Content uploaded by Ing-Shiou Hwang
Author content
All content in this area was uploaded by Ing-Shiou Hwang on Nov 10, 2018
Content may be subject to copyright.
Journal of Electromyography and Kinesiology 11 (2001) 327–335
www.elsevier.com/locate/jelekin
Quantitative EMG analysis to investigate synergistic coactivation
of ankle and knee muscles during isokinetic ankle movement.
Part 2: time frequency analysis
I.S. Hwang
a,*
, L.D. Abraham
b
a
Department of Physical Therapy, National Cheng Kung University, Tainan 701, Taiwan
b
Kinesiology and Health Education Department, and the Biomedical Engineering and Neuroscience Graduate Programs, The University of
Texas at Austin, Austin, TX 78712, USA
Received 1 June 2000; received in revised form 23 February 2001; accepted 15 March 2001
Abstract
Fundamental to intralimb coordination in the lower extremity, ankle–knee synergy induced by motor irradiation has long been
employed to secure facilitation of paralyzed muscles. This study, a companion research subsequent to the time amplitude analysis
of surface electromyography in part 1, was to investigate the recruitment strategy of irradiated muscles and prime movers during
ankle isokinetic contraction at different contraction speeds (30, 60, 120 and 240°/s) with time frequency analysis. The results
indicated the recruitment strategies of the major irradiated muscles (ipsilateral rectus femoris/ipsilateral biceps femoris) and prime
movers (anterior tibialis/gastrocnemius) were time-dependent and significantly different in terms of the instantaneous median fre-
quency. In general, the prime movers for ankle isokinetic concentric contraction demonstrated a similar recruitment strategy, irrespec-
tive of different contraction speeds. This finding is consistent with the idea of generalized motor programs that speed is one of the
constraint parameters supplied to motor programs. Nevertheless, the recruitment strategies of the irradiated muscles were highly
inconsistent, varying across trials at different contraction speeds, and were not relevant to those of the prime movers. In addition,
the recruitment in the irradiated muscles seemly limited to motor units of low threshold, in spite of maximal voluntary contraction
of the prime movers. 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Motor irradiation; Electromyography; Isokinetic concentric contraction; Recruitment strategy; Dynamic spectrum
1. Introduction
In part 1 [1], investigation of synergistic coactivation
during isokinetic ankle movement based on time ampli-
tude analysis was described. The results contribute an
understanding of the way patterned irradiation might
assist in providing to joint stability. Similar to most
research of neuromuscular synergy, that study focused
only on the electromyograph (EMG) features of relative
intensity and muscle patterns. However, the information
provided by time-amplitude analysis is incomplete, and
more information about muscle control can be realized
with EMG spectral analysis. Both experimentally and
* Corresponding author. Tel.: +886-6-2353535x5932; fax: +886-
6-2370411.
E-mail address: ishwang@mail.ncku.edu.tw (I.S. Hwang).
1050-6411/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 1 05 0 - 6 4 1 1 ( 0 1) 0 0 0 1 3 - X
theoretically, researchers agree that the use of spectral
shifts of surface EMG allows assessment of the type of
motor units (MUs) recruited, as predominate conduction
velocities and action potential durations of muscle fibers
differ [2–5]. Among several spectral variation indices,
the median frequency (MF) is recommended as an
appropriate representation of recruitment strategy [6–9].
For a strenuous contraction, no studies have ever tried
to investigate how MUs are recruited with motor
irradiation as an effect. To explore new aspects of neuro-
muscular functioning for motor irradiation [10,11], we
used joint time frequency analysis, or dynamic spectral
analysis, to elucidate the inter-relationship of the MU
recruitment strategy for the knee irradiated muscles and
ankle prime movers when the subject performed ankle
isokinetic contraction at different contraction speeds.
With dynamic spectral analysis to reveal the time-local
properties of the EMG spectrum, two hypotheses were
examined:
328 I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
1. the recruitment strategies of the prime movers and
irradiated muscles during ankle isokinetic dorsiflexion
and plantar flexion are speed related; and
2. there exists an analogous recruitment strategy of the
ankle prime movers and of coactivated knee muscles
during motor irradiation.
