Feasible Ranges of Muscle Activity Quantify ...stanford.edu/~cssimps/WCB_Simpson_10.pdfAdduction...
1
Adduction GMED Hip Joint Torques (N−m) 20 −60 0 −20 −40 0 50 100 −5 0 5 Joint Angle (Degrees) 0 0.5 1 GMED2 GMED1 0 0.5 1 Hip Abduction Feasible Ranges Hip Abduction 0 50 100 Abduction Adduction Joint Angles/Torques Frontal Plane Right Leg Left Leg Time (% of Gait Cycle) Time (% of Gait Cycle) 0 0.5 1 RF 0 0.5 1 BFLH SM Extension Flexion 0 0.5 1 VI VL VM Extension MGAS LGAS BFSH 0 0.5 1 Flexion 0 50 100 0 0.5 1 SOL Plantarflexors 0 0.5 1 TA Dorsiflexors −20 −10 0 10 20 Joint Angle (Degrees) Joint Torques (N−m) 40 −40 0 Extension BFLH SM RF Hip −60 −40 −20 0 Joint Angle (Degrees) Joint Torques (N−m) 20 −60 0 -20 -40 Extension MGAS LGAS BFLH SM RF VM, VL, VI BFSH Knee Dorsiflexion TA MGAS LGAS SOL −5 0 5 10 Joint Angle (Degrees) Joint Torques (N−m) 0 0 50 100 −100 -50 Ankle Time (% of Gait Cycle) Time (% of Gait Cycle) Hip Knee Ankle 0 0.5 1 Plantarflexion/Flexion Feasible Ranges of Muscle Activation Stance Swing Upper Bound Lower Bound Experimental EMG Data Extension Flexion Stance Swing Flexion Plantarflex. Dorsiflex. Right Leg Left Leg Joint Angles/Torques Sagittal Plane Right Leg Left Leg Right Leg Left Leg [5] Time (% of Gait Cycle) Time (% of Gait Cycle) , Feasible Ranges of Muscle Activity Quantify Musculoskeletal Redundancy in Human Walking Cole Simpson 1 , M. Hongchul Sohn 1 , Jessica Allen 2 , and Lena H. Ting 1,2 1 School of Mechanical Engineering, Georgia Tech 2 Department of Biomedical Engineering, Georgia Tech and Emory University Musculoskeletal redundancy [1] allows for an infinite num- ber of combinations of muscle activation patterns for per- forming a task and is frequently resolved in musculoskel- etal modeling using optimization techniques. Optimization methods select a single set of muscle activations from the entire range of possible solutions to satisfy physiologically based criteria, such as minimizing muscle stress [2,3] . How- ever, such techniques do not account for the variability that is commonly observed in many motor tasks such as walk- ing, both within and across individuals [4,5,6] . Accordingly, optimal muscle activation solutions frequently deviate from experimentally recorded patterns [4,5,7] . How does the ner- vous system select these variable muscle activations? Are consistent trends the deterministic result of biomechanics or common neural strategies? In order to better understand the role that biomechanical versus neural constraints play in shaping muscle activation patterns for movement, the full range of possible muscle activation patterns based on biomechanics must first be defined. Several methods for computing biomechanical limitations have been explored in static simulations [8, 9, 10] . Our objective was to identify biomechanical limitations on muscle activation dur- ing a dynamic human walking task. Methods Introduction Results 1. Bernstein, N., 1967. The Coordination and Regulation of Movements. Pergamon Press, New York. 2. Crowninshield, R.D. and Brand, R.A., 1981. A physiologically based criterion of muscle force prediction in locomotion. Journal of Biomechanics 14, 793–801. 3. Thelen, D.G., 2003. Adjustment of muscle mechanics model parameters to simulate dynamic contractions in older adults. J Biomech Eng 125, 70-77. 4. Liu, M.Q., Anderson, F.C., Schwartz, M.H., Delp, S.L., 2008. Muscle contributions to support and progression over a range of walking speeds. Journal of biomechanics 41, 3243-3252. 