J Appl Physiol 107: 1276 –1284, 2009.
First published July 16, 2009; doi:10.1152/japplphysiol.91598.2008.
In vivo intramuscular fascicle-aponeuroses dynamics of the human medial
gastrocnemius during plantarflexion and dorsiflexion of the foot
David D. Shin,1 John A. Hodgson,2 V. Reggie Edgerton,3 and Shantanu Sinha2
1
Biomedical Engineering, University of California, Los Angeles; 2Radiology, University of California, San Diego;
and 3Physiological Sciences, University of California, Los Angeles, California
Submitted 15 December 2008; accepted in final form 15 July 2009
gear ratio; pennation angle; fascicle length; fascicle strain; cine
velocity encoded phase-contrast magnetic resonance imaging
in ultrasound and magnetic resonance
(MR) imaging (MRI) have improved our abilities to investigate
the mechanisms involved in the transfer of muscle fiber force
and length changes across joints, not least by providing a
means to make direct observations of muscle and tendon
deformations in conscious human subjects activating muscles
voluntarily. MRI offers an advantage of being able to sample
large areas of muscle simultaneously and the opportunity to
track the movement of any selected tissue location, even when
no anatomic markers exist (5, 19, 23). This facilitates simultaneous sampling of multiple regions of a contracting muscle at
a finer resolution than has previously been attempted. While
RECENT DEVELOPMENTS
Address for reprint requests and other correspondence: S. Sinha, Muscle
Imaging and Modeling Laboratory, Dept. of Radiology, Univ. of CaliforniaSan Diego School of Medicine, RIL, 3510 Dunhill St., San Diego, CA
92121-0852 (e-mail: [email protected]).
1276
the conventional view of muscle contraction (17, 18, 25) is
based on a simple geometry and isotropic material properties,
the findings of Papas et al. (19) demonstrated that there are
regional differences in the pattern of fiber length changes.
These findings are consistent with theoretical models of contracting muscle suggesting regional differences in muscle fiber
shortening (10) and even strain differences along the muscle
fibers themselves (17, 18). Subsequent attempts to simulate
nonuniform strains suggested that structural variants such as
differences in fiber length may be responsible (3). However,
regional differences in muscle composition may not be the only
contributors to regional differences in strain. Several models of
muscle have demonstrated regional differences in strain, even
with uniform material properties across the regions of differing
strain (10, 17, 26). These included differences in overall fiber
length change between fibers in the midregion of a muscle
compared with fibers located closer to the tendons of origin and
insertion (10). Interestingly, this work (10) also predicted that
fibers at the ends of a muscle may undergo lengthening in an
isometric contraction. Perhaps more intriguing, several models
have suggested that there may be mechanical justification for
questioning the common assumption that fibers strain uniformly along their length. Figures in some reports (17, 26) of
muscle models using a homogenous muscle material showed a
range of strain values along what may be assumed to be muscle
fascicles. In some cases, the strain values ranged from positive
to negative along the orientation of a fiber, resulting in the
unexpected prediction that one end of an isometrically contracting fiber may shorten as the other end lengthens (26). In
another case, intramuscular pressure differences showed a
similar inhomogeneity along what may be presumed to be the
fiber orientation, again suggesting that the stress distribution
along the fiber may be inhomogeneous (11). It thus seems
reasonable to question the common assumption that muscle
fibers shorten homogeneously along their length and instead
hypothesize that fibers may strain heterogeneously. Prior modeling efforts have also suggested that the greatest heterogeneity
along fibers may occur in fibers close to the musculotendinous
junctions, i.e., the regions where the muscle ends and tendon
begins (10, 26). Furthermore, the greatest stresses and strains
appear to occur close to the musculotendinous junction, leading to a prediction that the location of the greatest strain in
muscle fibers will change from an area close to the origin of the
muscle fibers in the proximal region of the muscle to an area
close to the insertion of the fibers in the distal region of the
muscle.
To test these predictions, we measured strain along individual muscle fascicles at different proximodistal locations along
the human medial gastrocnemius (MG) muscle as the ratio of
fascicular displacement at the site of insertion to that of origin.
8750-7587/09 $8.00 Copyright © 2009 the American Physiological Society
http://www. jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
Shin DD, Hodgson JA, Edgerton VR, Sinha S. In vivo
intramuscular fascicle-aponeuroses dynamics of the human medial
gastrocnemius during plantarflexion and dorsiflexion of the foot.
J Appl Physiol 107: 1276 –1284, 2009. First published July 16, 2009;
doi:10.1152/japplphysiol.91598.2008.—Velocity-encoded phase-contrast
magnetic resonance (MR) imaging techniques and a computer-controlled
MR-compatible foot pedal device were used to investigate the medial
gastrocnemius muscle and aponeurosis deformations during passive
and active eccentric movements of the plantarflexors. Intrafascicular
strain, measured as the ratio of strain in the fascicle segment at its
insertion to strain at its origin, was nonuniform along the proximodistal axis of the muscle (P ⬍ 0.01), progressively increasing from the
proximal to distal direction. The high intrafascicular strain regions
appeared to correlate with the muscle regions that are likely to
encounter high stress concentrations, i.e., the regions where the
muscle physiological cross section decreases close to the tendons. The
architectural gear ratio, i.e., the mechanical amplification ratio of
fascicle length displacement to that of tendon/aponeuroses in a pennate muscle, also exhibited significant regional differences, with the
highest ratios in the proximal region of the muscle accompanied by a
higher initial pennation angle and a larger range of fascicular rotation
about the origin. Values close to unity in the distal region of the
muscle suggest that the aponeurosis separation may decrease in this
region. Fascicle length and pennation angle changes were significantly
influenced by force generation in the muscle, probably due to a
shortening of the loaded muscle fibers relative to a passive condition.
