Eur J Appl Physiol (2006) 97: 737–744

DOI 10.1007/s00421-006-0232-7

OR I G I NA L A RT I C L E

Ashraf S. Gorgey · Edward Mahoney · Tracee Kendall

Gary A. Dudley

Effects of neuromuscular electrical stimulation parameters

on specific tension

Accepted: 10 May 2006 / Published online: 4 July 2006

© Springer-Verlag 2006

Abstract This study examined the eVects of altering increased torque without aVecting the activated CSA.
surface neuromuscular electrical stimulation (SNMES) Clinicians who administer SNMES should be aware of
parameters on the speciWc tension of the quadriceps fem- the magnitude of adaptations to a given amplitude, pulse
oris muscle. Seven able-bodied subjects had magnetic duration, and frequency.
resonance images taken of both thighs prior to and
immediately after four SNMES protocols to determine Keywords Electrical stimulation · Amplitude ·
the activated muscle cross-sectional area (CSA). The Frequency · Pulse duration · SpeciWc tension
four protocols were: (1) research (RES, 100 Hz, 450
s,
and amplitude set to evoke 75% of maximal voluntary
isometric torque, MVIT); (2) pulse duration (PD, Introduction
100 Hz, 150
s, same current as in RES); (3) frequency
(FREQ, 25 Hz, 450
s, and same current as in RES); (4) Surface neuromuscular electrical stimulation (SNMES)
amplitude (AMP, 100 Hz, 450
s, and current set to is frequently used in rehabilitation, despite insuYcient
evoke the average of the initial torques of PD and knowledge of the impact of its parameters on force
FREQ, 45 § 9% of MVIT). Reducing the amplitude of development in relation to the stimulated muscle. It has
the current from 75 to 45% of MVIT did not alter spe- been known for some time that the amplitude of current
ciWc tension, 25 § 8 N/cm2, suggesting that the ampli- can aVect the evoked torque; within limits, the higher the
tude probably aVects torque and the area of activated amplitude the greater the torque (Adams et al. 1993).
muscle proportionally. Shortening the pulse duration EVorts to optimize SNMES amplitude in order to evoke
from 450 to 150
s caused speciWc tension to drop from maximal torque have received minimal attention because
25 § 6 to 20 § 6 N/cm2 (P < 0.05), indicating that pulse the magnitude of the amplitude is generally limited by
duration increased torque and the activated CSA dispro- the tolerance of a subject (Delitto et al. 1992). It has also
portionally. Alternatively, reducing the frequency from been shown that the frequency and the duration of the
100 to 25 Hz decreased speciWc tension from 25 § 6 to pulses during SNMES aVect torque; within limits, the
17 § 4 N/cm2 (P < 0.05), suggesting that the frequency greater the frequency up to about 50 Hz and the greater
the pulse duration up to about 500
s, the greater the
evoked torque (Hultman et al. 1983; Lieber and Kelly
1993). Whether each of these parameters aVects the mass
of skeletal muscle that is activated has not been system-
A. S. Gorgey (&) atically investigated. It is thought nonetheless that the
Department of Physical Medicine and Rehabilitation,
University of Michigan, 325 E Eisenhower, amplitude of the current (mAMP) increases force by
Ann Arbor, MI 48108, USA stimulating more muscles when increased. In contrast,
E-mail: [email protected] increasing the frequency of individual pulses increases
Tel.: +1-734-6155922 force by the summation of the twitches of the same stim-
Fax: +1-734-2133827
ulated muscles. SpeciWcally, how increases in pulse dura-
A. S. Gorgey · E. Mahoney · T. Kendall · G. A. Dudley tion increase force is diYcult to speculate.
Department of Kinesiology, The University of Georgia, The relationship of torque to skeletal muscle cross-sec-
300 River Road, Ramsey Center, Athens, GA 30602, USA tional area (CSA) has often been referred to as speciWc
G. A. Dudley
tension (Edgerton et al. 1990). SpeciWc tension reXects the
Shepherd Center, Crawford Research Institute, relative contributions of neural and musculoskeletal
2020 Peachtree Road, NW Atlanta, GA 30309, USA adaptations to the loss or gain in strength with aging or
738

