Content uploaded by Ranu Jung
Author content
All content in this area was uploaded by Ranu Jung on Sep 20, 2015
Content may be subject to copyright.
W. Jensen et al. (eds.), Replace, Repair, Restore, Relieve – Bridging Clinical
and Engineering Solutions in Neurorehabilitation, Biosystems & Biorobotics 7,
529
DOI: 10.1007/978-3-319-08072-7_78, © Springer International Publishing Switzerland 2014
Information on Ankle Angle from
Intramuscular EMG Signals during
Development of Muscle Fatigue in an
Open-Loop Functional Electrical Stimulation
System in Rats
Line E. Lykholt1, Sahana Ganeswarathas1, Anil K. Thota2,
Kristian Rauhe Harreby1, and Ranu Jung2
1 Center for Sensory-Motor Interaction (SMI),
Department of Health Science and Technology (HST),
Aalborg University (AAU), Denmark
{llykho09,sganes09}@student.aau.dk,
krauhe@hst.aau.dk
2 Adaptive Neural Systems Laboratory,
Department of Biomedical Engineering,
Florida International University, USA
{rjung,athota}@fiu.edu
Abstract. Functional Electrical Stimulation (FES) is one method available for
rehabilitation of spinal cord injured subjects. Although FES is used in the clinic
today, reliable and robust feedback for a closed-loop system is limited.
The objective was to examine if intramuscular electromyographic (iEMG) re-
cordings (of tibialis anterior and gastrocnemius medialis) can provide reliable
information of functional movement (i.e. ankle angle) during development of
fatigue.
Four longitudinal intrafascicular electrodes (LIFEs) were implanted in two fas-
cicles of the sciatic nerve in three adult Sprague-Dawley rats. Open-loop FES was
applied to produce rhythmic ankle movement. The FES stimulation pulse widths
and amplitudes were determined for the individual rats based on the strength dura-
tion curve. Each frequency (30, 40, 50, 60 and 70 Hz) was applied to perform 100
step cycles followed by a 15 min rest period. Kinematic information on the ankle
angle and iEMG were recorded simultaneously.
The results showed that the ankle angle and the iEMG amplitude decreased
when the muscles fatigued. A correlation between the ankle angle and iEMG was
present, which indicates that iEMG information can be used as feedback for a
closed-loop system. The correlation was higher at higher stimulation frequencies
(>0.76 at stimulation frequencies above 40 Hz).
1 Introduction
Injury to the spinal cord may cause permanent loss of voluntary motor function
and sensation below the level of the lesion. Functional electrical stimulation (FES)
530 L.E. Lykholt et al.
is a technique that has been used for many years for the rehabilitation of subjects
with spinal cord injury. [1], [2], [3]
The aim of the FES is to electrically activate the paralyzed muscles in a con-
trolled way to restore motor function. FES can be applied to the subject using an
open-loop (feedforward) or a closed-loop (feedback) control strategy. [1], [2], [3]
FES is commonly applied in the clinic in an open-loop mode that operates with
fixed stimulation parameters. A clear advantage is that the paradigm is simple and
easy to use. However, prolonged stimulation leads to muscle fatigue. An appropri-
ate closed-loop stimulation strategy could alleviate this problem. However,
closed-loop FES systems are dependent on feedback from the part of the body that
is controlled, and the availability of sensors and signals to provide a reliable feed-
back signal from the controlled limb or organ is therefore essential. [1], [2], [3]
One source of feedback can be achieved by recording kinematic data that pro-
vides information on the joint angles. Kinematics would reveal when fatigue oc-
curs. [1], [4] However, the recording of kinematic information requires specialized
equipment and is a highly time consuming procedure, which is not suitable for
daily use in the clinic or at home. Therefore it will be important to have access to
another source of information. Information on movement and muscle fatigue can
also be obtained through electromyography (EMG) recordings. The use of EMG
recordings would be relatively easy to implement in the clinic or in a portable
system since it is cheap, quick to setup and is used routinely today. Surface EMG
has the advantage of being easy to record but suffers from cross talk and the daily
need to don and doff the electrodes. An alternative signal source is the use of
intramuscular EMG (iEMG), which is a more invasive technique. This may help
overcome some of the drawback associated with surface EMG. Previous studies
show that it is possible to extract information from the iEMG related to the force
during movement [5].
There is today limited knowledge on whether information on joint angles may
be extracted from iEMG during normal movement and how muscle fatigue may
influence this.
