Journal of Electromyography and Kinesiology 16 (2006) 144–148

www.elsevier.com/locate/jelekin

Innervation zone shift with changes in joint angle

in the brachial biceps
S. Martin *, D. MacIsaac
Institute of Biomedical Engineering, Department of Electrical and Computer Engineering, University of New Brunswick, 25 Dineen Dr.,
P.O. Box 4400, Fredericton, NB, Canada E3B 5A3

Received 8 November 2004; received in revised form 16 June 2005; accepted 20 June 2005

Abstract

Empirical evidence is presented suggesting that the innervation zone of the brachial biceps shifts relative to recording electrodes
with changes in joint angle. Myoelectric signal data were acquired from five subjects using a 16-channel linear electrode array, and
analyzed to determine a reversal in signal propagation direction indicating innervation zone location. An analysis of the effect of
joint angle changes on innervation zone location yielded statistically significant results (ANOVA, a = 0.05, p < 0.001) suggesting
that the innervation zone moves between 5 and 30 mm in a direction distal to the shoulder as the arm is extended, statistically inde-
pendent of force level (ANOVA, a = 0.05, p > 0.2).

2005 Elsevier Ltd. All rights reserved.

Keywords: Innervation zone; Electrode array; Myoelectric signal; Conduction velocity

1. Introduction the introduction of new fibers into the electrode detec-

Muscle fiber conduction velocity (CV) is a frequently
used measurement in muscle fatigue assessment strate-
gies [1,2] and the analysis of muscle pathology [11].
CV estimation using surface myoelectric signals (MES)
is often achieved by taking the cross-correlation of two
measurements separated along the MES propagation
path by a known distance. The time shift in the cross-
correlation from zero represents the time taken by the
propagating signal to travel the separation distance.
Dividing the inter-electrode distance by the time shift
yields CV [2].
Because the MES is a summation of activity across
multiple muscle fibers, CV estimation using the above
method will result in a CV averaged across the active fi-
bers. This estimate may be altered due to (1) fiber
recruitment during increased force production, or (2)
*
Corresponding author.
E-mail address: shawn.martin@unb.ca (S. Martin).
tion region because of changes in joint angle. CV accu-
racy also depends on certain assumptions, the first of
which is that the recording electrodes are perfectly
aligned along the length of the muscle fiber. Because
muscle fibers are not all parallel, this assumption can
never be completely valid. Limiting the inter-electrode
distance has been shown to minimize this effect [3].
The second assumption is that the fibers are sufficiently
long such that the effects of non-propagating end effects
may be ignored. Although this assumption may be real-
ized for certain joint angles, muscle geometry changes
during dynamic contractions may render the assumption
invalid.
The third assumption is that both recording elec-
trodes are on the same side of the muscle innervation
zone (IZ). Through the use of judicious electrode place-
ment, this assumption may be met under static condi-
tions. However, when considering dynamic conditions,
changing muscle geometry may cause the IZ to move
relative to the recording electrodes, which in turn may

1050-6411/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jelekin.2005.06.010
 S. Martin, D. MacIsaac / Journal of Electromyography and Kinesiology 16 (2006) 144–148 145

result in one or more electrodes being placed over the IZ

[4]. Thus, an initial step in obtaining an accurate CV
estimation technique, particularly for use in dynamic
contraction scenarios, is to determine the amount of
innervation zone shift with changes in joint angle, and
whether this shift is affected by the contraction level.

2. Materials and methods

2.1. Subjects

Five healthy voluntary participants were used for

data collection; one female and four males, aged 22,
23, 24, 25 and 38, with no known instances of neuro-
muscular disorders. The muscle of interest was the bra-
chial biceps in the dominant arm. Each participant
signed an informed consent form before data collection
began.