5. van der Krogt, M.M., Delp, S.L., Schwartz, M.H., 2012. How robust is human gait to muscle weakness? Gait Posture 36, 113-119 6. Winter, D.A., Yack, H.J., 1987. EMG profiles during normal human walking: stride-to-stride and inter- subject variability. Electroencephalogr Clin Neurophysiol 67, 402-411. 7. Buchanan, T.S. and Shreeve,D.A.,1996. An evaluation of optimization techniques for prediction of muscle activation patterns during isometric tasks. Journal of Biomechanical Engineering 118, 565–574. 8. Kutch, J.J., Valero-Cuevas, F.J., 2011. Muscle redundancy does not imply robustness to muscle dysfunction. Journal of biomechanics 44, 1264-1270. 9. Martelli, S., Calvetti, D., Somersalo, E., Viceconti, M., Taddei, F., 2013. Computational tools for calculating alternative muscle force patterns during motion: a comparison of possible solutions. Journal of biomechanics 46, 2097-2100. 10. Sohn, M.H., et al., 2013. Defining feasible bounds on muscle activation in a redundant biomechanical task: practical implications of redundancy. Journal of Biomechanics 46, 1363-1368. 11. John, C.T., et al., 2013. Stabilisation of walking by intrinsic muscle properties revealed in a three- dimensional muscle-driven simulation. Comput Methods Biomech Biomed Engin 16, 451-462. 12. Delp. S.L., 1990. Surgery simulation: A computer-graphics system to analyze and design musculoskeletal reconstructions of the lower limb. Stanford University, Ph.D. Thesis. 13. Delp, S. L., et al., 2007. OpenSim: open-source software to create and analyze dynamic simulations of movement. IEEE Trans Biomed Eng 54(11): 1940-1950. Ting Lab: Jeff Bingham, PhD; Stacie Chvatal, PhD; Lucas McKay, PhD; Andrew Sawers, PhD; Darren Bolger; Morris Huang; Kyle Blum; Harrison Bartlett Funding: Air Products Undergraduate Research Award Conclusions • Wide feasible ranges suggest that the selection of mus- cle activations is not the deterministic result of bio- mechanical considerations for this submaximal task ◦ Large amounts of variability are permitted in a redun- dant system (compared to finger model [8] ) ◦ Additional neural constraints are needed to select unique muscle activations • Sensitivity of feasible ranges to biomechanical variations is consistent with previous studies [7] • Small variations in experimental data compared with fea- sible ranges suggest that consistent neural strategies are used • Feasible ranges quantify the redundancy of muscle acti- vations during a dynamic task ◦ This method reveals the complete range of biomechan- ically allowed variability in muscle activity during a dy- namic task ◦ Additional constraints will further reduce the feasible ranges (ex: muscle synergies) to more closely predict experimental muscle activation ranges Feasible ranges were generally wide • Most muscles (73%) were not limited by their biomechanics at any time during the gait cycle (upper bound = 1, lower bound = 0) • Only two muscles were necessary (lower bound > 0 at any time point): the left TA and the left GMED1 • Most muscles (76%) were unconstrained (upper bound = 1 for every time point) Differences were observed in the feasible ranges of the right and left legs • The musculoskeletal model is symmetric • Differences were observed between the feasible ranges of the right and left legs (ex: GMED2) • Differences in feasible ranges between right and left legs are due to variations in joint angles and joint torques between the legs Experimental variability was much less than that permitted by the biomechanics • Experimental