2. Methods
2.1. Test administration and data processing
Detailed description of the test administration was
provided in [1]. In brief, 11 subjects performed ankle
isokinetic contraction in either direction (plantar flexion
or dorsiflexion) at four different angular velocities, 30,
60, 120 and 240°/s. EMG signals from the ipsilateral
tibialis anterior (TA), medial gastrocnemius (GS), and
ipsilateral/contralateral rectus femoris (RF) and biceps
femoris (BF) were recorded using preamplified bipolar
surface electrodes (Iomed, Inc., electrode spacing 2.5
cm; diameter 1.1 cm, a gain of 380 and a CMRR of
102 dB) placed over the muscle belly. The EMG signals,
conditioned by an analog bandpass filter (6–500 Hz),
were digitized at 1 kHz and processed off-line to yield
the time-varying spectrum [12]. General representation
of the time-frequency characterized by a cone-shape ker-
nel function was employed for off-line analysis of the
recorded EMG [13]. The selection of this time-finite
time frequency distribution was based on Devedeux and
Duchene’s [14] work in which different EMG time fre-
quency distributions were compared. The resolution in
the frequency domain of the time-varying spectrum
analysis was equal to the Nyquist frequency divided by
2
n
(n=9). The resolution in the time domain was 2
n
ms
(n=4), constrained by the uncertainty principle. The
instantaneous median frequency (IMF) was estimated by
extracting the MF from each time frame across the time
frequency distribution, and was further fitted with a
regression polynomial of the lowest order to enhance its
implicit trend (p!0.05).
2.2. Statistical analyses of the control strategies of the
activated motor units during ankle isokinetic
contraction
With respect to the control strategies of the active
MUs, the order of the regression polynomials of the
IMF, the shape of the regression polynomials of the
IMF, and the mean IMF of each prime mover and major
irradiated muscle were examined. The former two were
a quantitative description of the IMF that referred to the
change of recruitment pattern of the active MUs in the
time course. The latter one was the average magnitude
of the IMF that referred to the mean recruitment level
of the active MUs during the contraction period. The
main interests in regard to the regression polynomial of
the IMF were differences:
1. among different contraction speeds; and
2. between prime movers and major irradiated muscles.
To compare the independence of the speed effect, the
order of the regression polynomials of the IMF of the
prime movers and major irradiated muscles were exam-
ined with the Friedman two-way ANOVA by ranks, and
the mean IMF was examined using repeated measures
one way ANOVA. In the same manner, the shape and
the order of the smoothed IMF were examined to com-
pare differences in the recruitment strategy between
prime movers and irradiated muscles. The level of sig-
nificance was 0.05 for all statistical analyses.
3. Results
3.1. Joint time frequency analysis and the recruitment
strategy of activated motor units
Fig. 1 shows a typical example of EMG time-varying
power spectrum as a representation of recruitment strat-
egy for the TA muscle. Fig. 1(c) is the time-frequency
distribution of original EMG [Fig. 1(a)], as illustrated by
the staggered dark bands in the time-frequency plot. The
spiky IMF [Fig. 1(b)], obtained by extracting every MF
across the time frame, represents the temporal change of
spectral profile and recruitment pattern of the TA during
isokinetic concentric contraction. In Fig. 1(b), a lowest
order polynomial, or smoothed IMF, was fitted to
enhance the trend of the TA IMF (N=2, r=0.59, p!.01).
The smoothed IMF suggests that the EMG spectral pro-
file of the TA muscle shifted progressively toward higher
frequency components to the middle of the isokinetic
contraction for recruiting additional larger MUs, and
then decreased gradually as a result of decruitment.
