17 Deguchi
Transcript of 17 Deguchi
-
8/12/2019 17 Deguchi
1/9
Cooperative Regulation of Vocal Fold Morphologyand Stress by the Cricothyroid and ThyroarytenoidMuscles
*Shinji Deguchi,Yuki Kawahara, andSatoshi Takahashi, *Sendai andyOkayama, Japan
Summary:Voice is produced by vibrations of vocal folds that consist of multiple layers. The portion of the vocal fold
tissue that vibrates varies depending primarily on laryngeal muscle activity. The effective depth of tissue vibrationshould significantly influence the vibrational behavior of the tissue and resulting voice quality. However, thus far,
the effect of the activation of individual muscles on the effective depth is not well understood. In this study, a three-
dimensional finite element analysis is performed to investigate the effect of the activation of two major laryngealmuscles, the cricothyroid (CT) and thyroarytenoid (TA) muscles, on vocal fold morphology and stress distribution in
the tissue. Because structures that bear less stress can easily be deformed and involved in vibration, information on
the morphology and stress distribution may provide a useful estimate of the effective depth. The results of the analyses
indicate that the two muscles perform distinct roles, which allow cooperative control of the morphology and stress.
When the CT muscle is activated, the tip region of the vocal folds becomes thinner and curves upward, resulting in
the elevation of the stress magnitude all over the tissue to a certain degree that depends on the stiffness of each layer.
On the other hand, the TA muscle acts to suppress the morphological change and controls the stress magnitude in
a position-dependent manner. Thus, the present analyses demonstrate quantitative relationships between the two mus-cles in their cooperative regulation of vocal fold morphology and stress.
Key Words:Vocal foldLarynxCricothyroid muscleThyroarytenoid muscleFinite element analysis.
INTRODUCTION
Voice is produced at the larynx by structure-fluid interactions
between the vocal folds and respiratory airflow. Mechanical
stress in the vocal folds during these interactions is believed
to cause vocal fold injuries.1 The stress produced by external
loading has been previously studied in finite element
models.27 However, actual vocal folds are complex structures
that experience not only passive stress but also active stress
because of muscle contractions. The vocal folds primarily
consist of membranous, ligamentous, and thyroarytenoid (TA)
muscular layers
8
(hereafter referred to as the membrane, liga-ment, and TA muscle, respectively) and are also externally
stretched inthe anterior-posterior direction by the cricothyroid
(CT) muscle7 in addition to responding to the action of the TA
muscle inside them. Thus, even without airflow forces, thevocal
folds should undergo complicated stresses depending on the
way in which they are acted on by the attached muscles.It has been proposed that vocal folds use muscle-dependent
structural changes to control phonation onset and voice
pitch.811 This theory suggests that tension in the membrane,
regulated by the CT muscle, plays a dominant role in
controlling voice at small-amplitude vibrations, whereas the
TA muscle is involved in large-amplitude vibrations and acts
as an antagonist for the CT muscle.12
Because the mechanical
properties of each layer differ,13 the amplitude of vibration
and parts of the tissue that vibrate are critical. The effective
depth of tissue vibration might be regulated actively by the
attached muscles and passively by the air pressure. Although
such muscle-dependent active stress has been considered in pre-
vious analyses,2,14,15 there has been no study that describes the
relationship between muscle activation and the resultant stress
distribution and morphology, which would be useful as an
estimate of the effective depth. Because the physical depthof
each layer changes in response to the muscle activation,16 in
addition to a longitudinal length change, a three-dimensionalanalysis considering all directions should be used.
The present study evaluates the effect of muscle activation on
the morphology and stress in the prephonatory position using
a three-dimensional finite element analysis. We focus on the
roles of the CT and TA muscles because they are the principal
laryngeal muscles that directly influence vocal fold dynam-
ics.9,17 Static changes in the morphology and stress caused by
the muscle activation are investigated. Tissue thixotropy and
viscoelasticity are not considered in this study. The expected
effective depth is discussed in terms of the stress distribution
in the tissue without analyses of the structure-fluid interaction
because structures that bear less stress could be deformed
relatively easily in response to the driving air pressure.18,19
METHODS
Vocal fold geometry and boundary condition
Finite element calculations are performed using Ansys 11.0
(Cybernet Systems, Tokyo, Japan). A pair of vocal folds is stud-
ied (Figure 1) in which the membrane, ligament, and TA muscle
are included. In the absence of muscle activation, the model is
14-mm long intheZdirection and has the same shape inany cor-
onal (X-Y) plane along theZ-axis. The membrane geometry of
the coronal section is formed by a combination of two sectors,
Accepted for publication November 15, 2010.From the *Department of Biomedical Engineering, Tohoku University, Sendai, Japan;
and the yDepartment of Mechanical Engineering, Graduate School of Natural Scienceand Technology, Okayama University, Kita-ku, Okayama, Japan.