Overall, our data illustrate significant proximodistal intramuscular
heterogeneity as supported by a regionally variable end-to-end strain
ratio of fascicles and angle changes in the medial gastrocnemius
muscle during passive and active ankle movements. These observations emphasize the need to reassess current conceptual models of
muscle-tendon mechanics.
MG MUSCLE ARCHITECTURE
MATERIALS AND METHODS
Study Design
Displacement of the intramuscular tissue of the MG muscle and its
two surrounding aponeuroses were measured using velocity-encoded
(VE) phase-contrast (PC) MRI techniques. Based on the measured
displacements, changes in several architectural parameters were quantified during plantarflexor and dorsiflexor movement of the foot within
the normal range of ankle rotation. The following functional parameters were measured: muscle fascicle length, pennation angle, gear
ratio, and intrafascicle end-to-end strain ratio. These parameters were
measured under two separate exercise paradigms, i.e., during passive
foot rotation and during foot rotation with active eccentric muscle
contraction. The cyclic rotation of the foot (plantar and dorsiflexions)
inside the MR scanner was generated with a computer-controlled
MR-compatible muscle testing foot pedal device described in a
previous report (20).
Subjects
Six healthy male volunteers participated in this study [age: 31.5 ⫾
13.5 yr, height: 174.8 ⫾ 4.1 cm, and weight: 80.5 ⫾ 12.9 kg (means ⫾
SD)] who had no prior history of muscle and tendon pathologies. The
details of the experimental procedures were explained to the subjects
who then signed the Institutional Review Board consent form approved by the Institutional Human Research Protections Program.
Muscle Testing System
A muscle testing apparatus designed and developed by our group
was used for dynamic imaging of the MG muscle. The apparatus,
described in detail in another report (20), is made up of two main
modules, an actuating end and a driven end. The driven module
consisted of a contoured platform to fixate the leg with Velcro straps,
a rotatable foot pedal, an optical force sensor attached to the pedal,
and a custom-made hydraulic piston-cylinder pair. Since this module
is in direct contact with the subject’s lower limb inside the scanner,
the parts of this module were selected to be MR compatible (Plexiglas,
Delrin, etc.). The actuating module was made up of a linear motor
(Animatics, Santa Clara, CA) and a stainless steel piston-cylinder pair
(Parker Hannifin, Cleveland, OH). The module was controlled by a
computer in the console room away from the magnetic environment.
The two modules were connected by a high-pressure tube (McMaster
Carr, Los Angeles, CA) through a waveguide entry (hole) in the wall
J Appl Physiol • VOL
separating the patient room from the console room. Distilled water
was used as the hydraulic fluid. The foot pedal was designed to rotate
over a wide range of angles (65–135°) to accommodate the full range
of dorsi-/plantarflexions. The angle of rotation was monitored from
the output of the motor module, which provided a measure of the
piston displacement and indirectly the angular displacement. The
same output was also connected to the scanner ECG gating cable via
a digital output port of the motor and was used to gate or synchronize
the PC MRI acquisition to any desired point in the motor cycle by
appropriate programming of the motor.
Exercise Paradigm
The subject lay supine on the MR scanner bed, and the foot of the
dominant leg (self-reported) was positioned on the foot pedal of the
muscle testing system. Extra care was taken to ensure that the rotating
axis of the foot aligned with that of the foot pedal. The pedal was
programmed to complete a rotational cycle of 120 –75-120° with the
cyclic period of 2.86 s (21 cycles/min). The angular values here
represent the angle formed by the main axis of the tibia and the foot
plate, thus 120 and 75° correspond to most plantarflexed and dorsiflexed positions, respectively. This specific range was consistently
applied to all subjects with minor adjustments to ensure the safety and
comfort of each subject while maintaining consistency. To minimize
fluid inertia in the hydraulic system, the foot pedal was driven at a
constant rate of acceleration in each direction, resulting in linearly
changing velocities and approximately sinusoidal angular displacement. The average absolute angular velocity used during the data
acquisition of the MG muscle was ⬃31°/s (range: 0 – 63°/s) for all
subjects.
For the passive mode of exercise paradigm, the subject was
instructed to fully relax the leg while the foot rotation was driven by
the muscle testing system. For the eccentric mode, the subject exerted
40% of the maximum voluntary contraction (MVC) during the dorsiflexion phase of the foot rotation cycle, thereby activating the
plantarflexors under the stretching effect of the dorsiflexion. The
complete acquisition of MR images under each mode required 70
identical contractions. Ankle torque was measured with a force
transducer on the foot plate and the force signal along with a target
force level projected onto the face of the scanner to provide feedback
to the subject. The 40% maximum voluntary contraction level was
chosen as a target for the eccentric mode to ensure that the data
acquisition could be completed without fatigue of the subject muscles.
MRI Protocols
The location and orientation of an oblique sagittal slice was
determined from a set of localizer images and multiple oblique slices.
The slice was chosen to be oriented approximately perpendicular to
the planes of the aponeuroses in the thickest region of the muscle,
corresponding to the region and orientation recommended in a previous study (9) to best represent a section parallel to muscle fascicles.
Morphological imaging. The orientation of muscle fascicles was
estimated with a spin-lattice relaxation time (T1)-weighted pulse
sequence technique. In brief, T1-weighted MR images are generally
acquired with the proton resonance frequency centered on water
protons, with typically the protons in the fat suppressed with special
off-resonance radiofrequency pulses. In our experiment, we flipped
this procedure, with the imaging frequency centered on the fat
resonance frequency and suppressing the water proton signal. Thus,
instead of fat being suppressed in the first (conventional) method
compared with water, our images (as shown in Fig. 1A) have fatty
tissue brighter compared with water (muscle) tissue. The extra contrast that this approach provided over just T1-weighted images is
shown in Fig. 1, A and C, where the fatty tissue signal is enhanced
compared with the T1-weighted (although not to the fullest extent)
magnitude image that was produced along with the VE PC image
shown in Fig. 1B.