EMG data was superimposed on the feasible ranges for available muscles [5] • Variability in experimental EMGs was much less than that allowed by the biomechanics (ex: RF) Contact: [email protected] BS7 Muscle Name Abbreviation Biceps femoris long head BFLH Biceps femoris short head BFSH Gluteus medius GMED Lateral gastrocnemius LG Medial gastrocnemius MG Rectus femoris RF Semimembranossus SM Soleus SOL Tibialis anterior TA Vastus intermedius VI Vastus lateralis VL Vastus medialis VM Acknowledgments • Minimum muscle stress solution ( O): 50% m 1 and 0% m 2 Biomechanical Constraints on Variability: Redundant System: Leg with two opposing mus- cles, one large (m1) and one small (m2), acting on a knee with 1 degree of freedom Objective: Produce a flexion torque equal to half of the maximum torque producible by the system Upper and lower bounds define limits of variation (Feasible Range): • Upper bound (maximum possible activation) limited by relative strength of the antagonistic muscles Feasible limits can be determined for each measured time point during a dynamic task: • Upper bound indicates whether muscle is constrained (<1) or unconstrained (=1) Optimization yields one solution from many possibilities: • Time point B shows a narrow feasible range, which does not permit a lot of variability m1 m2 1/2 Maximum Torque 0 1 0.5 0 1 0.5 1 0 1 0 X X X X X X X X m1 (large muscle) m2 (small muscle) Normalized Joint Torque Muscle Activation e m1 e m2 O O Normalized Joint Torque 1 2 0 0.5 1 0 0.5 1 0 1 0 m1 m2 e m1 e m e e Upper Bound Lower Bound Feasible Range Normalized Joint Torque Muscle Activation Normalized Joint Torque 0.5 1 0 X X X X X X X X Muscle Activation Time A B References Experimental Data Scaling Information Kinematic Information Musculoskeletal Model Model Outputs Computing Feasible Ranges John et al. 2012 OpenSim 2392 Joint Angles Inverse Kinematics Muscle Parameters 2392 Model [12, 13] Ground Reaction Forces Joint Torques Inverse Dynamics Linear Programming linprog.m, Matlab • Lower bound indicates whether a muscle is optional (=0) or necessary (>0) • Time point A shows a wide feasible range, which permits a large amount of variability • Lower bound (minimum possible activation) determined by necessity for the task • Measured muscle activities (X) may vary significantly from O
Transcript of Feasible Ranges of Muscle Activity Quantify ...stanford.edu/~cssimps/WCB_Simpson_10.pdfAdduction...
Adduction
GMED
Hip
Join
t Tor
ques
(N−m
)
20
−60
0−20−40
0 50 100
−5
0
5
Join
t Ang
le (D
egre
es)
0
0.5
1
GMED2
GMED1
0
0.5
1
Hip
Abd
uctio
n
Feasible Ranges
Hip
Abd
uctio
n
0 50 100
Abd
uctio
n A
dduc
tion
Joint Angles/Torques
Frontal Plane
Right Leg Left Leg
Time (% of Gait Cycle)Time (% of Gait Cycle)
0
0.5
1RF
0
0.5
1BFLH SM
Ext
ensi
onFl
exio
n
0
0.5
1VI VL VM
Ext
ensi
on
MGAS LGAS
BFSH
0
0.5
1
Flex
ion
0 50 1000
0.5
1SOL
Pla
ntar
flexo
rs
0
0.5
1TA
Dor
sifle
xors
−20
−10
0
10
20
Join
t Ang
le (D
egre
es)
Join
t Tor
ques
(N−m
)
40
−40
0
Extension
BFLHSM
RF
Hip
−60
−40
−20
0
Join
t Ang
le (D
egre
es)
Join
t Tor
ques
(N−m
)
20
−60
0
-20
-40
ExtensionMGASLGAS
BFLHSM
RF
VM, VL,VI
BFSH
Kne
e
Dorsiflexion
TA
MGASLGAS
SOL
−5
0
5
10
Join
t Ang
le (D
egre
es)
Join
t Tor
ques
(N−m
)
0
0 50 100
−100
-50
Ank
le
Time (% of Gait Cycle) Time (% of Gait Cycle)
Hip
Kne
eA
nkle
0
0.5
1
Pla
ntar
flexi
on/F
lexi
on
Feasible Ranges of Muscle Activation
Stance Swing Upper Bound
Lower BoundExperimental EMG Data
Ext
ensi
onFl
exio
n Stance Swing
Flex
ion
Pla
ntar
flex.
Dor
sifle
x.