with exercise conditioning (Narici et al. 1992; Reeves the study. A timeline of events is presented in Fig. 1.
et al. 2004). Muscle CSA in able-bodied humans has been Written informed consent was obtained from each sub-
determined using a variety of techniques (Adams et al. ject after all the procedures, and associated risks with the
1993; Fukunaga et al. 1992, 1996; Trappe et al. 2001; study were explained. The institutional review board of
Wickiewicz et al. 1983, 1984). The non-invasive imaging The University of Georgia approved the experimental
techniques are appealing because of their accuracy and protocols of this study.
because they present minimal health risk, especially mag-
netic resonance images (MRIs). The spin–spin relaxation Familiarization session
time of MRI scans has been used to quantify the muscle
activated by SNMES (Adams et al. 1993; Bickel et al. The Wrst visit was performed at the Muscle Biology Lab-
2004; Hillegass and Dudley 1999) and to calculate physio- oratory at The University of Georgia. Each participant
logical CSA (PCSA) (Akima et al. 2000). During SNMES was asked to perform three trials of maximum voluntary
application, understanding how torque developed in rela- isometric torque (MVIT) on each thigh and the highest
tion to the activated area could help clinicians to under- torque was considered his/her MVIT. In addition, they
stand the magnitude of muscle adaptation to a given demonstrated their tolerance to SNMES with each thigh
parameter. Moreover, it could provide more knowledge (see below).
on how the anatomic Weld of the stimulated muscle
aVected by altering SNMES parameters. Maximum voluntary isometric torque
The SNMES amplitude of current, frequency, or
pulse duration aVects torque, but individual eVects on The subject sat on a custom-built chair, and the leg was
speciWc tension have not been examined. Manipulation strapped at knee joint angle of 60° below horizontal to a
of SNMES parameters and quantifying the activated rigid lever arm with an inelastic strap to ensure that the
muscle using MRI could be used to examine the eVect of knee extensors could perform only isometric actions
each parameter upon speciWc tension. We hypothesized (Adams et al. 1993; Hillegass and Dudley 1999). The
that: (1) The SNMES mAMP would not change speciWc lever arm was established by mounting a load cell (model
tension; increasing current from 45 to 75% of maximal 2000A; Rice Lake Weighing Systems, West Coleman
voluntary isometric torque (MVIT) recruits more muscle Street, Rice Lake, WI, USA), perpendicular to and 12 in.
tissue and evokes proportional increase in torque. (2) (33 cm) from the axis of rotation of the lever arm. Each
The increase in the frequency from 25 to 100 Hz would subject was asked to extend against the immobile lever
increase speciWc tension. (3) Finally, increasing pulse arm as quickly and hard as possible for 3 s. Three trials
duration from 150 to 450
s would increase speciWc ten- were performed and 1 min of rest was given between sub-
sion, as the relation between evoked torque and pulse sequent trials on each thigh. The participant was asked
duration is skewed up and to the left (Hultman et al. to repeat the trials if the diVerence in MVIT between two
1983). of the three trials was greater than 5%. A load cell that
interfaced with an MAC laptop was used to measure,
and record force to be analyzed at a later date. All force
measures were gravity compensated.
Methods
Dosing phase and tolerance to SNMES
Subjects
After determining the MVIT, all subjects were asked to
Seven subjects (one female, 28 § 4 years old, height determine their ability to tolerate SNMES. A Thera-
173 § 9 cm, and weight 68 § 9 kg, mean § SD) were touch 4.7 electric stimulation unit (Theramini model TM
recruited from the population of The University of 1; Rich-Mar Corporation, Inola, OK, USA) was used to
Georgia and the surrounding areas. Participants were determine the mAMP required to elicit 75% of MVIT on
healthy, and none had a history of knee or hip patholog- each thigh. A rectangular, symmetrical biphasic wave-
ical conditions. They were asked to refrain from per- form was used in the current study. The participants
forming any strenuous physical activity for 72 h before were asked to completely relax, and the mAMP was

Fig. 1 General experimental

design and timeline. Protocols Study day
Familiarization
are applied in a random order
session
and this diagram is only for
illustration. Immediately after 0800 0900 1000 1100 1200 1300 1400
each protocol, T2-MRI images
were taken to determine the acti- 7 days Resting PD AMP
vated skeletal muscle CSA MRI (Left mQF) (Left mQF)

