Journal of Neuroscience Methods 196 (2011) 238–246
Contents lists available at ScienceDirect
Journal of Neuroscience Methods
journa l homepage: www.e lsev ier .com
Reliabi ge
and coh
Annie Sc
a Plasticity and 13402
b Department o anada
a r t i c l
Article history:
Received 15 Se
Received in re
Accepted 5 Jan
Keywords:
Motor unit
Synchrony
Coherence
EMG determinism
Human
r used
ctrom
e hav
ed wi
ET an
the d
yses w
nt of
in the time and frequency domains, respectively. Recurrent quantification analysis was applied to EMG
activity to extract its DET. Overall, changes in EMG DET were poorly explained by changes in motor
unit synchronous impulse probability (6%) and frequency (5%), and by changes in motor unit coherence
in the 6–12 Hz (5%) and 25–40 Hz (8%) bands. Moreover, the comparison of the data obtained at the
weakest and the strongest contraction levels tested with each motor unit pair showed that EMG DET
1. Introdu
With str
active and
degree of c
influence th
Randall, 19
Christakos
charges inv
slow co-var
time scale o
nizing proc
motoneuro
time scale o
Motoneuro
that innerva
pendent inp
and Sears, 1
synchronou
∗ Correspon
E-mail add
0165-0270/$ –
doi:10.1016/j.remained unaltered with stronger contraction despite the occurrence of consistent changes in motor
unit synchrony in both time and frequency domains. This speaks strongly against the reliability of DET
in evaluating changes in motor unit synchronization during submaximal muscle contraction.
© 2011 Elsevier B.V. All rights reserved.
ction
onger muscle contraction more motoneurones become
those which are already active fire at faster rates. The
oupling between motoneurone discharges is liable to
e strength and fluctuation of the force output (Elble and
76; Halliday et al., 1999; McAuley and Marsden, 2000;
et al., 2006). Coordination between motoneurone dis-
olves the so-called common drive which underlies the
iation of the motoneurone firing frequencies within a
f 1–2 s (De Luca et al., 1982), as well as the synchro-
esses which are responsible for the coupling between
ne discharges above chance level within a much shorter
f few ms to few tens of ms (Kirkwood and Sears, 1991).
ne synchronous activity is attributed to common inputs
te conjointly a group of motoneurones and/or to inde-
uts that are synchronized pre-synaptically (Kirkwood
991). In the case of uncorrelated common input firing,
s firings occur randomly with no specific rhythm. In
ding author. Tel.: +1 819 3765011.
ress: martin.descarreaux@uqtr.ca (M. Descarreaux).
the case of common inputs synchronized pre-synaptically and/or
independent inputs driven by oscillatory sources, synchronous fir-
ings occur rhythmically in a correlated way. Coupling between
discharges of human motor units is explored in the time domain
by computing cross-correlation histograms in which firings occur-
ring in synchrony periodically or not are grouped into central peaks
(Farmer et al., 1997). The various types of periodicity generated by
co-modulating processes and/or by oscillatory sources of synchro-
nizing inputs are detected in the frequency domain by coherence
analysis of motor unit discharges (Farmer et al., 1993).
In view of the challenge of reliably discriminating single motor
unit action potentials as more motoneurones become active,
extracting an index of motor unit synchronous activity from sur-
face EMG signals might be a good alternative to investigate the
alteration of motor unit synchrony with stronger contraction. The
non-linear method of recurrence quantitative analysis (RQA) has
been recently introduced to quantify the amount of recurrence pat-
terns in EMG activity using an index named determinism (Webber
and Zbilut, 1994; Farina et al., 2002; Fattorini et al., 2005).