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
We selected the MG muscle because its anatomy suggests that
aponeuroses lie on the surface of the muscle and that the
intramuscular region has a relatively uniform structure devoid
of large, load-transmitting tendinous structures. Our hypothesis
was that strain in a single muscle fiber undergoing a length
change is inhomogeneous and related to the intramuscular
location of the fiber.
These data also provided us with an opportunity to directly
relate whole muscle fiber shortening to the resulting displacement of aponeuroses. It is widely acknowledged that the
aponeuroses shear is greater than the muscle fiber shortening
due to the special mechanics of the muscle resulting from the
observed movements of the aponeuroses that tend to be parallel
to each other rather than along the direction of muscle fiber pull
(8, 16, 18). This “amplification” of muscle fiber length changes
into aponeurosis displacement has been termed the “gear ratio”
(1, 2). In a previous report (9), we suggested that the gear ratio
is sensitive to slight changes in aponeurosis separation as well
as the muscle fiber pennation angle. We therefore also examined the gear ratios and their relationship to pennation angle
and aponeurosis separation in the MG muscle.
1277
1278
MG MUSCLE ARCHITECTURE
using a vendor-supplied product PC sequence (field of view: 16.5 ⫻
30 cm, matrix: 141 ⫻ 256, slice: 5 mm, repetition time/echo time/flip
angle: 16.5 ms/7.7 ms/20°, bandwidth: ⫾15.6 kHz, velocity encoding:
10 cm/s in three directions, views per segment: 4, and number of
excitations: 2). The details of this technique have been previously
described (22, 23).
Data Processing
This technique was used to elucidate the structural orientation of
fatty layers running parallel to muscle fibers, facilitating the selection
of regions of interest (ROIs) along the direction of fibers over the
entire length of the MG muscle. The muscle fascicle ends were
directly localized along the length of the MG muscle using this image.
It should be noted that the images were obtained with the ankle at the
extreme plantarflexed position used in our experiments (120°).
Image registration. A 1:1 mapping between the water saturated
image and the corresponding PC image was necessary to translate the
location of ROIs selected on the T1-weighted image onto the PC
image. This was accomplished by an indigenously developed image
registration algorithm (MATLAB, Mathworks, Natick, MA), which
determined the x and y translations corresponding to the best shape
alignment between the two images (Fig. 1).
PC imaging. The ROIs selected based on the T1-weighted images
were tracked using the VE PC MRI data. Images were acquired on a
1.5-T Signa HDx MR scanner (GE Medical Systems, Milwaukee, WI)
J Appl Physiol • VOL
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
Fig. 1. Superposition of spin-lattice relaxation time (T1)-weighted and phasecontract (PC) magnitude images before (A) and after (B) image registration.
Vertical and horizontal offsets were used to create a 1:1 map between the two
images. The T1-weighted image was used to determine the ends of the muscle
fascicles, as shown by the red symbols in C. Those end points and the
connecting lines were translated onto the PC images for displacement tracking
of the muscle during dynamic foot rotation and the subsequent calculation of
fascicle length and pennation angle changes (D).
The tracked ROIs were used to compute several measures of
relevance to our study of muscle-aponeurosis complex dynamics.
Scanning was initiated at the extreme plantarflexed position (120°)
through dorsiflexion to the extreme dorsiflexed position (75°) and then
back through plantarflexion to the start position. The measures of
change reported in this study compare values measured at the extreme
plantarflexed and extreme dorsiflexed positions of the ankle, accomplished with an ankle excursion of 45°. ROIs along the superior and
deep aponeuroses were used to define the orientation and deformation
of the aponeuroses. Muscle fascicles were characterized by similar
ROIs arranged between their origins and insertions and orientated
parallel to the fatty layers identified by using the water-saturated
images as described above. Generally, between three and five equally
spaced ROIs per fascicle were identified in addition to the ROIs
defining their attachment to the aponeurosis. Fascicle strains were
measured between each pair of adjacent ROIs. In addition, the
intrafascicle end-to-end strain ratio was calculated as the ratio between the strain in the fascicle segment closest to its insertion and
strain in the segment closest to its origin. This measure summarizes
any systematic inhomogeneity in fascicle strain and yields a value less
than or greater than unity depending on which end of the fascicle
strained the most. Pennation angles were calculated as the angle
between the line defined by the fascicle end points and the line created
by the two points on the aponeurosis adjacent to the fascicle end point.
The pennation angles are reported for each end of the fascicle. The
changes in pennation angle that we report are the averages of the
pennation angles at each end of the fascicle. Our results indicated that
the ROIs tracking each fascicle remained oriented along a line joining
the fascicle origin and insertion, suggesting that the fibers do not curve
significantly during active or passive muscle length changes. Further
support for the notion that MG fascicles bend minimally comes from
spin-tagged images where the tag lines in the MG muscle remain
parallel as they move during a length change (12). Muscle fascicle
length was calculated as the distance between the two ROIs at the
intersections of the fascicles and the aponeuroses of origin and
insertion. The same ROIs were used to compute shear between the
two aponeuroses. The ratio between fascicle length change and
aponeurosis shear at the extremes of plantarflexion and dorsiflexion
were used to calculate the architectural gear ratio (2) (Fig. 2, A and B).
A critical question in the investigation of fiber properties in dynamically moving muscle relates to the reliability of the results. We
therefore assessed the test-retest repeatability of several measured
parameters, i.e., gear ratio, end-to-end fascicle strain ratio, and
fascicle length changes, based on the selected ROIs along the
length of the MG muscle (Fig. 2, B and D). Two separate
measurements using the same set of ROIs were compared in two
subjects based on the intraclass correlation coefficient. Student’s
paired t-test was used as an additional tool to assess repeatability. A
further error may arise if estimates of fascicle locations are erroneous.