The objective of this study was therefore to examine if information extracted
from iEMG recordings can provide reliable information on a functional movement
and during development of muscle fatigue in a rat model.
To investigate the objective, FES using longitudinal intrafascicular electrodes
(LIFEs) was used to produce a cyclic movement of the hindlimb of the rat while
recording iEMG. The FES was applied in an open-loop mode to induce muscle
fatigue over time. Kinematic data was also recorded as a reliable measure of the
movement.
2 Methods
Data was obtained from three adult healthy male Sprague-Dawley rats (294-615
g). The experimental procedures were approved by Florida International Universi-
ty Institutional Animal Care and Use Committee.
Information on Ankle Angle from Intramuscular EMG Signals 531
2.1
Animal Preparation
After induction of anesthesia with isoflurane gas (5 %), a single injection of So-
dium Pentobarbital (40 mg/kg ip) was given. Anesthesia was maintained with
isoflurane (0.5-2.0 %), throughout the experiment. The level of anesthesia was
assessed with toe pinch, and observation of eye blink and the respiration rate. To
prevent dehydration regular subcutaneous injections of isotonic saline in the dorsal
cavity were administered.
The left sciatic nerve was exposed and four single channel LIFEs were inserted
into the fascicles innervating the Tibialis Anterior (TA) and Gastrocnemius Me-
dialis (GM) muscles. The TA and GM are the main muscles involved in the
movement of the ankle and can be activated from one nerve. To verify that the
electrodes were placed correctly inside the fascicles, electrical stimulation was
applied while observing TA and GM muscle twitch. The LIFEs were sutured to
the epineurium and the incision was closed.
Table 1 Applied LIFE stimulation parameters
Amplitude (ߤܣ) Pulse width (ߤݏ)
Rat 1: GM 50 30
Rat 1: TA 30 30
Rat 2: GM 50 60
Rat 2: TA 250 100
Rat 3: GM 30 40
Rat 3: TA 20 30
2.2
Experimental Setup and Data Acquisition
The rat was placed in a prone position on an elevated platform so that the hin-
dlimbs were hanging freely.
3-D kinematic data was recorded by a Peak Motus System (Peak Performance
Technologies, Inc. Centennial, CO) by placing cone shaped three-dimensional
reflective markers on the hip, knee, ankle and toe. The system included two infra-
red cameras focused on the rat at an oblique angle of approximately 45° each. The
kinematic data was sampled at 60 Hz.
To record differential iEMG, two stainless steel fine wire electrodes were in-
serted with an average interelectrode distance of 3.5 mm in the TA and GM mus-
cles. The reference electrode was placed under the skin at the back of the rat. The
iEMG data was amplified (A-M systems Model 1700, gain = 100) filtered (band
pass: filter 100 Hz -10 kHz, notch filter at 60 Hz), sampled (10 kHz, NI USB-
6259, National Instruments, USA) and saved in a PC using a custom LabView
routine.
To determine the stimulation pulse width and amplitude a strength duration curve
was first established by consecutively stimulating the four implanted LIFEs with
different pulse widths (30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 250 and 300 µs)
532 L.E. Lykholt et al.
while increasing the stimulation pulse amplitudes until a muscle twitch was seen. A
pulse width and 1.5 times the amplitude at rheobase were selected for stimulating
the fascicles (see Table 1).
To determine the stimulation frequencies the muscle contraction was visually
observed. The stimulation frequency was chosen such that it provided a fused
contraction, which was later confirmed from kinematic data.
The open-loop stimulation was applied to the fascicles innervating TA and GM
muscles alternately to produce a rhythmic movement. One hundred step cycles
were performed at each stimulation frequency (30, 40, 50, 60 and 70 Hz) followed
by a 15 min period of rest. Kinematic data from ankle movement and correspond-
ing iEMG were recorded simultaneously.
Fig. 1 Shows the range of movement of the ankle angle and iEMG amplitude envelope over
time for the GM muscle. The maximal range of ankle angle and iEMG amplitude envelope
was normalized to 100 %. A gradual decrease in the ankle angle and iEMG was observed
for the GM over time.
2.3
Data Analysis
The kinematic markers were identified using the Peak Motus software. The data
were digitized and filtered with a fourth-order band-pass Butterworth filter [6]. To
find the range of movement for both the GM and TA individually the maximum
range of the movement (maximal extension to maximal flexion) was calculated for
each cycle.