Address correspondence and reprint requests to Shinji Deguchi, Department ofBiomedical Engineering, Tohoku University, 6-6-11-1306-2 Aramaki-Aoba, Sendai980-8579, Japan. E-mail:[email protected]
Journal of Voice, Vol. 25, No. 6, pp. e255-e2630892-1997/$36.00 2011 The Voice Foundationdoi:10.1016/j.jvoice.2010.11.006
mailto:[email protected]://dx.doi.org/10.1016/j.jvoice.2010.11.006http://dx.doi.org/10.1016/j.jvoice.2010.11.006mailto:[email protected] -
8/12/2019 17 Deguchi
2/9
X 5:52Y 22 22 20 q 0; (1)
X2 Y2 7:72 90 q< 20; (2)
where the origin of theX-Ycoordinates is located at the upper-right corner of the plane, and the angle q is measured from the
X-axis. The shapes of the ligament and TA muscle without
activation are geometrically similar to that of the membrane.
The membrane and ligament are 0.35- and 0.8-mm thick,
respectively, according to the literature.8,11 Although different
mechanical properties are assigned to each of these three
layers, relative displacement at any of the interfaces is not
considered. The geometry of this vocal fold model was
determined on the basis of anatomic images anddrawings1,2022 and other modeling studies.3,23 According to
these anatomic data, the entire vocal fold surface is lined with
the vocal ligament layer, as modeled inFigure 1. However, the
ligament is relatively thick around the tip region (which vibrates
the most), and such a localized ligament, without any lining
between the membrane and the TA muscle, has been used in
some previous modeling studies.2,57,24 Thus, the geometry of
the ligament could be a point of controversy; however, in the
present study, we focus on aspects that are essentially
unaffected by whether the ligament is localized or distributed
along the layer.
It is assumed that the relative positions among the elements
in each of the most anterior (the plane initially at Z 14 mm),posterior (Z 0 mm), or lateral (X 0 mm) flat surfaces,where the thyroid and arytenoid cartilages are located, remain
unchanged at each muscle activation level. This boundary
condition may result in an unnaturally high stress concentration
near those planes. However, in this study, we consider themedial
portion involved in actual vibration, and the other laryngealmuscles responsible for adduction/abduction of the vocal folds
are not included.
Constitutive equation
The vocal fold is assumed to consist of a linear elastic material.
A tensor expression of Hookes law is written as
sij Cijkl3kl i;j; k; l x;y; z; (3)
where sij, 3kl, and Cijkl are the stress, strain, and stiffness ten-
sors, respectively. Because of the symmetry of the tensors, 21
independent constants appear in the stiffness tensor compo-
nents. The vocal fold is assumed to be orthotropic with two or-thogonal planes of symmetry, thereby reducing the number ofindependent constants to nine. As the fiber tissues run in the
Zdirection, the vocal fold is further assumed to be transversely
isotropic, that is, having material properties that are symmetric
about the Z-axis. Equation 3can thus be written explicitly as
whereEx(Ey)andEzare the Youngs moduli in the transverse(X or Y) and longitudinal (Z) directions, respectively;
Gyz (Gzx) is the shear modulus in the longitudinal direction;and nx (ny) and nz are the Poisson ratios in the transverseand longitudinal directions, respectively. The number of inde-
pendent constants is now five, which is consistent with Alipour
et al.
2
Because the data available on Poisson ratio are few, forsimplicity, it is assumed that nx nz (n). Consequently, fourindependent parameters (Ex, Ez, n, and Gyz) are required for
each layer.
Muscle activation
Contraction of the CT muscle shortens the distance between the
anterior parts of the cricoid and thyroid cartilages, resulting in
an elongation of the vocal fold and accumulation of tensile
strain in its fiber direction (Figure 2A). In the present study,
activation of the CT muscle is approximated by a stretching
of the entire vocal fold in the Z direction. To evaluate the
FIGURE 1. Vocal fold model.