To determine the differences between actual fibers and the ones traced
from the T1-weighted images shown in Fig. 1A or inaccurate placement of ROIs representing the ends of fascicles, errors in orientation
of the fibers were artificially introduced and the consequent effects
were investigated.
To analyze possible uncertainties in visualizing fascicles by our
imaging technique, we performed additional analyses by simulating a
change in the orientation of the fiber by first displacing one of its ends
by 5 pixels (⬃5.8 mm) in one direction and the other end in the
1279
MG MUSCLE ARCHITECTURE
opposite direction and then repeating the process by switching the
directions of displacements. This effectively rotated the fascicles
around their midpoints, in one case reducing the pennation angle by
⬃5° and in the other increasing the pennation angle by ⬃8°. The
effect of the altered fascicle orientation did not qualitatively change
the overall spatial dependence of fascicle strain (Fig. 2C). Large
changes were observed in some fascicles (e.g., fascicle 2 in Fig. 2C),
probably due to local mechanical effects of noncontractile components, which changed as the sample regions changed.
To test for regional differences in the behavior of muscle fascicles,
they were grouped according to location in the muscle. The two
fascicles located most proximally in the muscle formed the proximal
group, the two fascicles located most distally were allocated to the
distal group, and the remaining fascicles were allocated to the midregion group (Fig. 1D). The region and type of exercise paradigm
(passive vs. eccentric) as described above were considered as two
independent factors. The effects of these two factors on the measured
variables were assessed using two-way ANOVA. A post hoc multicomparison test (Bonferroni) was used where applicable. Significance
was set at P ⬍ 0.05.
RESULTS
Reliability of ROI Trajectories
The intraclass correlation coefficient values of the gear ratio,
end-to-end fascicle strain ratio, and fascicle length changes were 0.89,
0.80, and 0.86, respectively. There were no significant differences in
any of these parameters between mean values of the two repeated
measurements (P ⬍ 0.05).
Figure 3A shows the trajectories of seven pairs of ROIs selected to
show deformations in several regions of the MG muscle during the
passive ankle movements. A general observation common to most
J Appl Physiol • VOL
points plotted is the consistent pattern over time of the movement of
the ROIs and adjacent regions of the muscle. Also to be noted is that
the tracked trajectories moved along the boundary of the aponeuroses,
an indication that the two-dimensional tracking algorithm provided
accurate tracking of tissue displacement. In addition to other validations such as spin tagging (23) and findings reported in previous
publications (12, 21), these data add to our confidence in the validity
of our measurements. Figure 3B shows the range of displacement at
most distal (red), middle (green), and most proximal (blue) ROIs at
both ends of the muscle, i.e., deep and superficial aponeuroses.
Fascicle Length and Pennation Angle
Figure 4 shows the initial fascicle lengths (A) and pennation angles
(B) in a relaxed muscle at an ankle angle of 120°, corresponding to a
fully plantarflexed position. Under these conditions, we would expect
pennation angles to be greater and fascicle lengths to be shorter than
a majority of reports in the literature, where the ankle is usually at 90°.
The values, however, appear to be consistent with other reports (14,
15) of these values measured in a plantarflexed position. There were
no statistically significant regional differences in fascicle length, but
pennation angles increased from the distal to proximal regions. The
pennation angles measured from the ends (insertion and origin) of
each fascicle were similar, as shown in Fig. 4B.
Figure 5A shows the average fascicle length change ⫾ SD of all the
identified fascicles in a representative MG muscle during one dorsiflexion-plantarflexion cycle. The two curves correspond to two different modes of the exercise paradigm (passive vs. eccentric) and
show the significantly lower fascicle length change during the eccentric mode.
Figure 5B shows the range of fascicle length changes as a function
of muscle region (distal, middle, and proximal) and the type of
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
Fig. 2. Estimates of errors and reproducibility
in estimates of gear ratio and fascicle strain.
A: fascicle shortening and aponeurosis
shear along the main axis of force generation. Data were from a representative subject showing the regional difference in the
relationship between the two measures.
The architectural gear ratio is the ratio
between the shear displacement of the aponeuroses and the fascicle length change.
Note that the gear ratio increases proximally as the aponeurosis shear becomes
greater than fascicle shortening. B: testretest repeatability of the gear ratio calculated from the same set of regions of interest (ROIs) positioned on two separate MRI
datasets obtained within the same day. The
graph shows data from two subjects. The
repeated data from the two subjects are denoted by two different symbols (squares vs.
triangles). C: the end-to-end strain ratio illustrates a progressive shift in the region of
maximal strain in the fascicle, from the end
of the fascicle closest to the superficial aponeurosis in the proximal region to the region
closest to the deep aponeurosis (Achilles tendon) in the distal region of the muscle. The
three different curves show the robustness of
the observation relative to potential errors in
our estimates of fascicle orientation. They
were obtained from the same subject by varying the angle of the selected fascicles relative
to the original orientation (default, ⬃default ⫺ 5° for shallow, and ⬃default ⫹ 8° for
steep changes). D: test-retest repeatability of
the end-to-end strain ratio between two separate measurements on two subjects.
1280
MG MUSCLE ARCHITECTURE
Fig. 3. Displacement of ROIs during a passive
ankle dorsiflexion-plantarflexion cycle. A: seven
pairs of ROIs corresponding to the ends of
muscle fascicles on deep (D) and superficial
(S) aponeuroses. The deep aponeurosis is the
proximal extension of the Achilles tendon and
therefore moves substantially with ankle rotation. The superficial aponeurosis is the tendon
of origin for the medial gastrocnemius muscle
and moves minimally during the contraction.
B: graph showing the vertical displacement of
the origin and insertion points of muscle fascicles on the aponeuroses. Downward (distal)
movements of the ROIs are plotted as positive
displacement. The colors correspond to the
ROIs in A. Note that the origins of the fascicles
are almost stationary relative to movement of
the insertions of the fascicles.
multiple-comparison test found a significant difference of pennation
angle between distal and proximal regions.