2.2. Test procedure

The apparatus used to enable proper joint angle posi-

tion consisted of a large disc attached to a central pulley
at its axis as previously described in [8], and as shown in
Fig. 1(a). This apparatus allowed the subjects to rest
their upper arm on a platform while flexing and extend-
ing at the elbow to create the desired joint angle. By
attaching a mass to the pulley, a specified force could
be applied perpendicular to the forearm regardless of
joint angle.
The data acquisition stage was divided into two ses-
sions. During the first session, maximum voluntary con-
traction (MVC) measurements were obtained for the
dominant arm of each subject for joint angles of 50,
70, 90, 110 and 130 (relative to 180 at full elbow
extension). To accomplish this, the central pulley appa-
ratus was used in a fastened mode by attaching the pul-
ley to the frame via a LC105 Aluminum ÔSÕ Beam Load
Cell. The load cell output a voltage proportional to the
amount of tension being exerted. Subjects were encour-
aged to pull as hard as possible against the fastened pul-
ley for approximately 5 s. The process was repeated
three times, with two minute intervals between each con-
traction. The highest value for each joint angle across
the three trials was used as the MVC for that angle.
The order of the joint angles used in each trial was
randomized.
Subjects returned for a second session at least one
hour after the MVC test. This delay was deemed ade-
quate to allow subjects to recover from the MVC mea-
surements, and thus not corrupt the data set with
fatiguing conditions [10]. Using the central pulley appa-
ratus, subjects were instructed to hold the disc at 50,
70, 90, 110 and 130 against loads equivalent to
approximately 20%, 40% and 60% of their MVC for
Fig. 1. (a) Central pulley apparatus. (b) 16-channel linear electrode
array (left) and adhesive attachment strip (right).
each angle. Each angle–force combination was held for
10 s, with two minute intervals between each contrac-
tion. The order of the joint angles used in each trial
was randomized.
The MES was recorded from the brachial biceps
using a flexible, adhesive 16-channel linear electrode ar-
ray as shown in Fig. 1(b), with an inter-electrode dis-
tance of 10 mm. Each electrode was rectangular and
approximately 5 mm · 1 mm. An Ag|AgCl Red-dot
electrode was placed on the triceps muscle as a reference.
Before electrodes were placed, the skin surface on which
they were mounted was cleansed with rubbing alcohol.
The array was placed along the length of the bicep mus-
cle on the dominant arm, with 20 lL of electrode paste
being applied to each electrode using a method similar
to that described in [5].
2.3. Data acquisition
The MES data from the electrode array were pro-
cessed first through a Prima EMG16 acquisition system
in a single differential mode with a gain of 5000 and
using a 10–500 Hz, fourth-order Bessel band-pass filter.
146 S. Martin, D. MacIsaac / Journal of Electromyography and Kinesiology 16 (2006) 144–148

The CMRR for the acquisition system was >96 dB, with served, resulting in IZ locations (assumed to be centered
an input impedance >90 MX over the entire bandwidth. on these boxes) of 55, 75, 75, 75 and 85 mm for joint an-
The data were then sampled using a National Instru- gles of 50, 70, 90, 110 and 130, respectively.
ments 12-bit DAQ-6024E PCMCIA acquisition card A plot of the mean IZ estimate at joint each angle
with a sampling frequency of 1024 Hz. averaged across subjects for each MVC level is shown
in Fig. 3. The IZ estimates are shown in terms of a
2.4. Data processing change from the estimate at 50 (taken as 0 mm). This
plot shows a statistically significant angle effect (ANO-
In total, 45 data records were recorded for each sub- VA, a = 0.05, p < 0.001), but no statistically significant
ject. To determine the location of the IZ estimate, force effect (ANOVA, a = 0.05, p > 0.2). The total
100 ms segments of raw MES from adjacent channels amount of IZ shift observed was between 5 and
were plotted to observe the location of the reversal in 30 mm over the range of joint angles.
signal propagation direction [6]. The location of the IZ Table 1 summarizes the CV estimates (in m/s) for
under the electrode array was estimated by determining each joint angle for subject P2 for 20% MVC, and shows
which electrodes encompassed this reversal. An estimate the influence of the IZ shift on the variability of CV
of the IZ location for each subject was made at each an- measurements as the joint angle changes.
gle for the three trials, and then averaged across trials,
giving an estimate for the location of the IZ at each joint
angle for each subject. This procedure was repeated for
each MVC level. To determine the amount of shift, the
channel most proximal to the shoulder was designated
to be at 0 mm, and each of the remaining channels
was located in reference to it. To demonstrate the effects
of any shift on CV estimation, an estimate of the CV
between channels for each joint angle was calculated
using the cross-correlation method, and then averaged
across trials.