26666664
sxx
syy
szz
syz
szx
sxy
37777775
2666666664
Ex1vz1vz
12v2zvx1vxExvxv2z
12v2zvx1vxExvz
12v2zvx0 0 0
Exvxv2z12v2zvx1vx
Ex1vz1vz
12v2zvx1vxExvz
12v2zvx0 0 0
Ezvz12v2zvxEzvz12v2zvx
Ez1vx12v2zvx 0 0 0
0 0 0 Gyz 0 00 0 0 0 Gyz 00 0 0 0 0 Ex
21vx
3777777775
2
666664
3xx
3yy
3zz
3yz
3zx
3xy
3
777775; (4)
Journal of Voice, Vol. 25, No. 6, 2011e256
-
8/12/2019 17 Deguchi
3/9
activity of the CT muscle, a parameterAct is introduced (follow-
ing previous analytical studies
9
) that ranges from 0%(no stretch) to 100% (maximum stretch). In the previous study,
maximum CT muscle activation with no TA muscle activation
was reported to yield a tensile strain of approximately 0.3 in the
vocal fold length direction. Here, the maximum stretch is
assumed to be 4 mm, corresponding to a strain of 0.29 (roughly
4 mm divided by the original or nonstress length L0of 14 mm).
Consequently, an enforced displacement DLct(04 mm) of the
following form is given to the entire vocal fold in the positive
Zdirection:
DLct 4Act=100 mm: (5)
To assess the effect of the TA muscle activation, the muscle
contraction is modeled as a shortening of the reference (or non-
stress) length in theZdirection (ie, muscle fiber direction) from
L0toL00(Figure 2B). If a muscle contracts without a constraint
that prevents a free displacement, the active shortening results
in no appearance of strain or tension even with substantive de-
crease in length. A tensile strain (3active >0) is generated if there
are constraints, that is, the thyroid and arytenoid cartilages. Inthis manner, the real TA muscle will achieve an isometric con-
traction in which a certain tension is present while keeping the
apparent length unchanged. To quantify the activity of the TA
muscle, another parameter, Ata, is introduced that ranges from
0% (no contraction) to 100% (maximum contraction). Accord-
ing to previous studies of the TA muscle,25,26 considerable
variability exists between samples in terms of active strain
that appears when the muscle fully contracts. However,
a strain of 0.20.3 seems to be reasonably typical as the
maximum value. Here, the maximum strain is assumed to be
0.3; more specifically, an active tensile strain of 0.3 (because
of internal sliding of the muscle fibers to achieve the
reference length L00 in the Z direction) is produced atAta 100% in the Z direction by the TA muscle while theoriginal length L0 in the Z direction is kept unchanged (ie,
isometric contraction). Thus,
Ata 3active100
0:3
L0 L00
L00$
100
0:3%: (6)
A resultant strain in the TA muscle, resulting from the
activities of the two muscles, is the sum of the active strain
caused by the TA muscle and passive strain caused by the
CT muscle (Figure 2B). Both strains are included in the
calculation of the strain tensor for the TA muscle. In separate
analyses, we confirmed that the resultant force caused by theactions of the two different muscles were actually a linear
sum of the active and passive components (Supplementary
Figure S1). Similarly, in previous numerical studiesthat con-
sidered the effect of active stress/strain,14,15,2730 a single
term for active strain (or stress) was linearly added to
another strain (or stress) caused by external loading. Thus,
the current model essentially follows the same approach as
in previous studies.The muscle under contraction thickens,24 and this lateral
expansion or bulging because of longitudinal shortening can
be mimicked by artificially setting new reference lengths in
the X and Y directions, which are longer than the original
lengths. The cover and ligament that line the bulging TA mus-
cle are stretched in the X-Yplane, whereas the TA muscle is
compressed in the same plane because the cover and ligament
function as a constraint that prevents the free lateral expansion
of the muscle inside. At the same time, as long as tension is
generated in the Zdirection because of longitudinal muscle
contraction, the muscle also tends to shrink within the trans-
verse plane to a certain degree, depending on the value of
the Poisson ratio. Therefore, performing a realistic simulation
of the bulging effect requires a reasonable setting of the refer-
ence lengths in theXand Ydirections to yield an expansion in
A
B
FIGURE 2. Active and passive strains. A. Tensile strain appears
when material is stretched (from i to ii). B. Free active shortening with-
out a constraint (from iii to iv) does not yield strain. If there is a con-
straint at the two ends, strain is produced in the material and is thus
under isometric contraction (v). Here, the reference length is reduced
from the original length L0 to L00, which results in the appearance of
the strain. If the material with the active strain is further stretched
externally or passively, the resulting total strain becomes the sum of
the active and passive strains (vi).