Fascicle Strain Distribution
Evaluation of strain along the length of a fascicle indicated that the
relative changes in strain between the two ends of the muscle reversed
along the proximodistal axis of the muscle (Figs. 2C and 6, A–C). This
was summarized by calculating the ratio of the strain changes in the
regions at each end of the muscle [strain values near the deep
aponeurosis (left) and at the superficial aponeurosis (right) in Fig. 6,
A–C]. This ratio was termed the end-to-end strain ratio and represents
the shortening of one end of a muscle fascicle relative to its other end.
The end-to-end strain ratio data are shown in Fig. 6D. The greater than
unity value of the ratio implies that the deep end (insertion) of a given
fascicle strained more than the superficial end (origin). The opposite
is true when the ratio value is ⬍1. A value of unity implies that a
fascicle strained equally at both ends. Substantial differences of strain
occurred at the two ends of muscle fascicles, ranging from a twofold
greater strain near the superficial aponeurosis in proximal fascicles to
almost equal strains in the midregion of the muscle. A 1.5-fold greater
strain occurred near the deep aponeurosis in the distal portion of the
muscle. Two-way ANOVA found a highly significant effect of the region
(F ⫽ 21.48, P ⬍ 0.001) but no effect of exercise mode. The post hoc test
indicated that the ratio was different in all three regions.
Gear Ratio of the Fascicle to Aponeurosis
Fig. 4. A: initial fascicle lengths in three regions (distal, mid, and proximal) at
the fully plantarflexed ankle. B: initial pennation angle of the fascicles in the
three regions. Since each fascicle yields two pennation angles at its two ends,
the initial pennation angles were grouped into insertion and origin angles at
each region. Values are means ⫾ SD; n ⫽ 12 (distal), 30 (mid), 12 (proximal).
J Appl Physiol • VOL
For the purposes of our investigation, we redefined the architectural
gear ratio as the relationship between changes in fascicle length and
the relative movement (shear) of the aponeuroses, making the assumption that fascicle length is a reasonable estimate of muscle fiber length
and that similar absolute length changes occur in both. A gear ratio
greater than unity indicates that the aponeurosis displacement is
greater than the fascicle length change due to the geometry of muscle
deformations. Gear ratios were statistically different among regions
(F ⫽ 12.94, P ⬍ 0.001) and progressively increased in the distal to
proximal direction, with the highest value in the proximal region (Fig. 7). The
Bonferroni multiple-comparison test showed that all regions were
significantly different. The mode of exercise paradigm (passive vs.
eccentric) had no significant effect on gear ratio (F ⫽ 2.9, P ⫽ 0.092).
A large SD of gear ratios was observed in the proximal region, which
reflected greater variability between subjects in this region.
Table 1 shows the comprehensive summary of the statistical test
results of all measured parameters reported in this study. Note that the
reported P values are quite low, particularly for comparisons between
regions (P ⬍ 0.001 in all cases), suggesting highly significant regional
effects upon the mechanical responses of fascicles within the MG
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
exercise paradigm (passive vs. eccentric). Two-way ANOVA found a
significant effect of exercise mode (F ⫽ 17.54, P ⬍ 0.001), indicating
that the eccentric mode generated a significantly smaller range of
fascicle length change than the passive one. There was no significant
effect of muscle region (F ⫽ 2.43, P ⫽ 0.094), implying that the
fascicle length change did not differ significantly between the three
regions. Pennation angles, however, differed both as a function of
muscle region (F ⫽ 7.88, P ⫽ 0.007) and the type of exercise (F ⫽
9.56, P ⫽ 0.0027), as shown in Fig. 5C. The Bonferroni post hoc
MG MUSCLE ARCHITECTURE
1281
Intrafascicle Strain
Fig. 5. Changes in fascicle length and pennation angle during passive and
eccentric ankle rotation in three different muscle regions. A: average length
change of all fascicles (n ⫽ 8) in a representative subject muscle throughout a
dorsiflexion-plantarflexion cycle. The fascicle length change was smaller
during the eccentric contraction. B: regional changes in fascicle length showing
no significant effect of region but significantly lower fascicle length changes in
the eccentric mode of exercise. C: regional changes in pennation angle
showing significant effects of region and of exercise mode. Values are
means ⫾ SD; n ⫽ 12 (distal), 30 (mid), and 12 (proximal).
muscle. Significant differences between eccentric and passive movements further suggest that the mechanical response of fascicles and
muscle is significantly modified by the mode of exercise paradigm.
DISCUSSION
We observed considerable heterogeneity in several measures of strain within the MG muscle and its aponeuroses. A
significant factor seems to be location within the muscle, but
J Appl Physiol • VOL
The strain levels observed in the extremities of some
fascicles were as high as 200% during passive dynamic
ankle rotation, suggesting that some MG sarcomeres may
triple their length during a large amplitude ankle rotation
(Fig. 6C). These rather high fascicle strains include changes
relative to the extreme plantarflexion position, when the
muscle fibers would be the shortest. A 1.5-fold change in
sarcomere length over a normal range of passive wrist
movement (13) has been reported for human wrist extensor
muscles. The present data would correspond to this range if
the initial fascicle length was readjusted to twice its length
measured at maximum plantarflexion (i.e., corresponding to
fascicle length at the midrange of the overall range of
movement). Attempts to fit the wrist extensor data to a
hypothetical length-tension curve suggest that the sarcomeres may shorten by an additional 50% before reaching the
descending portion of the length-tension curve. Ultrasound
measures of MG fiber length changes during ankle flexion
and maximum voluntary contractions indicate fiber lengths
ranging from ⬃20 mm in maximum isometric contraction at
a neutral ankle position to ⬃55 mm at 15° of dorsiflexion
(14). In each of these cases, it is highly likely that a
significant contraction at the shortest muscle lengths would
have resulted in even greater sarcomere and fiber shortening. It should also be noted that while our data include
strains from highly strained regions of a fascicle, the overall
fascicle length changes we observed were ⬃100% of the
shortest fiber length (at full plantarflexion).