3. Results

A plot of a sample of raw MES for subject P2 at 20%

MVC is shown in Fig. 2. Boxes indicate the channels be-
tween which a reversal in propagation direction was ob-

Fig. 3. The mean IZ location estimate for each joint angle across
subjects for MVC levels of 20%, 40% and 60%.

Table 1
CV estimates in m/s for subject P2
Channel Angle ()
50 70 90 110 130
1 Inf 22.8 18.5 14.2 12.8
2 14.2 8.5 8.5 8.5 9.1
3 9.1 12.8 11.9 11.9 9.7
4 7.1 8.1 8.1 7.7 6.8
5 11.0 5.5 5.2 5.0 4.7
6 9.2 5.3 5.0 5.1 5.0
7 18.5 7.0 5.7 5.1 5.0
8 18.5 0.7 3.0 4.5 7.0
9 11.9 11.9 14.2 24.2 5.0
10 8.7 6.4 5.5 5.5 5.5
11 8.5 6.7 5.7 5.5 5.7
12 11.1 7.4 7.0 5.5 5.1
13 15.6 14.2 10.5 7.8 6.2
Fig. 2. Raw MES collected from channels 5–12 for subject P2. Boxes
14 12.8 11.9 11.1 9.7 9.1
indicate channel range in which reversal in propagation direction was
15 2.2 2.4 2.4 3.0 3.4
identified.
 S. Martin, D. MacIsaac / Journal of Electromyography and Kinesiology 16 (2006) 144–148
4. Discussion
Results from this experiment indicate that there is a
statistically significant IZ movement in a direction distal
to the shoulder as the arm is extended from 50 to 130.
This corresponds to what we would expect if the shift
was the result of muscle lengthening as joint angle
changes. The amount of movement is subject dependent
due to the varying nature of bicep geometries and arm
lengths. Similar results have been previously reported
in [9].
There are multiple possible explanations for the ob-
served changes in IZ location. One explanation is that,
due to muscle shortening/lengthening, the IZ may shift
relative to the recording electrodes or fiber terminations.
A second explanation is that, due to increased force, more
motor units (MU) are contributing to the contraction,
and thus the IZ as detected by the electrode array may ap-
pear to be increasing in size. This may occur if the neuro-
muscular junctions of the recruited fibers are outside of
the IZ range estimated during the lower contraction level.
However, increased force (and thus increased MU
recruitment) did not result in significantly different esti-
mates for IZ location; therefore, increased MU recruit-
ment may be discounted as a possible explanation.
An analysis of the CV estimates yields an interesting
trend; as joint angle increases from 50 to 130, the vari-
ability of CV estimates across channels decreases. It is
hypothesized that this decrease is because, as the muscle
is lengthened, there is more distance between the IZ and
the end points, thus affording more acceptable electrode
locations for measurement. Also, it may be observed that
the CV estimates from the electrodes most proximal to
the shoulder are heavily biased, which is expected due
to the fiber termination effects. The estimates from the
electrodes most distal from the shoulder are somewhat
less biased, perhaps due to the fact that the electrode
array did not extend across the elbow, and thus did not
fully reach the fiber termination point. While these results
are only reported for one subject, they are typical of all
subjects in the study, and reflect the importance of IZ
avoidance, particularly during dynamic contraction sce-
narios where the joint angle may be varied significantly.
While these results are intriguing, there are a few
experimental issues that require consideration. As men-
tioned, there was a total IZ shift of between 5 and
30 mm for the subject group. As the array had an in-
ter-electrode distance of 10 mm, the IZ location could
only be estimated as being within a 10 mm wide area,
and thus IZ size and shift estimates may be subject to
some inaccuracies. A decrease in electrode separation
would yield more accurate estimates of IZ location.
The presence of multiple IZs was not explored in de-
tail. While only one change in MES propagation direc-
tion was observed in the centremost eight channels,
this does not preclude the presence of multiple, though
147
small, IZs distributed across a small region of the biceps
[7]. Again, a decrease in inter-electrode distance may
yield more information about the presence, if any, of
multiple IZs.
5. Conclusions