Shinji Deguchi, et al Muscle Cooperative Regulation on VF Morphology and Stress e257
-
8/12/2019 17 Deguchi
4/9
the transverse plane that exceeds the lateral shrinkage caused
by the nonzero Poisson ratio. The fact that the degree of
expansion is related to the action of the TA muscle, because
the expansion is caused by the muscle contraction, leads to
further complications. Because the data available on the
degree of expansion intrinsic to the TA muscle are few, the
lateral expansion is not included here to avoid incorporating
an arbitrary assumption. Later, at the end of the Effect onShapesection, the effect of muscle expansion is investigated
separately.
Material properties
Unless otherwise specified, the material properties given in this
section are used in the finite element models. The longitudinal
elastic modulusEzis 41.9 kPa for the membrane and 20.7 kPa
for the TA muscle, based on data from canine vocal folds in
which a linear approximation of stress-strain curves within
a strain range 00.15 was used.13 However, stress-strain curves
for human vocal ligaments have been found to be significantly
nonlinear.
31
AnEzvalue of 30 kPa is chosen for the ligament onthe basis of data approximated linearly within a strain range
00.15. We have found a quantitative measurement of the
transverse elastic modulus Ex reported by Kakita et al,32 but
their value (40 kPa) in canine cadavers seems high when com-
pared with the above longitudinal value, possibly because of
postrigor stiffening. As the transverse modulus should be lower
than the longitudinal modulus in each layer, we simply multi-
ply the value ofEzby 0.4 to approximateEx. Thus, the values of
Exused are 16 kPa for the membrane, 12 kPa for the ligament,
and 8 kPa for the TA muscle. The Poisson ratio n is assumed to
be 0.3 for all the layers, following the literature,6 and its effect
is investigated separately later on at the end of the Effect on
Shapesection. The longitudinal shear modulus Gyzis assumedto be 10 kPa for the membrane, 40 kPa for the ligament, and
12 kPa for the TA muscle, according to the literature.2 The
tension in the TA muscle of excised feline larynges was
reported to increase by at most 260% based on tetanic stimula-
tions.26 Indeed, a 260% increase in tension in the Zdirection is
achieved in the present model using Ata 100%.
Mesh sensitivity
The mesh sensitivity was examined to ensure that the results
were independent of the size of the computational mesh.
Three mesh sizes were used: the coarse, medium, and fine
meshes consisting of 1049, 2306, and 3025 nodes, respec-tively, distributed over 1719, 3619, and 4724 tetrahedron
elements in the membrane; 872, 2857, and 5180 nodesdistributed over 1371, 3494, and 5700 tetrahedron elements
in the ligament; and 746, 905, and 1784 nodes distributed
over 828, 1033, and 1828 tetrahedron elements in the TA
muscle. The three meshes gave similar patterns of results.
At all muscle activation levels, the maximum difference
among the three computational meshes was 0.5% in thick-ness (defined below in Figure 3A). Therefore, all the results
presented in the remainder of this article are obtained with
the medium mesh.
RESULTS AND DISCUSSION
Effect on shape
In this section, we investigate the effect of muscle activation on
vocal fold morphology and normal stress in the Zdirection
(Figure 4). With an increase in Act, which stretches the entire
vocal fold in the Zdirection, tensional stress is generated in
all the layers. In addition, the middle coronal section shrinks
FIGURE 3. Definitions of parameters. A. Thickness and radius ofcurvature, used in Figures 6 and 9. B. Abscissas used in Figures
5 and 10.