Following this track, by combining RQA applied to EMG activity
and cross-correlation analysis of single motor unit discharges, Del
Santo et al. (2006, 2007) recently reported the existence of a tight
co-variation between EMG DET and an increase in motor unit syn-
see front matter © 2011 Elsevier B.V. All rights reserved.
jneumeth.2011.01.005lity of EMG determinism to detect chan
erence during submaximal contraction
hmieda, Martin Descarreauxb,∗
Pathophysiology of Movement, UMR 6196 CNRS 31, Chemin Joseph Aiguier, Marseilles
f Chiropractic, Université du Québec à Trois-Rivières, C.P. 500, Trois-Rivières, Québec, C
e i n f o
ptember 2010
vised form 9 December 2010
uary 2011
a b s t r a c t
The determinism (DET) is a paramete
patterns in a signal. Applied to the ele
of motor unit synchrony in human. W
firings above chance level was enhanc
we aimed at determining if (1) EMG D
DET was more specifically related to
Cross-correlation and coherence anal
various force levels to assess the amou/ locate / jneumeth
s in motor unit synchrony
Cedex 20, France
G9A 5H7
in nonlinear analysis to quantify the occurrence of recurrent
yographic activity (EMG), DET has been proposed as an index
e recently shown that the amount of motor unit synchronous
th stronger submaximal muscle contraction. Using these data,
d motor unit synchrony varied in the same way and (2) EMG
egree of oscillatory coupling between motor unit discharges.
ere applied to the discharges of 30 motor unit pairs tested at
synchronous impulses and the strength of oscillatory coupling

A. Schmied, M. Descarreaux / Journal of Neuroscience Methods 196 (2011) 238–246 239
chronous activity induced pharmacologically. These results were
taken as a major argument in favour of the use of RQA to evalu-
ate changes in motor unit synchronization. The reliability of RQA
as a tool for assessing single motor unit synchronization has been
recently cha
values were
in motor un
the present
Upon tes
motor unit
we have re
chronous fi
increases in
Using this d
tested by D
mine if EM
in motor u
ses with str
2010).
The occ
likely to be
ingly, chan
fatigue, stre
as reflectin
motor unit
et al., 2001
objective w
between EM
of motor un
various leve
2. Materia
The exp
mittee of U
Experiment
aged 23–33
form prior t
laration of H
2.1. Experim
The prot
and Descar
their wrist
against a fo
wick, RI, US
screen facin
trace on a ta
was set at a
discharging
tested at va
periods, sep
set in rando
force, allow
ducer force
recording p
2.2. Data re
Surface
muscles wa
the belly of
units were r
(Frederick H
muscles. EM
(band-pass: 30 Hz to 1 kHz, 300–3000 Hz, respectively). Force, EMG
and single motor unit signals were digitized (sampling rates: 1 kHz,
5 kHz and 30 kHz, respectively) and stored on a computer by means
of a1401plusacquisitiondevicedrivenbySpike2-5software (Cam-
Elec
valu
ordi
sked
ist ex
ghest
uent
ary m
ngle
gle m
eir d
ing b
frequ
2) in
s fre
e ha
pro
rigge
unit
s. Re
the
ami
thro
mot
r-spi
(abo
unit
I we
otor
unit
s of
h mo
alys
tor u
ting
in F
ution
s bin
ce (t
ak in
zer
zatio
ing i
comp
100
ut in
bitra
chro
us im
nchro
the b
nchr
ount
ordin
as e
976)llenged, however, in a study in which similar EMG DET
observed in two muscles showing a striking difference
it synchrony (Dideriksen et al., 2009). The purpose of
work was to further investigate this controversial issue.
ting the coupling between discharges of wrist extensor
pairs at different contraction levels in human subjects,
cently shown that the occurrence of motor unit syn-
rings above-chance level was enhanced with moderate
contraction strength (Schmied and Descarreaux, 2010).
ata obtained in the wrist extensor muscles originally
el Santo et al. (2006), our first objective was to deter-
G DET was enhanced in association with the increase
nit synchrony established by cross-correlation analy-
onger muscle contraction (Schmied and Descarreaux,
urrence of recurrent patterns in EMG time series is
promoted by rhythmical motor unit synchrony. Accord-
ges in EMG DET have been observed in relation to
ngth-training or Parkinsonian tremor and interpreted
g an increase in the oscillatory synchronous activity of
s (Filligoi and Felici, 1999; Ikegawa et al., 2000; Felici
; Liu et al., 2004; Fattorini et al., 2005). Our second
as therefore to determine if there were any relationship
G DET and the changes in the oscillatory components
it synchronization assessed by coherence analyses at
ls of submaximal muscle contraction.
ls and methods
erimental procedure was approved by the Ethics Com-
niversity of Marseilles-II (CCPPRB protocol 03005).
s were performed on 8 healthy human subjects (males,
years, right-handed) who signed an informed consent
o the experimental procedures, as required by the Dec-
elsinki.