RES FREQ
(Right mQF) (Right mQF)
 739

gradually increased, using 1-s actions at 100 Hz and assigned to either thigh. These protocols were: (1)
450
s pulses given at 1 min apart until achieving 75% of Research protocol (RES): 100 Hz, 450
s pulses, and
their MVIT. Three trials of increasing amplitude were mAMP set to evoke 75% of MVIT. (2) Pulse duration
required to evoke a torque of 75% of MVIT. After 2– protocol (PD): 100 Hz, 150
s, the same current as in
3 min of rest, an SNMES protocol was delivered to RES. (3) Frequency protocol (FREQ): 25 Hz, 450
s,
evoke 75% of MVIT at 100 Hz and 450
s biphasic and the same current as in RES. (4) Amplitude protocol
pulses to either thigh for 3 min. Failure to tolerate this (AMP): 100 Hz, 450
s, and amplitude set to evoke the
protocol resulted in exclusion from the study. Stimula- average of the initial torques of PD and FREQ, 45 § 9%
tion at 75% of MVIT was selected because this mAMP of MVIT. The four protocols were applied for 3 min at
could activate more than 50% of the muscle and could be 3 s on and 3 s oV. It should be noted that as a result of
tolerated by most participants (Adams et al. 1993). Four the diVerence in the MVIT between right and left mQFs,
protocols of electrical stimulation were randomly deliv- the mAMP required to evoke 75% of MVIT was slightly
ered to both quadriceps femoris muscle (mQF) (see but not statistically diVerent in RES, PD, and FREQ (see
below). Recovery of the force was examined by asking Results). A period of 2 h was allowed to separate any
each participant to perform one MVIT contraction two protocols delivered on the same mQF to ensure the
before starting a new protocol on the same thigh. full recovery of the force.
Approximately 2 h was suYcient between the two proto-
cols on the same mQF for the recovery of force in all Torque analysis
participants.
The Wrst train of electrical stimulation in each protocol
Study day was used to express the torque value to avoid the inXu-
ence of muscle fatigue during SNMES on our measure-
All subjects arrived at the MRI facility at 8.00 a.m. They ments (Bickel et al. 2004). To analyze the Wrst train, a
were asked to rest for 30 min with no physical activity to 250–750 ms window was used that began when torque
reduce the error that could result from the Xuid shift on rose above baseline (Fig. 2). This window covered the
T2-signal intensity (Conley et al. 1996). Immediately after plateau phase of torque to establish peak torque in each
this resting period, a resting T2-weighted MRI scan was protocol (Dudley et al. 1990; Harris and Dudley 1994).
taken of each subject. They were then asked to perform
three MVITs. After reassessing the MVIT, the SNMES Magnetic resonance imaging
protocols were randomly administered to either thigh
with T2-MRI captured immediately after each protocol. Standard spin-echo images of the thighs were collected
Two protocols were administered to the right and the using a 1.5 T super-conducting magnet (Signa, General
other two to the left thigh. After scanning, each partici- Electric, Milwaukee, WI, USA) as done before (Adams
pant was asked to exit the scanner and to lie down for et al. 1993; Akima et al. 2002; Bickel et al. 2004). Subject
60 min before the second protocol was applied to the was positioned supine within the magnet using the whole
opposite thigh. Before staring a new protocol on the body coil. MRI scans were taken of both thighs prior to
same thigh, a 1-s contraction of a current that evoked the SNMES. Following each protocol, images were
75% of MVIT was administered to ensure the recovery immediately (3 § 0.2 min) collected again. At least
of the force to the starting force. 60 min separated any two scanning sessions to allow the
return to the pre-SNMES state. Each subject was asked
SNMES protocols

All stimulation protocols were performed on one day. 250

SNMES was applied to the mQF via surface electrodes
to induce intermittent, isometric actions. Two 8 £ 10 cm2 200 RES
(Uni-Patch, 1313 West Grant Boulevard, Wabasha, MI,
Torque (Nm)

150 PD
USA) adhesive carbon electrodes were placed on the skin
FREQ
over mQF. One electrode was placed 2–3 cm above the 100
superior aspect of the patella over m. vastus medialis, AMP
and the other lateral to and 30 cm above the patella over 50
m. vastus lateralis as done previously (Adams et al. 1993;
Bickel et al.2004). The edges of the electrodes were 0
1 2 3 4 5 6 7
marked to ensure exact location in the following proto- Time (sec)
col. The current was increased once every 1 min and
three trials were required, to re-establish the amplitude Fig. 2 Typical example of 3-s initial contractions of electrically
of current required to evoke 75% of MVIT by using 1 s, evoked torque in the four stimulation protocols. Five hundred-mil-
lisecond window was placed on each trace to identify the location of
100 Hz trains of 450
s biphasic, square pulses. Once the torque measurement. The traces for PD, FREQ, and AMP were
mAMP was determined for both left and right thighs, it slightly displaced from their original starting point for the purpose
was then recorded and the four protocols were randomly of clarity
740