Journal of Voice, Vol. 25, No. 6, 2011e258
-
8/12/2019 17 Deguchi
5/9
because of the nonzero Poisson ratio because a stretched mate-
rial tends to shrink in the directions transverse to the direction
of stretching. Meanwhile, with an increase in Ata, which con-
tracts the TA muscle and then produces tension in the Zdirec-
tion, a remarkable increase in stress appears only in the TAmuscle. The stress in the cover/ligament is reduced because lat-
eral shrinkage because of muscle contraction and the nonzero
Poisson ratio compresses the cover/ligament lying alongside
the muscle in the X-Y plane. The compression within the X-Y
plane causes the cover/ligament to expand in the Zdirection
to partially retain the volume; however, as the straight-line dis-
tance between the most anterior and posterior boundaries is
kept unchanged at each Act, the longitudinal tissue expansion
locally relaxes the magnitude of the tensile stress in theZdirec-
tion and may even produce compressive stress as discussed later
in theEffect on stress distributionsection andFigure 5. Thus,
even if a certain set value of longitudinal strain is given accord-
ing to the combination ofAta andAct (Figure 2), the actual stress
observed is different from the set value, particularly around the
peripheral region of the vocal folds containing the cover and
ligament (Figure 4).As muscle contraction proceeds, the tip of the vocal fold
(where vibration primarily occurs in real phonation) becomes
thinner and curves upward, exactly as seen in the detailed obser-
vations of human vocal fold geometry.33 To quantify these
changes, two parameters, thickness and radius of curvature
(Figure 3A), are measured. The thickness is defined as the ver-
tical length of the vocal fold at the leftmost edge of the TA mus-
cle, and the radius of curvature is defined as the radius of
curvature of an imaginary circle tangent to the lowermost por-
tion of the upper line defining the vocal fold. Both Actand Atadecrease the thickness (Figure 6A), and the radius of curvature
FIGURE 4. Vocal fold morphology and normal stress in Zdirection as a function of muscle activation characterized by parameters Actand Ata.
Because the deformation is symmetric with respect to the middle coronal section, only the rear half of the vocal fold is shown. Red represents a stress
magnitude of more than 8 kPa.
Shinji Deguchi, et al Muscle Cooperative Regulation on VF Morphology and Stress e259
-
8/12/2019 17 Deguchi
6/9
(Figure 6B) indicating that the tip region curves upward. The
decrease in the thickness is caused by the nonzero Poisson ratiobecause a tensed elastic material shrinks compared with the
nonstress state in the directions transverse to the direction of
tension. The decrease in the radius of curvature has two distinct
causes. Shrinkage deformation in the coronal section is in-
hibited at the rightmost surface of the model. This makes the
upper portion of the vocal fold curved, leading to a reduction
in the radius of curvature. In addition, a characteristic stress
distribution because of geometric asymmetry contributes to
the decrease in the radius of curvature. This fact is particularly
explained inFigure 7, which shows the normal stresses in the
Xand Ydirections in the midcoronal section under the maxi-
mum values ofActand Ata. The upper portion of the cover/lig-
ament is compressed in theXdirection because of shrinkage in
the X-Yplane (as shown by the blue area in Figure 7A). The
lower portion of the same layers running obliquely in the planeis also compressed to a similar extent. On the other hand, in the
Ydirection, only the lower tip portion is compressed. The upper
portion is not compressed in the Ydirection (as shown by the
green area in Figure 7B) because it runs perpendicular to
the Y direction. Because of this geometric asymmetry, only
the lower edge resists deformation in the Ydirection. Conse-
quently, a concave curve appears in the upper portion where
there is less resistance to deformation in the Ydirection.
An interesting question is which ofActandAtahas a stronger
effect on inducing the characteristic morphological change? As
mentioned at the end of the Muscle Activation section, lateral
expansion of the TA muscle because of longitudinal contraction
was not included in the deformation shown inFigure 4. A sep-arate analysis demonstrates that the lateral expansion could
suppress the morphological change caused by longitudinal con-
traction withAta 100% (Figure 8). Here, 3exrepresents an ac-tive expansion strain applied to the TA muscle in bothXand Y
directions by setting new reference lengths for the two direc-
tions that are longer than the original reference lengths. At
3ex 0.25, the decrease in the volume of the TA muscle becauseof longitudinal contraction is approximately equal to the in-
crease in volume because of lateral expansion. Therefore, if
the actual TA muscle contracts while maintaining its volume,a case where 3ex 0.25 would simulate a realistic deformation.A comparison of the cases for 3ex 0.25 and Ata 100%(Figure 8) and 3ex 0 and Ata 0% (shown in the upper leftinFigure 4) shows that the stress magnitude is elevated only
in the muscle in the former case, and the shape of the vocal
fold remains almost unchanged. Taken together, the results de-
scribed so far show that CT muscle activation makes the tip re-
gion thinner and upward oriented. In contrast, a similarmorphological shrinkage caused by the TA muscle would be
suppressed in reality because of the lateral expansion discussed
above.