Statistically significant results for the effect of joint
angle on IZ shift were presented. The movement was ob-
served to take place in a direction distal to the shoulder
for each of the five subjects, with a maximum shift of
30 mm. This effect was also shown to be independent
of an increase in force level.
This work provides a starting point to implement a
variety of dynamic contraction CV estimation tech-
niques, such as the use of a gating algorithm to selec-
tively turn off electrodes that are deemed to be within
the IZ. Future iterations of this study could incorporate
an electrode array with a more dense distribution of
electrodes, which may provide more accurate IZ estima-
tions, as well as a larger number of force levels to im-
prove the analysis of force effects.
Acknowledgment
This work was supported in part by NSERC Grant
261794-04.
References
[1] L. Arendt-Nielsen, M. Zwarts, Measurement of muscle fiber
conduction velocity in humans: techniques and applications,
Journal of Clinical Neurophysiology 6 (1989) 173–190.
[2] H. Broman, G. Bilotto, C.J. De Luca, Myoelectric signal
conduction velocity and spectral parameters: influence on force
and time, Journal of Applied Physiology 58 (5) (1985) 1428–1437.
[3] C.J. De Luca, M. Knaflitz, Surface Electromyography: WhatÕs
New? CLUT, Torino, Italy, 1992.
[4] D. Farina, R. Merletti, M. Nazzaro, I. Caruso, Effect of joint angle
on EMG variables in leg and thigh muscles, IEEE Engineering in
Medicine and Biology Magazine 20 (6) (2001) 62–71.
[5] D. Farina, M. Pozzo, E. Merlo, A. Bottin, R. Merletti, Assess-
ment of average muscle fiber conduction velocity from surface
EMG signals during fatiguing dynamic contractions, IEEE
Transactions on Biomedical Engineering 51 (8) (2004) 1383–1393.
[6] T. Masuda, H. Miyano, T. Sadoyama, The position of innervation
zones in the biceps brachii investigated by surface electromyog-
raphy, IEEE Transactions on Biomedical Engineering BME-32 (1)
(1985) 36–42.
[7] T. Masuda, T. Sadoyama, Distribution of innervation zones in
the human biceps brachii, Journal of Electromyography and
Kinesiology 1 (2) (1991) 107–115.
[8] D. MacIsaac, P. Parker, R.N. Scott, K. Englehart, A. Cechetto,
Influence of dynamic factors on myoelectric signal estimation, IEEE
Engineering in Medicine and Biology Magazine 20 (6) (2001) 84–89.
[9] K. Saitou, M. Okada, T. Sadoyama, T. Masuda, Effect on surface
EMG waveforms of electrode location with respect to the
neuromuscular junctions, in: Proceedings of 8th Congress of
ISEK, Baltimore, USA, 1990.
148 S. Martin, D. MacIsaac / Journal of Electromyography and Kinesiology 16 (2006) 144–148
[10] I. Yaar, Innervation-zone width in disease and its changes with
fatigue, Journal of Electromyography and Kinesiology 2 (4) (1992)
252–256.
[11] M.J. Zwarts, G. Drost, D.F. Stegeman, Recent progress in the
diagnostic use of suface EMG for neurological diseases, Journal
of Electromyography and Kinesiology 10 (2000) 287–291.
Shawn Martin is a M.Sc.E.E. candidate at
the Institute of Biomedical Engineering and
the Department of Electrical and Computer
Engineering, University of New Brunswick,
Fredericton, Canada. He received a B.Sc.E.
in Computer Engineering from the Uni-
versity of New Brunswick in 2004. His
current research interests are in the area of
biological signal processing.
Dawn MacIsaac received a PhD from the
University of New Brunswick in 2004, a
M.Sc.E. in Electrical Engineering from the
University of New Brunswick in 1999, and
received a BEng in Electrical and
Computer Engineering from McMaster
University in 1996. She also holds a BEd
from QueenÕs University, which she
received in 1991 and a BPE from McMas-
ter University in 1990. She is currently an
Assistant Professor in the Department of
Electrical and Computer Engineering, and
the Faculty of Computer Science at the University of New Brunswick.
Her research interests are in EMG signal processing and applications
of time–frequency analysis to biosignals. She is a member of the IEEE
Engineering in the Medicine and Biology Society, the IEEE Signal
Processing Society, and the Canadian Medical and Biological
Engineering Society.