to exit the MRI scanner and to rest completely until the ent SNMES protocols by the patellar tendon moment
delivery of next protocol. Transaxial T2 images (TR/TE arm length of the mQF. The moment arm was derived
= 2,000/30, 60 ms, 1 cm thick, 1 cm apart, 40 cm Weld of from cadavers and knee joints of able bodies, 0.04 m, at
view, a 256 £ 256 matrix size, and one NEX) were the same knee joint angle used in our study (Baltzopou-
acquired at 4 min and 40 s. An ink mark drawn proximal los 1995; Wickiewicz et al. 1983, 1984). Because the
to the upper border of patella by 6 in. allowed for similar moment arm increases with muscle contraction and
positioning in the magnet bore after each stimulation moment arm from cadavers shrinks after death, the per-
protocol. Fourteen to 18 slices for each subject were ana- centage increase in the moment arm during ankle exten-
lyzed beginning with the Wrst proximal slice visualizing sion, 18%, was used in the calculation of the moment
the four heads of the mQF group and continued distally arm length for this lower extremity, extensor muscle
until the slice just before the top of the patella. Images group (Maganaris et al. 2001). A moment arm length of
were analyzed using NIH Image 1.62 available on the 0.047 m was used to correct torque to force. Dividing the
web at http://www.rsb.info.nih.gov/nih-image/down- torque by the moment arm provided the force applied by
load.html. Test–retest reliability for repeat measures of the knee extensors to the tendon (Reeves et al. 2004). The
the knee extensor muscle group CSA by slice was previ- resulting force was divided by the PCSA of activated skel-
ously documented to be r > 0.95 (P < 0.05) (Adams etal muscle. The results were speciWc tension in N/cm2.
et al. 1993). The PCSA of mQF was calculated as follows.
PCSA = [muscle volume £ COS (pennation angle)]/
Calculation of activated CSA Wber length (Edgerton et al. 1990; Fukunaga et al. 1992;
Akima et al. 2000). Muscle volume (cm3) was calculated
MRIs were transferred to a computer to measure skeletal by multiplying the activated CSA (cm2) in each
muscle CSA before SNMES and muscle that was acti- slice £ 2 cm (slice thickness + interslice space). The slices
vated by SNMES. A region of interest was deWned by were then added together to calculate the volume of acti-
tracing the outline of the mQF. Activated muscle was vated tissue. A pennation angle of 15.6° was acquired
assessed as pixels with an elevated T2 in the post- from previously published data, which is the pennation
SNMES images. Pixels with a T2 between 20 and 35 ms angle of m. vastus lateralis at 60° of knee joint below
represented muscle in the pre-SNMES images. The mean horizontal during isometric contraction (Fukunaga et al.
and SD T2 of these pixels in each pre-SNMES image was 1997; Ichinose et al. 1997). Vastus lateralis muscle can be
calculated. Pixels in matching post-SNMES images with used as surrogate of mQF (Trappe et al. 2001; Reeves
a T2 greater than the mean plus 1 SD of the T2 of pre- et al. 2004). Fiber length was identiWed by the following
electrically stimulated images were considered elevated equation after measuring muscle length (Wber length/
or to show a contrast shift in the post-SNMES images. muscle length = 0.23) (Wickiewicz et al. 1983).
The CSA of these pixels was determined. Values were Muscle length was measured as the distance between
averaged over slices to determine the average anatomical the most proximal and the most distal points in which
CSA (ACSA) of skeletal muscle that was stimulated as the muscle was visible in three scout views captured with
done before (Adams et al. 1993; Akima et al. 2002; Bickel MRI (Fukunaga et al. 1992; Narici et al. 1992). Adjust-
et al. 2004; Hillegass and Dudley 1999). ments were made to account for Wber length during iso-
metric actions at diVerent percentages of MVIT
SpeciWc tension (Ichinose et al. 1997). The resting length was multiplied
by the percentage changes in Wber length when torque
SpeciWc tension was calculated by dividing the torque was 75, 30, 60, and 45% of MVIT for RES, PD, FREQ,
(Nm) obtained for the Wrst contraction of the four diVer- and AMP, respectively (Table 1).

Table 1 Architectural characteristics of activated mQF after stimulation

Total RES PD FREQ AMP

Muscle length (cm) 31 § 0.3 – – – –

Fiber length (cm) 7.1 § 0.3* 5.6 § 0.3** 6 § 0.3*,** 5.7 § 0.3** 5.8 § 0.3**
ACSA (cm2) 66.4 § 15* 30 § 7** 18 § 10*,** 36 § 8** 21.5 § 12*,**
PCSA (cm2) 253 § 49* 145 § 33** 83 § 47*,** 171 § 35** 101 § 55*,**
Muscle volume (L) 1.9 § 0.4* 0.9 § 0.2** 0.5 § 0.3*,** 1 § 0.25** 0.6 § 0.3*,**
SpeciWc tension-PCSA 20 § 3* 25 § 6** 20 § 7* 17 § 4* 26 § 8**
(N/cm2)
Muscle length was determined by MRI, and Wber length pre-SNMES was calculated by the ratio previously reported. Necessary adjustments
were made to Wber length at each protocol based on the percent of MVIT achieved. PCSA, muscle volume, and speciWc tension for total and
for the activated mQF (SNMES - torque/activated PCSA)
*DiVerent from RES (P < 0.05)
**DiVerent from total (P < 0.05)
 741

Statistics 250
RES
Statistical analyses were run using SPSS (V 10.0). * PD
200
*

Torque (Nm)
Repeated-measures ANOVAs were used to examine the * FREQ
eVects of SNMES on torque production, activated area, 150 AMP
and speciWc tension. Planned contrasts were used to test
the diVerences between any two protocols. Alpha level 100
was adjusted for pair-wise comparisons using the Bon-
ferroni correction. Simple linear regression was used to 50
examine the relationship between selected variables. Sta-
tistical signiWcance was set at a level of P < 0.05. Values 0
are presented as means § SD. SNMES protocols