3.2. Recruitment strategies of the prime movers and
irradiated muscles at different contraction speeds
As indicated in [1], the major irradiated muscles of
the knee joint during isokinetic dorsiflexion and plan-
tarflexion were the ipsilateral rectus femoris (IRF) and
both sides of the biceps femoris, respectively. However,
synergistic coactivation of the contralateral biceps fem-
oris during medium and fast isokinetic ankle dorsiflexion
was insignificant for some of the subjects. For the pur-
pose of generality, dynamic spectral analysis was applied
only on data from the major irradiated muscles IRF and
ipsilateral biceps femoris (IBF) and the prime movers of
the ankle dorsiflexion and plantarflexion, TA and gastro-
cnemius (GS). Figs. 2 and 3 show the IMF and smoothed
329I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
Fig. 1. A schematic expression of the dynamic EMG spectrum and its relationship with raw surface EMG data and traditional power density
function. The dynamic spectrum on (c) is an example of joint time frequency analysis of the TA surface EMG during slow isokinetic dorsiflexion.
(a) Original EMG data, (b) instantaneous median frequency, the unbiased estimation of the mean power frequency of the dynamic EMG spectrum
power spectrum, (d) traditional power spectrum based on the FFT algorithm). In (b), the instantaneous median frequency was fitted by a smooth
second-order polynomial function. The simple correlation coefficient and significance level are shown on the plot.
IMF of the TA and irradiated IRF during steady isoki-
netic dorsiflexion at four contraction speeds. It was inter-
esting to note that a second order polynomial was an
appropriate representation of the IMF of the TA for most
dorsiflexion trials, whereas IMF of the IRF muscle was
atypical and could not be fitted with the same order of
polynomials, regardless of the contraction speeds. Table
1 shows that regression orders for both the TA and
irradiated IRF during isokinetic ankle dorsiflexion were
not significantly different for contraction speeds
(p"0.05), however remarkable variance across subjects
of the regression order for the IRF was noted. In the
case of ankle isokinetic plantarflexion, in a sense, the
results were analog to those for isokinetic dorsiflexion.
Fig. 4 displays a typical example of the IMF and
smoothed IMF with a linear trend for the GS during iso-
kinetic plantarflexion contraction. On the other hand, the
IBF IMF at different contraction speeds, fitted with poly-
nomials of different orders, was not consistent in shape
(Fig. 5). Table 1 again shows that regression orders of
IMF for both the GS and IBF were insensitive to speed
effect (p"0.05), but the regressive order of the IMF for
the IBF was rather variant across subjects. Table 2 shows
that mean IMF, similar to the regression order, was not
a function of the contraction speed (p"0.05). Based on
the findings of the regression order and mean IMF level,
recruitment patterns for the prime movers were similar
for all contraction speeds, but recruitment patterns for
the irradiated muscles were atypical.
4. Discussion
4.1. The modulation of the primer mover IMF
One of the main attractions for the use of the surface
EMG spectrum is to provide a promising window for
understanding control strategies for a group of activated
MUs. As the conduction velocity of muscle fiber deter-
mines the profile of MU action potential associated with
MUs of different sizes, the MF can be an indicator of
MU recruitment [8]. The idea was supported by the work
of Solomonow et al. [6,15] who examined the effects of
firing rate and MU recruitment by a dual channel com-
puter-controlled system. They found that linear increase
in the MF resulted from orderly recruitment of large
MUs in the cat gastrocnemius. In this study, we therefore
proposed the IMF to assess time-dependent recruitment
strategy. For isokinetic dorsiflexion, the second order
smoothed IMF illustrated that recruitment strategy for
330 I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
Fig. 2. A typical example of the IMF and smoothed IMF of the TA muscle during isokinetic dorsiflexion at four different contraction speeds.