Right Leg Left Leg
Joint Angles/Torques
Sagittal Plane
Right Leg Left Leg
Right Leg Left Leg
[5]
Time (% of Gait Cycle)
Time (% of Gait Cycle)
,
Feasible Ranges of Muscle Activity Quantify Musculoskeletal Redundancy in Human Walking
Cole Simpson1, M. Hongchul Sohn1, Jessica Allen2, and Lena H. Ting1,2 1 School of Mechanical Engineering, Georgia Tech 2 Department of Biomedical Engineering, Georgia Tech and Emory University
Musculoskeletal redundancy[1] allows for an infinite num-ber of combinations of muscle activation patterns for per-forming a task and is frequently resolved in musculoskel-etal modeling using optimization techniques. Optimization methods select a single set of muscle activations from the entire range of possible solutions to satisfy physiologically based criteria, such as minimizing muscle stress[2,3]. How-ever, such techniques do not account for the variability that is commonly observed in many motor tasks such as walk-ing, both within and across individuals[4,5,6]. Accordingly, optimal muscle activation solutions frequently deviate from experimentally recorded patterns[4,5,7]. How does the ner-vous system select these variable muscle activations? Are consistent trends the deterministic result of biomechanics or common neural strategies? In order to better understand the role that biomechanical versus neural constraints play in shaping muscle activation patterns for movement, the full range of possible muscle activation patterns based on biomechanics must first be defined. Several methods for computing biomechanical limitations have been explored in static simulations[8, 9, 10]. Our objective was to identify biomechanical limitations on muscle activation dur-ing a dynamic human walking task.
Methods
Introduction Results
1. Bernstein, N., 1967. The Coordination and Regulation of Movements. Pergamon Press, New York.2. Crowninshield, R.D. and Brand, R.A., 1981. A physiologically based criterion of muscle force prediction
in locomotion. Journal of Biomechanics 14, 793–801.3. Thelen, D.G., 2003. Adjustment of muscle mechanics model parameters to simulate dynamic
contractions in older adults. J Biomech Eng 125, 70-77.4. Liu, M.Q., Anderson, F.C., Schwartz, M.H., Delp, S.L., 2008. Muscle contributions to support and
progression over a range of walking speeds. Journal of biomechanics 41, 3243-3252.5. van der Krogt, M.M., Delp, S.L., Schwartz, M.H., 2012. How robust is human gait to muscle weakness?
Gait Posture 36, 113-119
6. Winter, D.A., Yack, H.J., 1987. EMG profiles during normal human walking: stride-to-stride and inter-subject variability. Electroencephalogr Clin Neurophysiol 67, 402-411.
7. Buchanan, T.S. and Shreeve,D.A.,1996. An evaluation of optimization techniques for prediction of muscle activation patterns during isometric tasks. Journal of Biomechanical Engineering 118, 565–574.
8. Kutch, J.J., Valero-Cuevas, F.J., 2011. Muscle redundancy does not imply robustness to muscle dysfunction. Journal of biomechanics 44, 1264-1270.
9. Martelli, S., Calvetti, D., Somersalo, E., Viceconti, M., Taddei, F., 2013. Computational tools for calculating alternative muscle force patterns during motion: a comparison of possible solutions. Journal of biomechanics 46, 2097-2100.
10. Sohn, M.H., et al., 2013. Defining feasible bounds on muscle activation in a redundant biomechanical task: practical implications of redundancy. Journal of Biomechanics 46, 1363-1368.
11. John, C.T., et al., 2013. Stabilisation of walking by intrinsic muscle properties revealed in a three-dimensional muscle-driven simulation. Comput Methods Biomech Biomed Engin 16, 451-462.
12. Delp. S.L., 1990. Surgery simulation: A computer-graphics system to analyze and design musculoskeletal reconstructions of the lower limb. Stanford University, Ph.D. Thesis.
13. Delp, S. L., et al., 2007. OpenSim: open-source software to create and analyze dynamic simulations of movement. IEEE Trans Biomed Eng 54(11): 1940-1950.