The fascicle end closest to the deep aponeurosis exhibited
a progressive increase in strain moving from the proximal to
distal regions of the muscle (Fig. 6A). The reverse was true
at the other end, i.e., fascicles adjacent to the superficial
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
the mode of exercise paradigm i.e., passive versus active
eccentric, also plays a significant role. The major effects of
the two exercise modes are reflected in fascicle length and
pennation angle changes. Our observation of the velocity
data from the VE PC MRI indicates successively higher
velocities going from eccentric to passive to concentric
exercise paradigms (20). This is to be expected since contracting fibers exert forces that stretch series elastic elements, thus reducing fiber elongation under the eccentric
mode of exercise while contributing to the shortening in the
concentric mode (14). The concentric mode here refers to
voluntary activation of plantarflexors when the plantarflexors are shortening. The velocities in the passive mode would
be intermediate between those in the concentric and eccentric modes. The same argument would hold for the lower
pennation angle changes during the eccentric mode relative
to the passive condition, as shown in this study (Fig. 5C).
One surprising observation was the relative uniformity of
fascicle length changes in the three muscle regions compared. Huijing et al. (10) predicted that midregion fibers of
an isometrically contracting muscle would shorten more
than fibers in other regions and even predicted that fibers at
both ends of the muscle would lengthen. We were unable to
verify this prediction with our current observations in the
human MG muscle undergoing passive or eccentric length
changes.
1282
MG MUSCLE ARCHITECTURE
aponeurosis (Fig. 6C). This resulted in a significant and
progressive change in the ratio of strains (Figs. 2D and 6D)
at the origins and insertions of the fascicles from ⬃1.5 in the
distal region to ⬃0.5 in the proximal region of the muscle
(P ⬍ 0.001). These high strain regions in the fascicles were
observed in the regions where the muscle thins near the
musculotendinous junctions of its origin and insertion and
corresponds to the regions of high strain and intrafiber pressure
in some models of contracting muscle (11, 26). It is currently
not possible to determine the precise distribution of stress
throughout a muscle, but it seems reasonable to assume that
the total (integrated) force at any cross section of the muscle
and tendon remains fairly constant along the proximodistal
axis. The smaller cross-sectional areas as the muscle thins and
becomes tendon will thus result in a higher stress concentration
to accommodate the same stress over a smaller area and
therefore potentially higher strains if material properties remained constant. It should be noted that in Pappas et al.’s work
J Appl Physiol • VOL
(19), the regions of differential strains tended to concentrate in
the regions of the central aponeurosis, where there were larger
differences in the intramuscular mix of passive and active
tissues. But, in addition, they observed differences in fiber
strain well beyond the region identified as mixed with aponeurosis, consistent with the modeling prediction that the intramuscular location itself may have some influence on fiber
strain.
Gear Ratio
The translation of work from muscle fiber shortening to
movement of the aponeurosis seems to involve the interactions
of multiple force vectors, and, as a result, a range of displacement outcomes can occur. It is generally accepted that force
exerted by fibers is reduced by at least the cosine of the fiber
pennation angle and that there is a corresponding increase in
aponeurosis movement to balance the work done by fibers and
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
Fig. 6. Intrafascicular strain during dorsiflexion phase of the foot rotation under passive exercise mode. A: strain along the most distal fascicle in a representative
subject. Data points represent the strain measured at different locations along the fascicle. Left, region closest to the fiber insertion (deep aponeurosis/Achilles
tendon); right, region closest to the fiber origin (superficial aponeurosis). Individual lines identify the elapsed time (noted in the insets) from the initiation of
dorsiflexion. B: same as in A except from a fascicle in the midregion of the same subject. C: same as in A and B except from a fascicle in the most proximal
location. D: intrafascicular end-to-end strain ratio for all subjects showing significant regional differences but no difference between the passive and eccentric
mode. Values are means ⫾ SD; n ⫽ 12 (distal), 30 (mid), and 12 (proximal).
1283
MG MUSCLE ARCHITECTURE
muscle (8, 16, 18). This and other work has suggested that the
gear ratio may vary depending on the initial pennation angle,
the precise mechanics of aponeurosis separation, and even the
level of forces generated by the muscle relative to the opposing
forces (2, 9). Our present observations demonstrate that the
mechanics of each component of the muscle-tendon complex
can vary substantially based on the contractile conditions under
which the muscle operates and even on regional differences in
a muscle contraction. For example, we found a significant
proximodistal decrease in the gear ratio, suggesting a changing
characteristic of muscle along the proximodistal axis (Fig. 7).
The gear ratio increases with pennation angle and decreases as
the aponeurosis separation decreases, suggesting either lower
(acute) pennation angles distally, or inward movement of
aponeuroses during a contraction or muscle length change. One
clear finding of this study was that the initial pennation angle
of the distal region was significantly lower than the mid and
proximal regions (Fig. 4B). The occurrence of gear ratios equal
to or less than unity, however, cannot be explained by the
pennation angle, suggesting that there may be less aponeurosis
separation in the distal region than in proximal region of the
muscle (7). A reduction of interaponeuroses distance of a few
millimeters during shortening would accomplish the observed
drop in the gear ratio (9). These findings challenge the commonly held assumptions that the distance between aponeuroses
remains constant when a muscle shortens. The present finding
not only suggests that this distance can change but that this
change can vary along the length of the muscle.
The higher gear ratio in the proximal region of the muscle
can be explained by higher pennation angles in that region
(Fig. 4B). Consistent with this finding is the greater pennation
angle change in the region (Fig. 5C) permitting higher gear
ratios by increasing the rotation of muscle fibers while maintaining a similar range of fiber stretching/shortening (Fig. 5B).