ental procedure
ocol has been described in detail elsewhere (Schmied
reaux, 2010). Briefly, subjects were asked to contract
extensor muscles by pushing the back of their hand
rce transducer device (FT03, Grass Telefactor, W. War-
A). The force signal was displayed on an oscilloscope
g the subjects who were required to maintain the force
rget, without feedback of motor unit activity. The target
level high enough to record the activity of two steadily
motor units over 1.5–3 min. Each motor unit pair was
rious force levels during at least 2 (up to 7) recording
arated by 2-min resting periods. The force target was
m order between the lowest and the highest levels of
ing reliable discrimination of both motor units. Trans-
output, calibrated in Newton, was averaged across each
eriod.
cording
EMG activity on the right extensor carpi radialis (ECR)
s recorded by two electrodes placed 2 cm apart over
these muscles. The discharge patterns of two motor
ecorded concurrently by two tungsten microelectrodes
aer & Co., Bowdoinham, ME, USA) inserted into the ECR
G and motor unit activities were amplified and filtered
bridge
(RMS)
the rec
were a
the wr
The hi
subseq
volunt
2.3. Si
Sin
and th
The fir
neous
and MU
taneou
no spik
ination
spike-t
motor
period
ted on
was ex
tested
The
of inte
300 ms
motor
each IS
each m
motor
in term
for bot
2.4. An
Mo
correla
shown
distrib
in 1-m
referen
tral pe
around
chroni
occurr
1978)
from −
clear-c
was ar
time.
Syn
chrono
The sy
above
The sy
peak c
the rec
peak w
et al., 1tronic Design, Cambridge, UK). The root mean square
es of ECR EMG activity were averaged across each of
ng periods. At the end of the experiment the subjects
to produce 3 bouts of maximal isometric contraction of
tensor muscles, under strong verbal encouragements.
level of ECR EMG activity assessed in these bouts was
ly used to normalize EMG as a percentage of maximal
uscle contraction (% MVC).
motor unit firing pattern analysis
otor unit action potentials were discriminated off-line
istribution was analyzed using the Spike 2-5 software.
ehavior of each motor unit was plotted on an instanta-
ency curve, as illustrated for the two motor units (MU1
Fig. 1A. The presence of abnormally low or high instan-
quency values was carefully monitored to ensure that
d been missed or erroneously included in the discrim-
cess. Motor unit macro-potentials were extracted by
red averaging of ECR EMG activity, using discriminated
action potentials as triggers across each of the recording
producibility of the motor unit macro-potentials, plot-
right of the instantaneous frequency curves in Fig. 1A,
ned to ensure that the same motor units were being
ughout the various levels of force.
or unit firing patterns were characterized on the basis
ke interval (ISI) durations, excluding those longer than
ut 4–5 times the mean), taken to result from pauses in
tonic activity. Instantaneous firing rates associated with
re calculated and averaged across each recording for
unit (FRmean = mean (1/ISI)). The firing pattern of the
pairs at each contraction level was thereafter described
the instantaneous firing rates geometric mean obtained
tor units (FRgeo = sqrt (FRmean 1 × FRmean 2)).
is of motor unit synchronous activity
nit synchronous activity was analyzed by cross-
the 2 spike trains in each of the recording periods, as
ig. 1B and C. Cross-correlation histograms yielded the
of impulses produced by the analyzed motor units
s, 100 ms before and after impulses generated by the
rigger) motor unit. Synchronous firings formed a cen-
the cross-correlograms marked by a white bar located
o in the examples illustrated (Fig. 1B and C). The syn-
n peak was delimited on the basis of rising inflection
n the central region of the cumulative sum (Ellaway,
uted with respect to mean count in a baseline located
to −20 ms in the cross-correlograms. In the absence of
flection around time 0, the strength of synchronization
rily calculated over a 20-ms window centered on trigger
nization strength was evaluated in terms of both syn-
pulse probability and synchronous impulse frequency.
nous impulse probability index is given by peak counts
aseline mean divided by the number of trigger spikes.
onous impulse frequency index is derived from the
above the baseline mean divided by the duration of
g period. Statistical significance of the synchronization
valuated at P < 0.05 on the basis of the z-score (Garnett
.