The Poisson ratio was set to 0.3 throughout the calculations
shown here, as mentioned in the Material Properties section;
however, we confirmed the effect of Poisson ratio on thickness
in separate analyses (Figure 9). As the Poisson ratio increases,the effect ofAtaon thickness increases because a low Poisson
ratio does not significantly affect the lateral deformation (or
thickness measured within theX-Yplane) even with the longi-
tudinal stretching caused byAta. As expected, the cross section
around the middle coronal section under longitudinal tension
shrinks as the Poisson ratio increases.
Effect on stress distribution
Next, we investigate the normal stress in the Z direction
(Figure 5) in the element immediately beneath the vocal fold
profile, as defined inFigure 3B. AtAct 0%,Atahas virtually
FIGURE 5. Stress distribution in cover as a function of muscle
activation.
FIGURE 6. Effect of muscle activation on thickness (A)and radius
of curvature (B). Note that the valuesof radius of curvature atAct 0%
andAta 0% are infinite in magnitude and are hence omitted from the
graph.
Journal of Voice, Vol. 25, No. 6, 2011e260
-
8/12/2019 17 Deguchi
7/9
no influence on the stress magnitude around the tip region,
where significant vibration occurs in real phonation (at about
8 mm along the abscissa in Figure 5), because only a minor
stress induced by Ata in the TA muscle is transmitted to the
cover. In the upper (10 mm, subglottis) regions, Ata induces considerable com-pression in the cover within theX-Yplane (Figure 7). This trans-
verse compression also causes a local relaxation of the tension
in theZdirection because the longitudinal straight-line distance
between the most anterior and posterior boundaries is kept con-
stant at eachAct. With an increase in Act, the stress magnitude is
elevated over the vocal fold, and then the compression isreplaced by positive tensional stress. Thus, although Act and
Ata show distinct effects, they cooperatively allow regulation
of tension in the cover.
To assess how the cooperation betweenActandAtaaffects the
expected effective tissue depth, the distribution of normal stress
in the Zdirection across a line from the medial border to the
lateral margin (Figure 3B) is investigated (Figure 10). An in-
crease in Act with a constant value ofAta (33%) increases the
stress magnitude almost equally in all layers (Figure 10A).
The relative magnitude in the layers is determined essentially
by the longitudinal Youngs modulus and Ata. The effect of
Ata on the stress along the same line is examined with
Act 100% (Figure 10B). As TA muscle activation proceeds,tensile stress builds in the muscle, whereas the magnitude of
the stress in the ligament decreases. Therefore, vibration may
be limited to the surface region at high values ofAta.
As the Youngs modulus of the cover is higher than that of the
ligament in this analysis, significant increases in depth becauseof TA muscle contraction appear almost entirely in the liga-
ment, and hence, the magnitude of the stress in the cover
remains relatively unchanged. In real vocal fold tissues, nonlin-
earity ofthe stress-strain curve is most profound for the vocal
ligament,31 and hence, the incremental elastic modulus of the
ligament would in fact increase as the Act-induced stretch
FIGURE 7. Stresses at Act 100% andAta 100%.A. Stress in Xdirection.B. Stress in Ydirection.
FIGURE 8. Effect of lateral expansion of TA muscle because of longitudinal contraction.
Shinji Deguchi, et al Muscle Cooperative Regulation on VF Morphology and Stress e261
-
8/12/2019 17 Deguchi
8/9
proceeds. Therefore, when the TA muscle is not activated, most
of the longitudinal stress at high strains (ie, for large values of
Act) would be borne by the ligament. This will cause the cover
to loosen, thereby facilitating vibrations localized near thesurface. Taken together, the analyses suggest that the TA mus-
cle regulates how deep the vibration reaches into the tissue. The
CT muscle would also regulate the effective depth, but it takes
advantage of different nonlinear stress-strain relationships for
each layer because the principal stress-bearing component
could be replaced by stretching the vocal fold.
CONCLUSIONS
In previous studies, changes in longitudinal vocal fold length in
response to muscle activation were investigated using simple
analytical models,1,9,10,17 and the expected effective depth
of tissue vibration and its role in regulating the voicefundamental frequency were discussed. The present study
provides not only results consistent with those of previous
studies but also more detailed explanations and new findings.