In more general terms, evidence is accumulating that the
intramuscular mechanics are considerably more complex than
is generally assumed, i.e., muscles are not neatly constructed in
J Appl Physiol • VOL
Limitations
A significant limitation of this study pertains to selection of
fascicle location and orientation along the length of the MG
muscle. It can be argued that the determination of fiber architecture is not possible with the conventional magnetic resonance imaging, although the diffusion tensor imaging of muscle holds promise (4, 24), since the spatial resolution offered by
MRI is a few orders of magnitude larger than the scale of fibers
and even fascicles. However, a reasonable assumption was
made in regards to their orientation based on fatty layers of
connective tissue aligned in parallel with the fascicle arrangement in the MG muscle using enhanced imaging of fat compared with water. Even though the tissue tracking as used in
this study would not have resolved the movement of a single
muscle fiber, it is reasonable to conclude that the local strain
measured reflects a large collection of fibers moving in tandem.
The measurements were shown to be sensitive enough to detect
Table 1. Summary of statistical differences of the measured
parameters relative to muscle regions and modes of the
exercise paradigm
Region (Distal, Mid,
and Proximal)
Contraction Mode
(Passive and
Eccentric)
Parameter
F value
P value
F value
P value
Initial fascicle length
Initial pennation angle
Fascicle length change
Pennation angle change
Gear ratio
End-to-end strain ratio
0.53
55.86
2.43
7.88
12.94
21.48
NS
⬍0.001†
NS
⬍0.001*
⬍0.001†
⬍0.001†
NA
NA
17.5
9.56
2.9
0.35
NA
NA
⬍0.001
⬍0.01
NS
NS
NS, not significant; NA, not applicable. *Significant difference between
proximal and distal regions; †significant difference between all three regions.
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
Fig. 7. Architectural gear ratio as a function of muscle region and mode of
contraction. Summary data from all subjects are shown. #P ⬍ 0.05, significant
differences between all regions. Values are means ⫾ SD; n ⫽ 12 (distal), 30
(mid), and 12 (proximal).
an optimized way, which often seems to be implied in efforts
to understand their function. Variability in resting fiber length
relative to optimum length is becoming well accepted and even
recognized as a mechanism to broaden the range of operating
lengths of a muscle (6). Mechanically, it may be impossible to
integrate uniform fiber strain into a complex muscle structure
that, due to regional differences in stress levels and mixes of
passive and active materials, must exhibit a high degree of
mechanical heterogeneity. Perhaps most importantly, we must
recognize that the behavioral repertoire of an organism requires
a wide range of functional demands such as changing levels of
deformation and regional stresses. It is not difficult to imagine
that such functional flexibility would be difficult to implement
without significant changes in the intramuscular mechanical
environment, expressed as regional inhomogeneity and changing mechanics as the demand upon the muscle changes. For the
most commonly used behaviors, muscles seem to operate at
relatively low levels, leaving a large reservoir of unused
operational potential (9, 10). In that case, suboptimal operation,
such as fibers operating in different regions of the lengthtension curve, may not be a major concern in most movements.
With this in mind, it may be that much may be learned from
investigating the contributions of the diverse structural variation of certain muscles to the performance of an organism in
critical high-output performance that simulates survival conditions such as escape from predators.
1284
MG MUSCLE ARCHITECTURE
GRANTS
This work was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant RO1-AR-53343 and by General Electric
Grant TAB-A26.
REFERENCES
1. Azizi E, Brainerd EL. Architectural gear ratio and muscle fiber strain
homogeneity in segmented musculature. J Exp Zool Part A Ecol Genet
Physiol 307: 145–155, 2007.
2. Azizi E, Brainerd EL, Roberts TJ. Variable gearing in pennate muscles.
Proc Natl Acad Sci USA 105: 1745–1750, 2008.
3. Blemker SS, Pinsky PM, Delp SL. A 3D model of muscle reveals the
causes of nonuniform strains in the biceps brachii. J Biomech 38: 657–
665, 2005.
4. Damon BM, Ding Z, Anderson AW, Freyer AS, Gore JC. Validation of
diffusion tensor MRI-based muscle fiber tracking. Magn Reson Med 48:
97–104, 2002.
J Appl Physiol • VOL
5. Drace JE, Pelc NJ. Tracking the motion of skeletal muscle with velocityencoded MR imaging. J Magn Reson Imaging 4: 773–778, 1994.
6. Ettema GJ, Huijing PA. Effects of distribution of muscle fiber length on
active length-force characteristics of rat gastrocnemius medialis. Anat Rec
239: 414 – 420, 1994.
7. Finni T, Hodgson JA, Lai AM, Edgerton VR, Sinha S. Nonuniform
strain of human soleus aponeurosis-tendon complex during submaximal
voluntary contractions in vivo. J Appl Physiol 95: 829 – 837, 2003.
8. Gans C. Fiber architecture and muscle function. Exerc Sport Sci Rev 10:
160 –207, 1982.
9. Hodgson JA, Finni T, Lai AM, Edgerton VR, Sinha S. Influence of
structure on the tissue dynamics of the human soleus muscle observed in
MRI studies during isometric contractions. J Morphol 267: 584 – 601,
2006.
10. Huijing PA. Muscle, the motor of movement: properties in function,
experiment and modelling. J Electromyogr Kinesiol 8: 61–77, 1998.
11. Jenkyn TR, Koopman B, Huijing P, Lieber RL, Kaufman KR. Finite
element model of intramuscular pressure during isometric contraction of
skeletal muscle. Phys Med Biol 47: 4043– 4061, 2002.