Ting Lab: Jeff Bingham, PhD; Stacie Chvatal, PhD; Lucas McKay, PhD; Andrew Sawers, PhD; Darren Bolger; Morris Huang; Kyle Blum; Harrison Bartlett
Funding: Air Products Undergraduate Research Award
Conclusions• Wide feasible ranges suggest that the selection of mus-
cle activations is not the deterministic result of bio-mechanical considerations for this submaximal task
◦ Large amounts of variability are permitted in a redun-dant system (compared to finger model[8]) ◦ Additional neural constraints are needed to select unique muscle activations
• Sensitivity of feasible ranges to biomechanical variations is consistent with previous studies[7]
• Small variations in experimental data compared with fea-sible ranges suggest that consistent neural strategies are used
• Feasible ranges quantify the redundancy of muscle acti-vations during a dynamic task
◦ This method reveals the complete range of biomechan-ically allowed variability in muscle activity during a dy-namic task ◦ Additional constraints will further reduce the feasible ranges (ex: muscle synergies) to more closely predict experimental muscle activation ranges
Feasible ranges were generally wide
• Most muscles (73%) were not limited by their biomechanics at any time during the gait cycle (upper bound = 1, lower bound = 0)
• Only two muscles were necessary (lower bound > 0 at any time point): the left TA and the left GMED1
• Most muscles (76%) were unconstrained (upper bound = 1 for every time point)
Differences were observed in the feasible ranges of the right and left legs• The musculoskeletal model is symmetric• Differences were observed between the feasible
ranges of the right and left legs (ex: GMED2)• Differences in feasible ranges between right and left
legs are due to variations in joint angles and joint torques between the legs
Experimental variability was much less than that permitted by the biomechanics• Experimental EMG data was superimposed on the
feasible ranges for available muscles[5]
• Variability in experimental EMGs was much less than that allowed by the biomechanics (ex: RF)
Contact: [email protected] BS7
Muscle Name AbbreviationBiceps femoris long head BFLHBiceps femoris short head BFSHGluteus medius GMEDLateral gastrocnemius LGMedial gastrocnemius MGRectus femoris RFSemimembranossus SMSoleus SOLTibialis anterior TAVastus intermedius VIVastus lateralis VLVastus medialis VM
Acknowledgments
• Minimum muscle stress solution (O): 50% m1 and 0% m2
Biomechanical Constraints on Variability:
Redundant System: Leg with two opposing mus-cles, one large (m1) and one small (m2), acting on a knee with 1 degree of freedom
Objective: Produce a flexion torque equal to half of the maximum torque producible by the system
Upper and lower bounds define limits of variation (Feasible Range):• Upperbound(maximum
possible activation) limited by relative strength of the antagonistic muscles
Feasible limits can be determined for each measured time point during a dynamic task:• Upperboundindicates whether muscle is
constrained (<1) or unconstrained (=1)
Optimization yields one solution from many possibilities:
• Time point B shows a narrow feasible range, which does not permit a lot of variability
m1
m2
1/2 MaximumTorque
0 10.5 0 10.5
1
0
1
0
X
X
X
X
XXXX
m1 (large muscle) m2 (small muscle)
Normalized JointTorque
Mus
cle
Act
ivat
ion
e m1
e m2
O
O
Normalized JointTorque
1
2
0 0.5 1 0 0.5
1
0
1
0
m1 m2
e m1
e mee
Upper Bound
Lower Bound
Feasible Range
Normalized Joint Torque
Mus
cle
Act
ivat
ion
Normalized Joint Torque
0.5
1
0
X
X
X
X
XXXX
Mus
cle
Act
ivat
ion
Time
A
B
References
Experimental Data
Scaling Information
KinematicInformation
Musculoskeletal Model Model Outputs Computing Feasible RangesJohn et al. 2012 OpenSim 2392
Joint AnglesInverse Kinematics
Muscle Parameters2392 Model
[12, 13]
Ground ReactionForces
Joint TorquesInverse Dynamics
Linear Programminglinprog.m, Matlab
• Lowerboundindicates whether a muscle is optional (=0) or necessary (>0)
• Time point A shows a wide feasible range, which permits a large amount of variability
• Lowerbound(minimum possible activation) determined by necessity for the task
• Measured muscle activities (X) may vary significantly from O
![[JOINT COMMITTEE PRINT] 7 0 3 Union Calendar No.](https://static.fdocuments.us/doc/165x107/622c20c07741bf7bae3d265d/joint-committee-print-7-0-3-union-calendar-no.jpg)