Fig. 3 Initial torque (Nm) of four protocols of SNMES. *RES was

greater than PD, FREQ, and AMP (P < 0.0001)
Results

The MVIT of the right and left mQFs was 242 § 66 and 63 § 8, and 38 § 24% for RES, PD, FREQ, and AMP,
226 § 60 Nm, respectively. The mAMP delivered for respectively (Fig. 4). The activated ACSA and PCSA in
RES, PD, FREQ, and AMP was 74 § 18, 76 § 16, RES were greater than PD and AMP (P < 0.001) but
72 § 18, and 56 § 13 mA, respectively. The mAMP not diVerent from FREQ (P = 0.1). In addition, the acti-
required to evoke 75% of MVIT was not diVerent among vated ACSA and PCSA in FREQ were also greater than
RES, PD, and FREQ (P > 0.05), but was lower in AMP PD and AMP (P < 0.01). PCSA was about four times
(P < 0.05). Each protocol initially evoked 74 § 3, ACSA (P < 0.0001). Following stimulation, Wber length
31 § 12, 60 § 8, and 45 § 9% of MVIT, respectively decreased (P < 0.01) from the resting length by 21, 15.5,
(P < 0.05). Initial torque generated by RES was greater 20, and 20% after RES, PD, FQ, and AMP, respectively
than that evoked by PD (55%, P < 0.001), FREQ (18%, (Table 1).
P < 0.01), and AMP (34%, P < 0.001) (Fig. 3). The torque (Nm) was strongly correlated to the acti-
The average CSA of left (65 § 15 cm2) and right vated PCSA (P < 0.008), particularly when the mAMP
(66 § 14 cm2) mQFs was not diVerent before the applica- and pulse duration of SNMES were individually
tion of any of the four protocols (P = 0.1). The four increased. This relation was not evident (P = 0.5) when
stimulation protocols decreased the CSA involved in the the frequency of the pulses alone was altered (Fig. 5).
calculation of the speciWc tension (P < 0.05). Following SpeciWc tension decreased when pulse duration was
stimulation, the activated CSA was 54 § 14, 32 § 29, shortened to 150
s and frequency was lowered to 25 Hz

Fig. 4 Representative binary T2

maps of mQF from one subject
at rest (before stimulation) and
after the four SNMES proto-
cols. Note RES and AMP show
left thigh, whereas PD and
FREQ show the right thigh.
White regions indicate elevated
T2 or activated muscle

Before SNMES of left knee extensor RES AMP

Before SNMES of right knee extensor FREQ PD

Resting T2 Activated T2
742

250 duration from 450 to 150
s decreased torque, activated

area, and speciWc tension. Finally, reducing frequency
200 from 100 to 25 Hz did not inXuence the activated area,
but decreased torque and speciWc tension.
Torque (Nm)

150 The present Wndings show that increasing the mAMP

A that evoked 45 or 75% of MVIT did not change speciWc
100 tension. The relationship between torque and activated
RES PD CSA has been previously conWrmed (Adams et al. 1993;
50 B Bickel et al. 2004). In short, if the mAMP is increased, a
FREQ AMP
greater torque results. The mAMP and torque are line-
0 arly related; torque increases because more muscle is
0 50 100 150 200 250 activated. In the RES protocol, amplitude that evokes
activated PCSA (cm2) 75% of MVIT was chosen to increase the CSA activated.
Although it has been shown that it is rare to stimulate
Fig. 5 Torque (Nm) vs. activated PCSA (cm2), the results of PD,
FREQ, and AMP were mainly compared to RES results. A At diVer-
mQF at an amplitude greater than 55% of MVIT (Lieber
ent amplitudes, 45 vs. 75% MVIT, torque (Nm) = 0.78 (activated and Kelly 1991), others have shown it is possible to stim-
PCSA) + 46.86 (r2 = 0.49, P < 0.008). B At diVerent pulse dura- ulate grater than 60 or 90% of MVIT (Adams et al. 1993;
tions, 150 vs. 450
s, torque (Nm) = 0.94 (activated PCSA) + 15.35 Dudley et al. 1990; Harris and Dudley 1994; Westing
(r2 = 0.63, P < 0.001). C At diVerent frequencies, 25 vs. 100 Hz et al. 1990). Adams et al. showed that increasing the
(r2 = 0.03, P = 0.5), regression line C was discarded for clarity
mAMP required to evoke 25–75% of MVIT increases the
activated CSA from 18 to 54% (Adams et al. 1993). Part
of the diVerence could be explained by the selected pulse
but not when amplitude decreased to 45% of MVIT duration. Lieber and Kelly (1991) stimulated at a shorter
(Table 1 and Fig. 6). SpeciWc tension for RES was greater pulse duration (250
s) compared to the longer pulse
than that for PD and FREQ by 24 and 31%, respectively duration (450
s), which was used in the current study.
(P < 0.05), but not diVerent from that for AMP Therefore, it is unlikely to achieve stimulation greater
(P > 0.05). than 55% with a short pulse duration.
Increasing the pulse duration from 150 to 450
s
increased torque by 55% and the amount of muscle acti-
Discussion vated by 40%. The diVerences in the relative increases
could be explained if longer pulse durations activated
This study investigated the eVect of altering SNMES mainly fast-twitch motor units. SNMES at a constant
parameters on speciWc tension. The RES protocol was pulse duration recruits muscle Wber type randomly
used as a reference to the other three protocols (PD, (Gregory and Bickel 2005; KnaXitz et al. 1990). Further-
FREQ, and AMP), in which only one parameter was more, an increase in pulse duration increases the evoked
altered from the parameters of the RES. There were sev- torque (Hultman et al. 1983). Moreover, fast-twitch
eral major Wndings. Most importantly, the highest spe- Wbers generate greater force than slow twitch Wbers after
ciWc tension was occurred when the pulse duration and controlling the size of the Wbers (Bodine et al. 1987). If
the frequency of pulses were increased. Decreasing the pulse duration resulted in a proportional increase in
mAMP from 75 to 45% of MVIT did not reduce speciWc torque and activated area, speciWc tension would not
tension, but as is generally known decreased the torque have been diVerent between RES and PD. However, the
and the activated area. Moreover, shortening pulse results showed a greater speciWc tension after controlling
the activated area between RES and PD. Therefore, acti-
vation of fast-twitch Wbers with longer pulse durations
35 may explain the results of this study. The present Wndings
* show that increasing the pulse duration had the most
30
Specific tesnion (N/cm2)