(a) 30, (b) 60, (c) 120 and (d) 240°/s. Ris the correlation coefficient and Nis the regression order of the smoothed IMF.
the prime mover (TA) seemed to employ smaller MUs
of low threshold in the beginning, and then progressively
recruit larger MUs. The recruitment of high threshold
MUs during concentric contraction was also reported by
Moritani et al. [16]. At the end of contraction, MUs were
decruited in reverse order on account of biomechanical
constraint, a shorter muscle length. Interestingly, for
most of the subjects, the second order of smoothed IMF
was in parallel with the torque recruitment curve during
concentric isokinetic dorsiflexion [17] and the length-
relationship observed for a single muscle fiber [18], both
contain force build-up and force decay phases. These
findings suggest that the recruitment of the TA MUs was
a time and force function, conforming to the size prin-
ciple [19]. In addition, a common recruitment pattern
for the GS muscle during isokinetic plantarflexion was
represented by the first order smoothed IMF. The GS
muscle appeared to recruit small MUs first and then pro-
gressively to recruit larger MUs through the whole con-
traction period, excluding possible effect of muscle
length (discussed next). Our results were also in agree-
ment with orderly recruitment of MUs during low level
of concentric contraction using invasive approach [20],
which is in contrast to selective recruitment of high-
threshold MUs during eccentric contraction [21]. Gener-
ally, recruitment strategies of the TA and GS muscles
in the context of IMF profiles demonstrated remarkable
similarities across various contraction speeds. The
invariant features of the IMF (TA: second order poly-
nomial, GS: first order polynomial) support the idea of
a generalized motor program [22] in that recruitment pat-
tern of the prime movers was temporally compressed and
inherent speed-related parameters were crucial for isoki-
netic concentric contraction. Samples of generalized
motor program were reported elsewhere, such as a simi-
lar temporal structure of typing [23] and constant pro-
portion step cycles during locomotion [24].
Another main finding in the prime movers was recruit-
ment saturation during isokinetic contraction. With
maximal voluntary contraction, notwithstanding differ-
ent speeds, there was no statistical difference in mean
IMF and mean EMG RMS values for the prime movers
during ankle isokinetic contraction. According to the
torque–velocity relationship and empirical measure-
ments [17], it is known that the torque output of the
ankle joint depends on the contraction speed. Slower
contractions result in larger force output, and vice versa.
Although fast twitch fiber MU (large MU) was reported
to be important for force output during fast contraction
[25,26], apparently the speed related force output was
not commensurate with the SEMG, either in the time or
the frequency domain, or both in this study. There are
several possibilities for this finding: first, EMG-force
relationship is definitely not linear during isokinetic
331I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
Fig. 3. A typical example of the IMF and smoothed IMF of the IRF muscle during isokinetic dorsiflexion at four different contraction speeds.
(a) 30, (b) 60, (c) 120 and (d) 240°/s. Ris the correlation coefficient and Nis the regression order of the smoothed IMF.
Table 1
A Friedman two-way analysis of variance by ranks of the regression order of the smoothed IMF of the prime movers and major irradiated muscles
(TA, tibialis anterior; GS, gastrocnemius; IRF, ipsilateral rectus femoris; IBF, ipsilateral biceps femoris)
Speed (°/s) TA IRF
Mean SD Mode Mean SD Mode
Isokinetic dorsiflexion
30 1.91 0.94 2 2.00 2.29 1
60 1.73 0.47 2 2.56 2.51 2
120 1.82 0.87 2 2.67 3.54 2
240 1.91 1.14 2 2.78 2.54 1
Friedman test c
2
(3)=0.484 c
2
(3)=2.631
Sig. of asymp.=0.922 Sig. of asymp.=0.452
GS IBF
Isokinetic plantarflexion
30 1.27 0.47 1 3.73 2.97 1
60 1.55 1.21 1 2.45 2.21 2
120 1.55 0.52 1 2.27 2.41 1
240 1.36 0.47 1 2.18 1.54 2
Friedman Test c
2
(3)=1.875 c
2
(3)=3.037
Sig. of asymp.=0.599 Sig. of asymp.=0.386
movement. Linear relationship between EMG and force
exists only in the isometric contraction with certain cor-
rection of EMG variables [7,15]. Next, fast switch MUs
are not fully recruited at all contraction speeds in spite
of maximal effort. Westing et al. [27] reported that under
certain high-tension loading conditions the neural drive
to the agonist muscles was reduced to protect the muscu-
loskeletal system from an injury. On the other hand, we
332 I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
Fig. 4. A typical example of the IMF and smoothed IMF of the GS muscle during isokinetic plantarflexion at four different contraction speeds.
