?” European
Journal of Applied Physiology,vol.,no.,pp.–,.
[]D.Farina,R.Merletti,andR.M.Enoka,“eextractionof
neural strategies from the surface EMG,” Journal of Applied
Physiology,vol.,no.,pp.–,.
[] P. J. Blijham, H. J. Ter Laak, H. J. Schelhaas, B. G. M. Van
Engelen, D. F. Stegeman, and M. J. Zwarts, “Relation between
muscle ber conduction velocity and ber size in neuromuscu-
lar disorders,” Journal of Applied Physiology,vol.,no.,pp.
–, .
[] K. G. Keenan, D. Farina, K. S. Maluf, R. Merletti, and R. M.
Enoka, “Inuence of amplitude cancellation on the simulated
surface electromyogram,” Journal of Applied Physiology, vol. ,
no. , pp. –, .
[] J. C. Stevens and W. S. Cain, “Eort in isometric muscular
contractions related to force level and duration,” Perception &
Psychophysics, vol. , no. , pp. –, .
[] M. K. Timmons, S. M. Stevens, and D. M. Pincivero, “e eect
of arm abduction angle and contraction intensity on perceived
exertion,” EuropeanJournalofAppliedPhysiology,vol.,no.
,pp.–,.
[] KM. Lagally, RJ. Robertson, KI. Gallagher, FL. Goss, JM. Jakicic,
SM. Lephart et al., “Perceived exertion, electromyography,
and blood lactate during acute bouts of resistance exercise,”
Medicine & Science in Sports & Exercise,vol.,no.,pp.–
, .