12. Kinugasa R, Shin D, Yamauchi J, Mishra C, Hodgson JA, Edgerton
VR, Sinha S. Phase-contrast MRI reveals mechanical behavior of superficial and deep aponeuroses in human medial gastrocnemius during isometric contraction. J Appl Physiol 105: 1312–1320, 2008.
13. Lieber RL, Loren GJ, Friden J. In vivo measurement of human wrist
extensor muscle sarcomere length changes. J Neurophysiol 71: 874 – 881,
1994.
14. Maganaris CN, Baltzopoulos V, Sargeant AJ. In vivo measurements of
the triceps surae complex architecture in man: implications for muscle
function. J Physiol 512: 603– 614, 1998.
15. Martin DC, Medri MK, Chow RS, Oxorn V, Leekam RN, Agur AM,
McKee NH. Comparing human skeletal muscle architectural parameters
of cadavers with in vivo ultrasonographic measurements. J Anat 199:
429 – 434, 2001.
16. Narici M. Human skeletal muscle architecture studied in vivo by noninvasive imaging techniques: functional significance and applications. J
Electromyogr Kinesiol 9: 97–103, 1999.
17. Oomens CW, Maenhout M, van Oijen CH, Drost MR, Baaijens FP.
Finite element modelling of contracting skeletal muscle. Philos Trans R
Soc Lond B Biol Sci 358: 1453–1460, 2003.
18. Otten E. Concepts and models of functional architecture in skeletal
muscle. Exerc Sport Sci Rev 16: 89 –137, 1988.
19. Pappas GP, Asakawa DS, Delp SL, Zajac FE, Drace JE. Nonuniform
shortening in the biceps brachii during elbow flexion. J Appl Physiol 92:
2381–2389, 2002.
20. Shin D, Kinugasa R, Mishra C, Hodgson JA, Edgerton VR, Sinha S.
Mapping of Movements in the Dynamically Contacting Human Triceps
Surae Muscles Using a Computer-Controlled Hydraulic Foot-Pedal Device with Velocity Encoded Phase Contrast MRI. Toronto, ON, Canada:
Proceedings of the 16th Annual Meeting of ISMRM, 2008.
21. Shin D, Finni T, Ahn S, Hodgson JA, Lee HD, Edgerton VR, Sinha S.
In vivo estimation and repeatability of force-length relationship and
stiffness of the human Achilles tendon using phase contrast MRI. J Magn
Reson Imaging 28: 1039 –1045, 2008.
22. Shin D, Finni T, Ahn S, Hodgson JA, Lee HD, Edgerton VR, Sinha S.
Effect of chronic unloading and rehabilitation on human Achilles tendon
properties: a velocity-encoded phase-contrast MRI study. J Appl Physiol
105: 1179 –1186, 2008.
23. Sinha S, Hodgson JA, Finni T, Lai AM, Grinstead J, Edgerton VR.
Muscle kinematics during isometric contraction: development of phase
contrast and spin tag techniques to study healthy and atrophied muscles. J
Magn Reson Imaging 20: 1008 –1019, 2004.
24. Sinha S, Sinha U, Edgerton VR. In vivo diffusion tensor imaging of the
human calf muscle. J Magn Reson Imaging 24: 182–190, 2006.
25. van Soest AJ, Huijing PA, Solomonow M. The effect of tendon on
muscle force in dynamic isometric contractions: a simulation study.
J Biomech 28: 801– 807, 1995.
26. Yucesoy CA, Koopman BH, Huijing PA, Grootenboer HJ. Threedimensional finite element modeling of skeletal muscle using a twodomain approach: linked fiber-matrix mesh model. J Biomech 35: 1253–
1262, 2002.
107 • OCTOBER 2009 •
www.jap.org
Downloaded from http://jap.physiology.org/ by 10.220.32.247 on June 18, 2017
changes in the mechanical output of these fascicles as a
function of anatomical position and loading conditions. Furthermore, to assess the sensitivity of fiber orientation on the
outcome of measured parameters (13), we repeated the intrafascicle strain ratio measurement two more times on the same data,
by increasing and decreasing the orientation of the selected fibers.
As shown in Fig. 2C, varying the angle relative to the original
orientation did not change the main trend in the intrafascicle strain
ratio.
It is also to be noted that the translation of linear displacement of the motor piston to the angular rotation of the foot
pedal was nonlinear. In fact, the piston oscillation was driven
by step acceleration functions of ⫾8 in./s2, causing the piston
displacement to be quadratic in both forward and backward
directions. However, the mean angular velocity of the foot
pedal was maintained at ⬃31.47°/s between rotational cycles
for all subjects and such nonlinearities (which are generally the
norm in natural movements) do not invalidate the observations
that we have reported in this study.
In conclusion, this study measured several architectural
parameters in vivo in the healthy human MG muscle. The
fascicular strains were translated into proximodistal movement
of the deep aponeurosis, whereas the superficial aponeurosis
remained approximately stationary. The fascicles positioned at
the distal end of the MG muscle exhibited the highest strain
close to the deep aponeurosis (insertion point), whereas those
at the proximal end experienced the greatest strain close to the
superficial aponeurosis (origin). The fascicles in the midregion
were strained similarly at both of their ends (Fig. 6, B and D).
A higher gear ratio was found in the proximal region (Fig. 7)
accompanied by a higher initial pennation angle (Fig. 4B), with
a higher range of fascicular rotation in the same region (Fig.
5C). The distinct intrafascicular strain pattern measured as the
intrafascicle end-to-end strain ratio along the proximodistal
direction (Fig. 6D) matched the output of some finite-element
models, which predicted high stress and strain concentrations
at both ends of the muscle as it thins and becomes the tendon
(11, 26). Although the PC MRI technique has previously been
used to study the mechanics of tendinous structures, the new
method established herein has demonstrated its ability to characterize muscle architectural parameters, which may be useful
not only for the further elucidation of structure and function but
also for studying the effects of clinical conditions including
aging, muscle atrophy, and surgical procedures.