In particular, given that the three-dimensional vocal fold mor-
phology and stress distributionwhich the previous analytical
studies had no means to obtainmust be one of the major
determinants of the effective depth, their close relationship
with muscle activation revealed here would be important in
evaluating the effect of muscles on the vibrational behavior
and resultant voice quality. Moreover, active and passive
stresses, which were produced in the vocal folds by the TA
and CT muscles, respectively, were considered separately inthe present analysis. This distinction between active and passive
contributions has not typically been established in previous
modeling studies, even when elongation in the longitudinal
vocal fold length was incorporated (eg, study by Tao and
Jiang7). The present study, thus, provides a likely mechanism
regarding the two muscles cooperative regulation of the effec-
tive depth of tissue vibration.The present results show that the activity of both muscles
makes the tip region of the vocal fold thinner and curve upward.
However, lateral expansion of the TA muscle because of its
longitudinal contraction suppresses the morphological change
induced by the TA muscle. These analyses suggest that the char-
acteristic morphology, which seems suitable for efficient vibra-
tion,34 is realized primarily by the CT muscle in actual vocal
folds. The CT muscle activation increases the stress magnitude
in each layer to an equal degree because only linear analyses are
performed in this study. If nonlinear stress-strain relationships
were incorporated instead, the ligament that has the most
FIGURE 9. Effect of Poisson ratio on thickness (A) and radius of
curvature (B).
FIGURE 10. Stress distribution in depth direction of vocal fold
model.A. Effect ofActat Ata 33%.B. Effect ofAtaat Act 33%.
Journal of Voice, Vol. 25, No. 6, 2011e262
-
8/12/2019 17 Deguchi
9/9
profound nonlinearityand is firmly attached to themuscle would
become the principal stress-bearingcomponent. Because the CT
muscle regulates the vocal fold, the principal stress-bearing
component and resulting effective depth of tissue vibration
vary according to the incremental elastic moduli of the layers.
In contrast, TA muscle activation produces a stress barrier
at the interface between the muscle and the ligament. This bar-
rier prevents the enlargement of movable portions of the vocalfold. When the TA muscle is relaxed, a relatively flat stress
distribution is obtained over the tissue. In this case, tissue
motions occur even in deep layers including the muscle, and
the regulation of the effective depth by the CT muscle described
above would dominate. When the TA muscle is sufficiently
activated and becomes the principal stress-bearing component,
tissue motion is limited to the surface region. Thus, the present
analyses demonstrate the relationship between the CT and
TA muscles in their cooperative regulation of the vocal fold
morphology, stress, and possibly effective depth as well. The
results obtained in this study should be useful in future stress
estimations in various phonation situations.
Acknowledgments
This work was supported by the Japan Science and Technology
Agency and the Ono Acoustics Research Fund.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jvoice.2010.11.006.
REFERENCES1. Titze IR.Principles of Voice Production. Upper Saddle River, NJ: Prentice-
Hall; 1994:1519.
2. Alipour F, Berry DA, Titze IR. A finite-element model of vocal-fold vibra-tion.J Acoust Soc Am. 2000;108:30033012.
3. Rosa Mde O, Pereira JC, Grellet M, Alwan A. A contribution to simulating
a three-dimensional larynx model using the finite elementmethod.J Acoust
Soc Am. 2003;114:28932905.
4. Gunter HE. Modeling mechanical stresses as a factor in the etiology of
benign vocal fold lesions. J Biomech. 2004;37:11191124.
5. Tao C, Jiang JJ, Zhang Y. Simulation of vocal fold impact pressures with
a self-oscillating finite-element model. J Acoust Soc Am. 2006;119:
39873994.
6. Tao C, Jiang JJ. Mechanical stress during phonation in a self-oscillating
finite-element vocal fold model. J Biomech. 2007;40:21912198.
7. Tao C, Jiang JJ. A self-oscillating biophysical computer model of the
elongated vocal fold.Comput Biol Med. 2008;38:12111217.
8. Hirano M. Morphological structure of the vocal cord as a vibrator and its
variations.Folia Phoniatr. 1974;26:8994.
9. Titze IR, Jiang JJ, Druker DG. Preliminaries to the body-cover theory of
pitch control. J Voice. 1988;1:314319.
10. Titze IR, Luschei ES, Hirano M. The role of the thyroarytenoid muscle in
regulation of fundamental frequency.J Voice. 1989;3:213224.
11. Story BH, Titze IR. Voice simulation with a body-cover model of the vocal
folds.J Acoust Soc Am. 1995;97:12491260.
12. Kataoka K, Kitajima K. Effects of lengthand depth of vibrationof thevocal
folds on the relationship between transglottal pressure and fundamental
frequency of phonation in canine larynges. Ann Otol Rhinol Laryngol.
2001;110:556561.