impact on torque compared to increasing the amplitude

25 * of current and the frequency. Although this Wnding is
20
RES limited to the range of SNMES parameters used in this
PD
study, it suggests the importance of pulse duration,
FREQ
15 which has often been overlooked in research and rehabil-
AMP
10 itation (Delitto et al. 1992; Maganaris et al. 2001). One
previous study that used SNMES to determine speciWc
5
tension with pulse duration of 100
s showed a value of
0 15 N/cm2 (Maganaris et al. 2001). If at least 450
s had
SNMES protocols been used, then the speciWc tension would have been
Fig. 6 SpeciWc tension (N/cm2) of mQF after four protocols of about one-third larger.
SNMES. SpeciWc tension for the activated mQF (SNMES-torque/ Stimulation at 75% MVIT and longer pulse duration
activated PCSA). *RES was greater than PD and FREQ (P < 0.05) are strong and sometimes painful, which may have
 743

resulted in evoked or reXex contraction (Collins et al. occurs over time in voluntary eVorts, electrical stimulation
2002). For these reasons, a pulse duration of 450
s was leads to increased fatigue. Thus, if the activated area repre-
selected to avoid recruitment of sensory Wbers and to sents the motor units that were stimulated/recruited
minimize discomfort. Sensory nerve Wbers are recruited regardless of fatigue, these motor units were still on from
with pulse duration greater than 500
s (Panizza et al. the start to the end of the protocol. Stimulation was
1998). We also encouraged our participants to relax dur- repeated to increase T2 to measure the activated area.
ing diVerent protocols to ensure that no tensing up SpeciWc tension has been studied in animals as well as
occurred that could have aVected our values (Collins in humans (Bodine et al. 1987; Edgerton et al. 1990;
et al. 2002). Some researchers have suggested that co- Narici et al. 1992; Reeves et al. 2004). To the best of our
contraction of the antagonist muscle is important factor knowledge, this is the Wrst study that has accounted for
to consider (Reeves et al. 2004). However, others have the activated muscle using T2-MRI. The importance of
shown that shape of the force–velocity curve is not diVer- considering the recruited muscle involved in exercise
ent between individuals with spinal cord injury (SCI) and when referring to variables such as oxygen uptake during
controls (Triolo et al. 1987). This means that co-contrac- physical activity is well known (Armstrong et al. 1992).
tion has a minimal eVect on the torque of mQF because Our data support the robustness of using MRI to diVer-
SCI rules out co-contraction. entiate in the activated area with diVerent stimulation
The eVect of the frequency of the pulses on torque protocols. The use of T2-MRI provides anatomic and
development has often been discussed in both animal quantitative distinction between activated and non-acti-
and human studies (Bridges et al. 1991; Lieber and Kelly vated region immediately post-exercise. In this study,
1993). There is an agreement that frequency increases multiple slices (n = 14–18) were analyzed because a sin-
torque development by the summation of twitches. The gle slice has been shown to underestimate the true CSA
present Wndings support this. Frequency operates via of the muscle (Edgerton and Roy 1991). Also the current
inXuencing the Ca2+ kinetics; with lower frequency, less spreads randomly to the whole mQF, which would be
Ca2+ is being released from the sarcoplasmic reticulum, hard to account for the total activated area by using a
which results in fewer cross-bridges attachments and single slice (Adams et al. 1993).
thus less force production (Fitts 1994). In this study, a In this study, the pennation angle of 15.6° was not mea-
frequency of 100 Hz was used to ensure maximum ten- sured, but derived from previously published papers
sion developed in the stimulated area (Maganaris et al. (Fukunaga et al. 1997; Ichinose et al. 1997). The error
2001). This frequency is unlikely to be used in a training associated with cos
is minimal in terms of its inXuence in
paradigm with the goal to minimize fatigue and enhance the calculation of PCSA. If a pennation angle of 14.1° or
functional performance during stimulation. The Wring 16° had been chosen instead of 15.6° (Reeves et al. 2004),
rate of mQF is most commonly »15–30 Hz (Conwit the result would always be 0.96, indicating that about 96%
et al. 1998); for this reason, a stimulation frequency of of the force exerted by the Wbers will be transmitted to the
25 Hz was chosen for the FREQ protocol. The increase tendon. Furthermore, a moment arm length of 0.047 m
in speciWc tension from 17 to 25 N/cm2 found in this was used to correct torque to force. Data from cadavers
study was not explained by increasing activated area, used to calculate speciWc tension showed that the moment
suggesting that the mechanism for increasing torque is arm of knee joint at 60° Xexion is 0.04 m. This was sup-
the summation of twitches. ported later by X-ray study that showed the maximum
In this study, large size (8 £ 10 cm2) and adhesive car- moment arm of knee joint is »0.04 m (Baltzopoulos 1995).
bon-Xexible electrodes were used to ensure that maximum However, none of these studies accounts for the eVect of
tension had developed in the activated area. This type of contraction on the moment arm. During PCSA calcula-
electrode has been shown to produce the maximum abso- tion, the eVect of contraction on the moment arm length,
lute and relative torque during stimulation of mQF (Lie- Wber length, and pennation angle was taken into account
ber and Kelly 1991). A duty cycle of 50% (3 on/3 oV) was as has been previously done (Fukunaga et al. 1997; Ichi-
chosen which is enough to enhance metabolic intensity in nose et al. 1997; Maganaris et al. 2001). It should be noted
the stimulated muscle, as it has been shown that T2 is that if the resting moment arm and the resting Wber length
tightly associated with the metabolic intensity of exercise had been used in this study, the speciWc tension of RES
(Fisher et al. 1990). Additionally, a 50% duty cycle was would have been 37 N/cm2. The increase in the moment
previously shown to produce greater average torque than arm length has been previously observed during contrac-
70% (Lieber and Kelly 1993). Fatigue commonly occurs as tion and attributed to muscle thickening and displace-
result of SNMES (Bickel et al. 2004). Therefore, fatigue ments of the center of rotation of the knee joint
output of each protocol could have inXuenced the speciWc (Maganaris et al. 1998). Although errors may have been
tension. However, this is unlikely given that the recruit- introduced because of the use of these parameters to cal-
ment pattern of SNMES did not change over the course of culate PCSA, the relative outcomes of the four protocols
repeated contractions. Electrical stimulation is thought to would still be the same because these parameters were
activate motor units within skeletal muscle in a spatially standardized across them.
Wxed manner (Adams et al. 1993), i.e., the same motor In summary, SNMES pulse duration and the fre-
units are repeatedly activated during exercise. By prevent- quency of the pulses augment speciWc tension when
ing the rotation of motor unit recruitment that often increased. SpeciWc tension was not altered by the mAMP,
744