To obtain the iEMG amplitude envelope to investigate when the amplitude de-
creased, the iEMG amplitude was full wave rectified and low-pass filtered (3rd
order low-pass Butterworth filter with 0.5 Hz cut off frequency). The maximal
range of ankle angle and the iEMG amplitude envelope were normalized to 100 %
and an average of the normalized data obtained for the three rats. These averaged
data were used for the rest of the data analysis. Fatigue was defined as a decrease
Information on Ankle Angle from Intramuscular EMG Signals 533
Fig. 2 Shows the range of movement of the ankle angle and the iEMG amplitude envelope
over time for the TA. The maximum range for ankle angle and the iEMG amplitude
envelope were normalized to 100 %. A correlation between the two signals can be ob-
served. The decrease in the iEMG and ankle angle can be observed for the TA after 10-15 s
indicating presence of fatigue.
in the ankle angle and the iEMG amplitude envelope. A correlation coefficient
was calculated between the iEMG and ankle angle for both muscles for each fre-
quency.
3 Results
The movement produced by the GM muscle for all the stimulation frequencies
was in the range of 5-100 % of the maximum ankle angle, see Fig 1. The ankle
angle increased during the first step cycles (approximately 10 s). After this it de-
creased during the rest of the step cycles. The response after 60 Hz decreased
more rapidly than the other frequencies. The iEMG amplitude decreased conti-
nuously and rapidly for approximately the first 25 s (40-60 %). After this the rate
of the decrease was less for the rest of stimulation (30-40 %).
Table 2 Correlations coefficients between ankle angle and iEMG for GM and TA for the
different LIFE stimulations
GM TA
30 Hz 0.36 0.22
40 Hz 0.63 0.95
50 Hz 0.77 0.96
60 Hz 0.85 0.96
70 Hz 0.85 0.96
534 L.E. Lykholt et al.
The movement produced by the TA muscle for all the stimulation frequencies
was in the range of 65-100 % of the maximum ankle angle, see Fig 2. The ankle
angle increased for the first 10 s. After this there was a decrease from 10-25 s
thereafter a plateau was reached for the rest of the step cycles (65-85 %). The
iEMG amplitude was found to be stable for the first 10 s. After this the 50, 60 and
70 Hz response rapidly decreased from 10-30 s (30-60 %), while the 30 Hz re-
sponse produced no changes for the rest of the step cycles (80 %). The 40 Hz
response also had a different tendency where it decreased less compared to 50, 60
and 70 Hz response (85 %). After this a plateau was reached for the 50, 60 and 70
Hz response (30-50 %). Here the 40 Hz response still decreased and did not reach
a plateau (60 %). The ankle angle and the iEMG amplitude demonstrated similar
response.
To quantify the degree of correlation between the ankle angle and the iEMG,
the correlation coefficient between the ankle angle and the iEMG data were calcu-
lated (see Table 2). In the case of the GM muscle, it was observed that the higher
the stimulation frequency that was applied, the higher the correlation observed
(mean and standard deviation of 0.76 +/- 0.1 for 40 HZ – 70 Hz frequencies). In
the case of the TA muscle the same tendency was observed, i.e. coefficients for
the TA was high for the 40, 50, 60 and 70 Hz (0.95 +/- 0.1). This indicated that
there was a good correlation between the ankle angle and the iEMG amplitude
except when applying 30 Hz stimulation.
4 Discussion
In the current study the ankle angle was measured with kinematic data. This was
compared to the iEMG amplitude to examine the correlation between these. The
results showed that there was a correlation between the ankle angle and the iEMG.
The correlation was higher for the TA than the GM.
4.1
Comparison of results with Other Studies
Previous studies from E. A. T. De Laat et al., S. G. Boe et al., J. R. Potvin et al.
revealed that the relation between muscle force and amplitude is present and that
the variable used for this was root mean square amplitude. Here they were looking
at the linear force and root mean square amplitude using surface electrodes. [7],
[8], [9]
4.2
Methodological Considerations
With the use of an animal model instead of a human model the physiological in-
fluences are not the same. When a human walks normally there is a force applied
to the leg due to maintaining balance and standing upright. In this experiment, this
was not taken into account since the leg of the rats was hanging freely and no
external force was applied.