(a) 30, (b) 60, (c) 120 and (d) 240°/s. Ris the correlation coefficient and Nis the regression order of the smoothed IMF.
Fig. 5. A typical example of the IMF and smoothed IMF of the IBF muscle during isokinetic plantarflexion at four different contraction speeds.
(a) 30, (b) 60, (c) 120 and (d) 240°/s. Ris the correlation coefficient and Nis the regression order of the smoothed IMF.
333I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
Table 2
A repeated measures ANOVA of the mean IMF of the prime movers and major irradiated muscles to examine the independence of the speed effect
(TA, tibialis anterior; IRF, ipsilateral rectus femoris; GS, gastrocnemius, IBF: ipsilateral biceps femoris)
Speed (°/s) Mean IMF (Hz)
Dorsiflexion Plantarflexion
TA IRF GS IBF
30 76.5±14.6 66.0±15.4 115.8±10.3 59.9±8.6
60 79.0±15.7 66.4±17.0 111.0±6.4 59.8±10.5
120 76.3±15.1 64.9±9.8 108.6±5.9 59.8±11.3
240 79.3±13.3 68.2±16.8 112.6±9.9 57.1±10.6
FStatistics FTA(3,30)=1.74, sig. of F=0.180
FIRF(3,24)=4.33, sig. of F=0.730
FGS(3,30)=16.75, sig. of F=0.147
FIBF(3,24)=0.74, sig. of F=0.535
favor that the larger torque output at slower contraction
speed relates to muscle mechanical properties such as:
1. the viscoelastic component in the muscle–tendon unit
which stores contractile energy depending on the con-
traction rate [28]; and 2. the speed related formation of
crossbridges [29].
4.2. The recruitment strategy of the irradiated muscles
In contrast to generalized recruitment of the prime
mover, the recruitment pattern of the irradiated muscle
was inconsistent across trials at different contraction
speeds. This fact was manifest by the varying order of
the smoothed IMF as well as irregular shape of the IMF.
In [1], a larger mean EMG RMS of the irradiated knee
muscles indicated more MUs were recruited during
slower ankle isokinetic contraction. However, this
additional recruitment, lacking a consistent temporal pat-
tern, was different from steady isometric contraction at
afixed level. Consequently, during ankle isokinetic con-
traction of various speeds, individuals may exhibit
reasonable consistency in the amount of irradiated mus-
cular activity [30,31], but not systematic recruitment of
MUs. Since cortical inhibitory mechanisms influence
activity of the αmotor neurons [30], we argue that the
inhibitory system regulates merely the amount of overall
inhibition of voluntary effort. The inhibition is not selec-
tive, so the recruitment patterns in terms of the IMF are
arbitrary and unlikely to be a recruitment function of the
prime mover. In consideration of a larger mean EMG
RMS but insignificant increase in mean IMF during
slower ankle isokinetic contraction, additional recruit-
ment of the irradiated muscle might exclude those larger
MUs. Hence, the recruitment strategy of the irradiated
muscles of fixed length is different from progressive iso-
metric contraction during which MUs are recruited in
order [2,3,9]. It is likely that irradiation may recruit
many intermittent MUs which occasionally stop firing
and later be recruited again [32]. Recruitment of inter-
mittent MUs was also reported in low level of aniotonic
contraction (20% MVC) [33]. This irradiated recruit-
ment, serving to joint stability as described in [1], delin-
eates a functional coupling of motor neuronal organiza-
tion within the central nervous system (CNS) to muscle
fiber organization peripheral to the CNS is time-variant,
because synaptic excitation entering to spinal motor pool
is rather atypical.