but a subject’s tolerance can limit SNMES when the cur- Fisher MJ, Meyer RA, Adans GR, Foley GM, Potchen EJ (1990)
rent is increased to activate as much muscle as possible. Direct relationships between proton T2 and exercise intensity in
skeletal muscle MR images. Invest Radiol 25:480–485
The Wndings in this study are important for diVerent rea- Fitts RH (1994) Cellular mechanisms of muscle fatigue. Physiol Rev
sons. First, it contributes to our basic understanding of 74:49–94
how muscle generates force during electrical stimulation. Fukunaga T, Roy RR, Shellock FG, Hodgson JA, Day MK, Lee PL,
Second, the results should be applied in rehabilitation as Kwong-Fu H, Edgerton VR (1992) Physiological cross-sectional
area of human leg muscles based on magnetic resonance imag-
they show how changes in SNMES parameters inXuence ing. J Orthop Res 10:928–934
muscle recruitment. Finally, clinicians applying SNMES Fukunaga T, Roy RR, Shellock FG, Hodgson JA, Edgerton VR
should appreciate that the success of rehabilitation will (1996) SpeciWc tension of human plantar Xexors and dorsiXexors.
depend on the magnitude of the adaptations to a given J Appl Physiol 80:158–165
frequency and pulse duration not just the mAMP. Fukunaga T, Ichinose Y, Ito M, Kawakami Y, Fukashiro S (1997)
Determination of fascicle length and pennation in a contracting
human muscle in vivo. J Appl Physiol 82:354–358
Acknowledgments The authors would like to thank all the subjects Gregory CM, Bickel CS (2005) Recruitment patterns in human skel-
participated in this study. We also appreciate Chris Black and Car- etal muscle during electrical stimulation. Phys Ther 85:358–364
olyn Sharp for their technical expertise and Drs. Kevin McCully and
Harris RT, Dudley GA (1994) Factors limiting force during slow,
Jill Slade for their helpful comments during the preparation of the
shortening actions of the quadriceps femoris muscle group in vi-
manuscript. This study was supported by NIH grants to G.A.D. vo. Acta Physiol Scand 152:63–71
(Grants No.: HD39679 and HD39676S2). Hillegass EA, Dudley GA (1999) Surface electrical stimulation of
skeletal muscle after spinal cord injury. Spinal Cord 37:251–257
Hultman E, Sjoholm H, Jaderholm-Ek I, Krynicki J (1983) Evalua-
tion of methods for electrical stimulation of human skeletal mus-
References cle in situ. PXugers Arch 398:139–141
Ichinose Y, Kawakami Y, Ito M, Fukunaga T (1997) Estimation of
Adams GR, Harris RT, Woodard D, Dudley GA (1993) Mapping of active force–length characteristics of human vastus lateralis
electrical muscle stimulation using MRI. J Appl Physiol 74:532–537 muscle. Acta Anat (Basel) 159:78–83
Akima H, Kuno S, Takahashi H, Fukunaga T, Katsuta S (2000) The KnaXitz M, Merletti R, De Luca CJ (1990) Inference of motor unit
use of magnetic resonance images to investigate the inXuence of recruitment order in voluntary and electrically elicited contrac-
recruitment on the relationship between torque and cross-sec- tions. J Appl Physiol 68:1657–1667
tional area in human muscle. Eur J Appl Physiol 83:475–480 Lieber RL, Kelly MJ (1991) Factors inXuencing quadriceps femoris
Akima H, Foley JM, Prior BM, Dudley GA, Meyer RA (2002) Va- muscle torque using transcutaneous neuromuscular electrical
stus lateralis fatigue alters recruitment of musculus quadriceps stimulation. Phys Ther 71:715–721
femoris in humans. J Appl Physiol 92:679–684 Lieber RL, Kelly MJ (1993) Torque history of electrically stimulated
Armstrong RB, Essen-Gustavsson B, Hoppeler H, Jones JH, Kayar human quadriceps: implications for stimulation therapy. J Ort-