13. Alipour F, Titze IR. Elastic models of vocal fold tissues. J Acoust Soc Am.
1991;90:13261331.
14. Ikeda T, Matsuzaki Y, Aomatsu T. A numerical analysis of phonation using
a two-dimensional flexible channel model of the vocal folds. J Biomech
Eng. 2001;123:571579.
15. Deguchi S, Matsuzaki Y, Ikeda T. Numerical analysis of effects of trans-
glottal pressure change on fundamental frequency of phonation. Ann Otol
Rhinol Laryngol. 2007;116:128134.
16. Hirano M. Phonosurgery: basic and clinical investigation. Otologia
(Fukuoka). 1975;21:239260.
17. Farley GR. A quantitative model of voice F0 control. J Acoust Soc Am.
1994;95:10171029.
18. Kucinschi BR, Scherer RC, DeWitt KJ, Ng TT. Flow visualization and
acoustic consequences of the air moving through a static model of the
human larynx. J Biomech Eng. 2006;128:380390.
19. Deguchi S, Hyakutake T. Theoretical consideration of the flow behavior in
oscillating vocal fold. J Biomech. 2009;42:824829.
20. Bailey BJ, Johnson JT, Newlands SD.Head & Neck SurgeryOtolaryngol-
ogy. Philadelphia, PA: Lippincott Williams & Wilkins; 2006:17211726.
21. Fritsch H, Kuhnel WIn:Color Atlas of Human Anatomy: Internal Organs,
Vol. 2. New York, NY: Thieme; 2007:108117.22. Ohashi M, Ota F, Ochiai A, Komori A, Takayanagi H, Hesaka H,
Moriyama H. Examination of atelocollagen implantation validity by max-
imum phonation time and mean airflow rate. Nihon Kikan Shokudoka
Gakkai Kaiho. 2007;58:365370.
23. Zheng X, Bielamowicz S, Luo H, Mittal R. A computational study of the
effect of false vocal folds on glottal flow and vocal fold vibration during
phonation. Ann Biomed Eng. 2009;37:625642.
24. Alipour F, Scherer RC. Vocal fold bulging effects on phonation using a bio-
physical computer model. J Voice. 2000;14:470483.
25. Alipour F, Titze IR, Perlman AL. Tetanic contraction in vocal fold muscle.
J Speech Hear Res. 1989;32:226231.
26. Johns MM, Urbanchek M, Chepeha DB, Kuzon WM Jr, Hogikyan ND.
Length-tension relationship of the feline thyroarytenoid muscle. J Voice.
2004;18:285291.
27. Deguchi S, Fukamachi H, Hashimoto K, Iio K, Tsujioka K. Measurementand finite element modeling of the force balance in the vertical section of
adhering vascular endothelial cells. J Mech Behav Biomed Mater. 2009;
2:173185.
28. Hunter EJ, Titze IR, Alipour F. A three-dimensional model of vocal fold
abduction/adduction.J Acoust Soc Am. 2004;115:17471759.
29. Napadow VJ, Kamm RD, Gilbert RJ. A biomechanical model of sagittal
tongue bending. J Biomech Eng. 2002;124:547556.
30. Stops AJ, McMahon LA, OMahoney D, Prendergast PJ, McHugh PE. A
finite element prediction of strain on cells in a highly porous collagen-
glycosaminoglycanscaffold.J BiomechEng. 2008;130:061001-1061001-11.
31. Min YB, Titze IR, Alipour F. Stress-strain response of the human vocal
ligament.Ann Otol Rhinol Laryngol. 1995;104:563569.
32. KakitaY, Hirano M, OhmaruK. Physical properties of thevocal fold tissue:
measurement on excised larynges In: Stevens K, Hirano M, eds. Vocal Fold
Physiology. Tokyo, Japan: University of Tokyo Press; 1981:377397.
33. Sidlof P, Svec JG, Horacek J, Vesely J, Klepacek I, Havlk R. Geom-
etry of human vocal folds and glottal channel for mathematical and
biomechanical modeling of voice production. J Biomech. 2008;41:
985995.
34. Hollien H. Vocal fold thickness and fundamental frequency of phonation.
J Speech Hear Res. 1962;5:237243.
Shinji Deguchi, et al Muscle Cooperative Regulation on VF Morphology and Stress e263
http://dx.doi.org/doi:10.1016/j.jvoice.2010.11.006http://dx.doi.org/doi:10.1016/j.jvoice.2010.11.006