4.3. Further consideration of dynamic spectral
analysis
Several possibilities for processing EMG during
dynamic contraction should be taken into consideration
before finalizing the conclusion in this study. First, mus-
cle length of the prime movers varied during isokinetic
contraction in a way that might affect the general trend
of the IMF. This is because muscle conduction velocity
is conditional to the change of length and diameter of
muscle fibers so that a corresponding alteration of the
MUAP profile occurs [34–36]. According to this argu-
ment, a shorter muscle length would lead to the EMG
power spectrum shifting toward high frequency compo-
nents at a fixed level of isometric contraction, and vice
versa. As a matter of fact, the change of the EMG power
spectrum might not be real change of the recruitment
patterns but physical phenomena. However, Vander Lin-
den et al. [37] used indwelling wire EMG to investigate
recruitment patterns of the MUs in the human TA mus-
cle. They concluded that the discharge pattern of the
MUs changes with muscle length. In addition, we also
noted that MF of the surface EMG during a fixed level
(50% MVC) of isometric contraction varied with muscle
length for both the TA and GS muscles (p!0.05)
[unpublished data (n=10)]. For a contraction at a shorter
muscle length, it took greater effort to maintain a certain
level of contraction resulting in higher MF. Hence,
whether the spectral change of the EMG represents the
analogous modulation of recruitment patterns or merely
334 I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
a measuring partiality is still controversial. However, it
is worthy of mentioning that spectral change of physical
origins cannot explain the significant reduction of the
IMF near the end of muscle contraction for most of the
TA muscles and some of the GS muscles (second order
polynomial of the smoothed IMF). With respect to the
major irradiated muscles, the length effect could be
ignored reasonably since the knee joint was fixed and
contraction of the irradiated muscle was isometric. Sec-
ondly, we cannot rule out the possibility of the spectral
variation being due to a geometrical change between
muscle fibers and the collecting electrodes. This
phenomenon could explain in part for the
intersubject/intertrial variability occurred in this study.
Third, the resolution of the EMG time-frequency analy-
sis may change with signal/noise ratio. As indicated in
[1], the irradiated muscle activity at the higher contrac-
tion speed was smaller than that at slower contraction
speeds. The accuracy and consistency of estimating the
IMF and smoothed IMF of the irradiated muscles might
be affected because of a relatively low signal/noise ratio
during very fast isokinetic ankle contraction.
4.4. Clinical implications
Muscle weakness and loss of coordination could be a
result of segmental or multiple injuries of the neuro-
muscular system. Stimulation of the proprioceptive sys-
tem, or proprioceptive neuromuscular facilitation (PNF),
has long been considered a promising treatment to attain
a greater motor response and to re-educate impaired
muscles. The central idea of the techniques is to employ
mass movement patterns induced by motor irradiation to
restore integrated motor functions. The findings of this
research characterize many interesting features of motor
irradiation, complex for rehabilitation clinicians while
applying the facilitation technique. First, the recruitment
of MUs in the irradiated muscles was neither patterned
nor consistent. Lack of patterning implies subsequent
reeducation programs are needed for paralyzed muscles.
Second, irradiated recruitment by means of the PNF
technique may limit to small MUs of lower threshold.
References
[1] Hwang IS, Abraham LD. Quantitative EMG analysis to investi-
gate synergistic coactivation of ankle and knee muscles during
isokinetic ankle movement. J Electromyogr Kinesiol
2001;11(5):319–325.
[2] Lindstrom L, Magnussum R. Interpretation of myoelectric spec-
tra. A model and its applications. Proceedings of IEEE
1977;65:653–62.
[3] Bilodeau M, Arsenault AB, Gravel D, Bourbonnais D. EMG
power spectra of elbow extensors during ramp and step isometric
contractions. Eur J Appl Physiol 1991;63(1):24–8.
[4] Bilodeau M, Cincera M, Gervais S, Arsenault AB, Gravel D,
Lepage Y, McKinley P. Changes in the electromyographic power
spectrum distribution caused by a progressive increase in the
force level. Eur J Appl Physiol 1995;71(2/3):113–23.