Information on Ankle Angle from Intramuscular EMG Signals 535
The markers were placed by visual inspection of the animal’s anatomical struc-
ture. Placement of the markers may therefore have varied slightly from animal to
animal. During the offline digitization of the kinematic video data it was possible
that some degree of error was present in the marker identification because of in-
distinct images. Especially the toe marker was difficult to distinguish in the video
during the extension phase of the movement.
During the stimulation the frequency was changed from 30-70 Hz. The stimula-
tion was done in that same order during all of the experiments. It is not possible to
judge if a particular stimulation frequency caused some cumulative influence on
the next stimulation sequence. This could be solved by randomization.
A factor that may have had an influence on the results is potentiation. This oc-
curs during continuous stimulation and also has a tendency to happen in fast
twitch fibers, and causes a positive staircase phenomenon [10]. This could likely
explain that some of the kinematic data had a tendency to not reach a maximum of
100 % just after the stimulation onset. Here the maximum movement range was
not reached until approximately 10 s after the onset of the stimulation.
5 Conclusion
In the present study it was investigated if information extracted from iEMG could
provide reliable information on a functional movement (i.e. the ankle angle) dur-
ing development of muscle fatigue. A higher degree of correlation was found
between iEMG and ankle angle when stimulation frequencies above 40 Hz were
applied to produce muscle contractions. Also, the results indicate that TA may be
a more reliable source of feedback than the GM since the TA had higher correla-
tion coefficient than the GM (GM: average of 0.76 +/- 0.01, TA: 0.96 +/- 0.01).
Further research should focus on developing an animal model where the corre-
lation coefficient would be higher. In a future perspective the improvement would
be beneficial for the clinical rehabilitation with the FES for subjects with spinal
cord injury.
Acknowledgment. We would like to thank Ph.D. student Ricardo Siu for manufacturing
the LIFEs. Supported in part by R01-EB008578 to Ranu Jung.
References
[1] Fairchild, M.D., Kim, S., Iarkov, A., Abbas, J.J., Jung, R.: Repetitive hindlimb
movement using intermittent adaptive neuromuscular electrical stimulation in an in-
complete spinal cord injury rodent model. Experimental Neurology 223, 623–633
(2010)
[2] Abbas, J.J., Triolo, R.J.: Experimental Evaluation of an Adaptive Feedforward Con-
troller for Use in Functional Neuromuscular Stimulation System. IEEE Transl. J. 5,
12–22 (1997)
536 L.E. Lykholt et al.
[3] Kim, S., Fairchild, M.D., Iarkov (Yarkov), A., Abbas, J.J., Jung, R.: Adaptive Control
of movement for Neuromuscular Stimulation-Assisted Therapy in a Rodents Model.
IEEE Transl. J 56, 452–461 (2009)
[4] Rose, J., Gamble, J.G.: Human Walking, 3rd edn., ch. 3. Lipincott Willams & Wil-
kins, Philadelphia (2006)
[5] Kamavuakoa, E.N., Rosenvanga, J.C., Bøga, M.F., Smidstrupa, A., Erkocevica, E.,
Niemeiera, M.J., Jensen, W., Farinaa, D.: Influence of the feature space on the estima-
tion of hand grasping force from intramuscular EMG. Biomedical Signal Processing
and Control 8, 1–5 (2013)
[6] Thota, A.K., Carlson, S., Knapp, E., Thompson, B., Jung, R.: Neuromechanical Con-
trol of Locomotion. Journal of Neurotrauma 24(2), 442–465 (2005)
[7] De Laat, E.A.T., Visser, C.P.J., Coene, L.N.J.E.M., Pahlplatz, P.V.M., Tavy, D.L.J.:
Nerve Lesion in Primary Shoulder Dislocation and Humeral Neck Fractures. J. Bone
Joint Surg. 76, 381–383 (1994)
[8] Boe, S.G., Rice, C.L., Doherty, T.J.: Estimation Contraction Level Using Root Mean
Square Amplitude in Control Subjects and Patients With Neuromuscular Disorders.
Arch. Phys. Med. Rehabil., 711–718 (2008)
[9] Potvin, J.R.: Effects of muscle kinematics in surface EMG amplitude and frequency
dur-ing fatiguing dynamic contractions. Department of Human Biology and Nutri-
tional Sciences, 144–151 (1997)
[10] Krarup, C.: Enhancement and Diminution of Mechanical Tension Evoked by Stair-
case and by the Teanus in Rat Muscle. J. Physiol. 311, 355–372 (1981)