SR, Laughlin MH, Lindholm A, Longworth KE, Taylor CR, hop Res 11:131–141
Weibel ER (1992) O2 delivery at VO2max and oxidative capacity Maganaris CN, Baltzopoulos V, Sargeant AJ (1998) Changes in
in muscles of standardbred horses. J Appl Physiol 73:2274–2282 Achilles tendon moment arm from rest to maximum isometric
Baltzopoulos V (1995) A videoXuoroscopy method for optical dis- plantarXexion: in vivo observations in man. J Physiol 510(Pt
tortion correction and measurement of knee-joint kinematics. 3):977–985
Clin Biomech (Bristol, Avon) 10:85–92 Maganaris CN, Baltzopoulos V, Ball D, Sargeant AJ (2001) In vivo
Bickel CS, Slade JM, Dudley GA (2004) Long-term spinal cord in- speciWc tension of human skeletal muscle. J Appl Physiol 90:865–
jury increases susceptibility to isometric contraction-induced 872
muscle injury. Eur J Appl Physiol 91:308–313 Narici MV, Landoni L, Minetti AE (1992) Assessment of human
Bodine SC, Roy RR, Eldred E, Edgerton VR (1987) Maximal force knee extensor muscles stress from in vivo physiological cross-
as a function of anatomical features of motor units in the cat tib- sectional area and strength measurements. Eur J Appl Physiol
ialis anterior. J Neurophysiol 57:1730–1745 Occup Physiol 65:438–444
Bridges CR Jr, Clark BJ III, Hammond RL, Stephenson LW (1991) Panizza M, Nilsson J, Roth BJ, Grill SE, Demirci M, Hallett M
Skeletal muscle bioenergetics during frequency-dependent fa- (1998) DiVerences between the time constant of sensory and mo-
tigue. Am J Physiol 260:C643–C651 tor peripheral nerve Wbers: further studies and considerations.
Collins DF, Burk D, Gandevia SC (2002) Sustained contractions Muscle Nerve 21(1):48–54
produced by plateau-like behavior in human motoneurones. J Reeves ND, Narici MV, Maganaris CN (2004) EVect of resistance
Physiol 1 538(Pt 1):289–301 training on skeletal muscle-speciWc force in elderly humans. J
Conley MS, Foley JM, Ploutz-Snyder LL, Meyer RA, Dudley GA Appl Physiol 96:885–892
(1996) EVect of acute head-down tilt on skeletal muscle cross- Trappe TA, Lindquist DM, Carrithers JA (2001) Muscle-speciWc
sectional area and proton transverse relaxation time. J Appl atrophy of the quadriceps femoris with aging. J Appl Physiol
Physiol 81(4):1572–1577 90:2070–2074
Conwit RA, Tracy B, Cowl A, McHugh M, Stashuk D, Brown WF, Triolo R, Robinson D, Gardner E, Betz R (1987) The eccentric
Metter EJ (1998) Firing rate analysis using decomposition-en- strength of electrically stimulated paralyzed muscle. IEEE Trans
hanced spike triggered averaging in the quadriceps femoris. Biomed Eng 9:651–652
Muscle Nerve 10:1338–1340 Westing SH, Seger JY, Thorstensson A (1990) EVects of electrical
Delitto A, Strube MJ, Shulman AD, Minor SD (1992) A study of stimulation on eccentric and concentric torque–velocity relation-
discomfort with electrical stimulation. Phys Ther 72:410–421 ships during knee extension in man. Acta Physiol Scand 140:17–
Dudley GA, Harris RT, Duvoisin MR, Hather BM, Buchanan P 22
(1990) EVect of voluntary vs. artiWcial activation on the relation- Wickiewicz TL, Roy RR, Powell PL, Edgerton VR (1983) Muscle
ship of muscle torque to speed. J Appl Physiol 69:2215–2221 architecture of the human lower limb. Clin Orthop Relat Res
Edgerton VR, Roy RR (1991) Regulation of skeletal muscle Wber 275–283
size, shape and function. J Biomech 24(Suppl 1):123–133 Wickiewicz TL, Roy RR, Powell PL, Perrine JJ, Edgerton VR (1984)
Edgerton VR, Apor P, Roy RR (1990) SpeciWc tension of human el- Muscle architecture and force–velocity relationships in humans.
bow Xexor muscles. Acta Physiol Hung 75:205–216 J Appl Physiol 57:435–443