[5] Fuglevand AJ, Winter DA, Patla AE, Stashuk D. Detection of
motor unit action potentials with surface electrode: influence of
electrode size and spacing. Biol Cybern 1992;67(2):143–53.
[6] Solomonow M, Baratta R, Zhou BH, Shoji H, D’Ambrosia RD.
The EMG-force model of electrically stimulated muscles: depen-
dence on control strategy and predominant fiber composition.
IEEE T Biomed Eng 1987;34(9):692–703.
[7] Solomonow M, Baten C, Smit J, Baratta R, Hermens H, D’Am-
brosia R, Shoji H. Electromyogram power spectra frequencies
associated with motor unit recruitment strategies. J Appl Physiol
1990;68(3):1177–85.
[8] Bernardi M, Solomonow M, Sanchez JH, Baratta RV, Nguyen
G. Motor unit recruitment strategy of knee antagonist muscles in
a step-wise, increasing isometric contraction. Eur J Appl Physiol
1995;70(6):493–501.
[9] Bernardi M, Solomonow M, Baratta RV. Motor unit recruitment
strategy of antagonist muscle pair during linearly increasing con-
traction. Electromyogr Clin Neurophysiol 1997;37(1):3–12.
[10] Zulch KJ, Muller N. Associated movements in man. In: Uniken
PJ, Bruyn GW, editors. Handbook of clinical neurology, vol. 1.
New York: Wiley, 1969:404–26.
[11] Dimitrijevic MR, McKay WB, Sarjanovic I, Sherwood AM,
Svirtlih L, Vrbova G. Co-activation of ipsi- and contralateral
muscle groups during contraction of ankle dorsiflexors. J Neurol
Sci 1992;109(1):49–55.
[12] Cohen L. Time frequency analysis. New Jersey: Prentice Hall,
1995.
[13] Zhao Y, Atlas LE, Marks RJ. The use of cone-shaped kernels
for generalized time-frequency representations of nonstationary
signals. IEEE T Acoustic, Speech, Signal Processing
1990;38(7):1084–91.
[14] Devedeux D, Duchene J. Comparison of various time/frequency
distributions (classical and signal dependent) applied to synthetic
uterine EMG signals. In: Proceedings of the IEEE-SP Inter-
national Symposium on Time-Frequency and Time-Scale Analy-
sis 1994:572–75.
[15] Solomonow M, Baratta R, Shoji H, D’Ambrosia R. The EMG-
force relationships of skeletal muscle; dependence on contraction
rate, and motor units control strategy. Electromyogr Clin Neuro-
physiol 1990;30(3):141–52.
[16] Moritani T, Muramatsu S, Muro M. Activity of motor units dur-
ing concentric and eccentric contractions. Am J Phys Med
1988;66(6):338–50.
[17] Dvir Z. Isokinetics —muscle testing, interpretation and clinical
applications. New York: Churchill Livingstone, 1995:1–39.
[18] Gordon AM, Huxley AF, Julian FJ. The variation in isometric
tension with sarcomere length in vertebrate muscle fibers. J Phy-
siol (London) 1966;184:170–92.
[19] Henneman E, Somjen G, Carpenter DO. Functional significance
of cell size in spinal neurons. J Neurophysiol 1965;50:560–80.
[20] Søgaard K, Christensen H, Jensen BR, Finsen L, Søgaard G.
Motor control and kinetics during low level concentric and eccen-
tric contractions in man. Electroencephal Clin Neurophysiol
1996;101:453–60.
[21] Nardone A, Romano
`C, Schieppati M. Selective recruitment of
high-level human motor units during voluntary isotonic lengthen-
ing of active muscles. J Physiol (London) 1989;409:451–71.
[22] Schmidt RA. Motor control and learning: A behavioral emphasis.
Champaign: Human Kinetics, 1988:240–57.
[23] Terzuolo CA, Viviani P. The center representation of learning
motor programs. In: Talbot RE, Humphrey DR, editors. Posture
and movement. New York: Raven, 1979.
[24] Shapiro DC, Zernicke RF, Gregor RJ, Diestel JD. Evidence for
generalized motor programs using gait-pattern analysis. J Motor
Behav 1981;13:33–47.
335I.S. Hwang, L.D. Abraham / Journal of Electromyography and Kinesiology 11 (2001) 327–335
[25] Thorstensson A, Grimby G, Karlsson J. Force–velocity relations
and fiber composition in human knee extensor muscles. J Appl
Physiol 1976;40(1):12–6.
[26] Cabri JMH. Isokinetic strength aspects of human joints and
muscles. Crit Rev Biomed Eng 1991;19(2,3):231–59.
[27] Westing SH, Cresswell AG, Thorstensson A. Muscle activation
during maximal voluntary eccentric and concentric knee exten-
sion. Eur J Appl Physiol 1991;62(2):104–8.
[28] Zhou BH, Baratta R, Solomonow M. Manipulation of muscle
force with various firing rate and recruitment control strategies.
IEEE T Biomed Eng 1987;34(2):128–39.
[29] Enoka RM. Neuromechanical basis of kinesiology. Champaign:
Human Kinetics, 1988:155–62.
[30] Todor JI, Lazarus JA. Exertion level and the intensity of associa-
ted movements. Dev Med Child Neurol 1986;28(2):205–12.
[31] Armatas CA, Summers JL, Bradshaw JL. Mirror movements in
normal adult subjects. J Clin Exp Neuropsychol
1994;16(3):405–13.
[32] Thomas CK, Ross BH, Calancie B. Human motor-unit recruit-
ment during isometric contraction and repeated dynamic move-
ments. J Neurophysiol 1987;57:311–24.
[33] Søgaard K. Motor unit recruitment pattern during low-level static
and dynamic contractions. Muscle Nerve 1995;18:292–300.
[34] Inbar GF, Allin J, Kranz H. Surface EMG spectral changes with
muscle length. Med Biol Eng Comput 1987;25(6):683–9.
[35] Okada M. Effect of muscle length on surface EMG wave forms
in isometric contractions. Eur J Appl Physiol 1987;56(4):482–6.
[36] Kossev A, Gantchev N, Gydikov A, Gerasimenko Y, Christova P.
The effect of muscle fiber length change on motor units potentials
propagation velocity. Electromyogr Clin Neurophysiol
1992;32(6):287–94.
[37] Vander Linden DW, Kukulka CG, Soderberg GL. The effect of
muscle length on motor unit discharge characteristics in human
tibialis anterior muscle. Exp Brain Res 1991;84(1):210–8.
Ing-Shiou Hwang received a B.S. degree in
physical therapy from the National Taiwan Uni-
versity, Taiwan, in 1988, and the M.S. degree
in 1993, and the Ph.D. degree in 1998, both in
biomedical engineering, from the National
Cheng Kung University, Taiwan and the Uni-
versity of Texas at Austin, USA, respectively.
He has been an assistant professor in the Depart-
ment of Physical Therapy, National Cheng Kung
University, Taiwan since 1999. His major inter-
ests are in biosignal processing and neuro-
muscular control.
Lawrence D. Abraham received the A.B.
degree in physical education from Oberlin Col-
lege in 1971, the M.S. degree in physical edu-
cation (motor learning) from Kansas State Tea-
chers College in 1972, and the Ed.D. degree in
physical education (motor learning and
biomechanics) from Teachers College, Colum-
bia University in 1975. He is an associate pro-
fessor of biomechanics and motor control in the
Department of Kinesiology and Health Edu-
cation at The University of Texas at Austin,
USA. He is also on the graduate faculty in the
Institute for Neuroscience and the Biomedical Engineering Program at The
University of Texas at Austin. His major interests are in biomechanical
and neurophysiological analysis of human motor performance coordi-
nation.