MUSCULOSKELETAL SIMULATION FROM MOTION CAPTURE … · M. USCULOSKELETAL SIMULATION TUTORIAL – 3DPH W. ORKSHOP – 2009.11.29. Musculoskeletal simulation ? What is it ? Musculo

TUTORIAL

MUSCULOSKELETAL SIMULATION :

FROM MOTION CAPTURE TO MUSCULAR ACTIVITYIN LOWER LIMB MODELS

Nicolas Pronost and Anders Sandholm

MUSCULOSKELETAL SIMULATION TUTORIAL – 3DPH WORKSHOP – 2009.11.29

Musculoskeletal simulation ?

What is it ?

2



What is it ? Musculo

3

Hen

ry G

ray,

Ana

tom

y of

the

hum

an b

ody,

191

8




4

Human anatomy Musculoskeletal representation

Sylv

ia S

. Ble

mke

r, St

anfo

rd U

nive

rsity

, 200

6




5

Human anatomy Musculoskeletalrepresentation

Action lines

Ope

nSim

, Uni

vers

ity o

f Sta

nfor

d



What is it ? Musculo Skeletal

6




7




8

l_mid_foot

r_hand

pelvis

l_hip

l_knee

l_ankle

l_subtalar

l_toe

r_hip

r_knee

r_ankle

r_subtalar

r_mid_foot

r_toe

vl2

vl3

vt4

vt5

vt6

head_top

r_clav

l_hand

Segments connected by

joints and hierarchically

organized

Rigid bodies with mass,

inertia matrix and CoM



What is it ? Musculo Skeletal Simulation

9



What is it ? Musculo Skeletal Simulation means analysis

10



What for ? Analyze athletic performance

11

3DAH

Mar

ie C

urie

Pro

ject

Ope

nSim

, Uni

vers

ity o

f Sta

nfor

d



What for ? Analyze athletic performance Design ergonomically safe environments

12

AnyB

ody

Tech

nolo

gy, A

albo

rg U

nive

rsity



What for ? Analyze athletic performance Design ergonomically safe environments Understand and/or treat movement disorders

13

3DAH

Mar

ie C

urie

Pro

ject

Ope

nSim

, Uni

vers

ity o

f Sta

nfor

d



What you do with ? Visualize complex movement patterns Test “what if” scenario Estimate data difficult to measure Identify cause-effect relationships

14


Outlines of the tutorial

Objective : To perform a musculoskeletal simulation from A to … V Acquisition of the data Definition of the model Inverse Kinematics solving Muscular activation estimation Validation of the simulation

Extra features How to create a model ? Interactions with medical imaging Towards more visualizations Simulating tendon transfer surgery

15


Context

Tools OpenSim

Open-source musculoskeletal simulation platform Based on SimTK (biological dynamics) Performs SCALE, IK, ID, RRA, CMC and FD Provided with validated musculoskeletal models GUI and command line based

Subject specific data Motion capture (crouch) with ground reaction forces EMG signals

16


STEP 1 : ACQUISITION OF THE DATA

17


Acquisition of the data ( 1 / 3 )

Motion capture 3D position of anatomical landmarks over time Skin markers vs. clusters vs. bone pins

18

VIC

ON

QU

ALYS

IS

QU

ALYS

IS

QU

ALYS

IS

C. N

este

r, U

nive

rsity

of S

alfo

rd, 2

007



Motion capture 3D position of anatomical landmarks over time Skin markers vs. clusters vs. bone pins

19

3DAH

Mar

ie C

urie

Pro

ject



Ground reaction forces 6D (force + moment) kinetics reaction of the body To solve the inverse dynamics analysis (through the

Newton’s laws of motion)

20

AMTI



Electromyography (EMG) signals As muscles contract, volt level electrical signals are

created within the muscle that may be measured from the surface of the body

21

NO

RAX

ON

3DAH

Mar

ie C

urie

Pro

ject

MUSCULOSKELETAL SIMULATION TUTORIAL – 3DPH WORKSHOP – 2009.11.29 22


STEP 2 : DEFINING A MODEL

23


Loading a model

Start OpenSim Menu FILE >> Open Model… Select /TutorialData/GenericModel.osim Manipulate Menu bar, 3D view, Coordinates and

Navigator panels

24


Scaling the model – Step 1

Scale factors are applied from ratios between markers distances in model and in mocap

25

Original model Standing pose Scaled model


Scaling the model – Step 2

The virtual markers are moved to match the positions of the experimental markers

26

Standing poseScaled model Subject model


Scaling the model

Menu Tools >> Scale Model… Settings >> Load Settings… Select /TutorialData/Setup_SCALE.xml Run then Close

27


STEP 3 : INVERSE KINEMATICS

28


Inverse Kinematics

Goal : to find the joint angles of the model that best reproduce the experimental kinematics of the subject’s motion Weighted least squares optimization solver with the goal

of minimizing marker errors

q = joint angles , xiexp = experimental position of marker ixi(q) = virtual position of marker i

29


Inverse Kinematics

Menu Tools >> Inverse Kinematics… Settings >> Load Settings… Select /TutorialData/Setup_IK.xml Run then Close

30


STEP 4 : MUSCULAR ACTIVATION ESTIMATION

31


Muscular activation estimation

Residual Reduction Algorithm (RRA) Dynamics inconsistency due to errors in kinematics and

kinetics measurements and in rigid body modeling Additional “residual” forces and moments are added

F + Fresidual = m . a Modification of the kinematics and the CoM to reduce

Fresidual without significantly altering the simulation

32

Ope

nSim

, Uni

vers

ity o

f Sta

nfor

d


Effect of reducing residuals

33

Ope

nSim

, Uni

vers

ity o

f Sta

nfor

d



Computed Muscle Control (CMC) To compute a set of muscle excitations tracking the

desired kinematics PD control law defines the desired accelerations Static optimization distributes the loads across actuators Forward dynamics conducts the simulation advancing in time Repeated until time is advanced to dt

34

Ope

nSim

, Uni

vers

ity o

f Sta

nfor

d



Menu Tools >> Computed Muscle Control… Settings >> Load Settings… Select /TutorialData/Setup_RRA.xml Run then Close

Menu Tools >> Computed Muscle Control… Settings >> Load Settings… Select /TutorialData/Setup_CMC.xml Run then Close

35


STEP 5 : VALIDATION OF THE SIMULATION

36


Validation of the simulation

Comparison against experimental data : EMG

37

right vastus medialiscrouch motion

right soleuscrouch motion

muscle activationfrom simulation raw EMG


Post processing of EMG

Electrical potential generated by muscle cells Measured in volt, about 90mV Signal need to be post–processed Noise Cross reading from other muscles Rectified

Filtering Box filtering

Can cancel out “real” signal Kalman filter/smoother

More computational intense

38


Post processing of EMG

39


Simulation vs. EMG

40


HOW TO CREATE A MODEL ?

41


How to create a model ?

We need Palpable bony landmarks

3D position (from mocap), definition of a coordinate system

Body parts Moment of inertia, mass, position of center of mass

The joints DoF, axis and center of rotation

Muscle and ligament attachment sites Origin and insertion (and via points) positions, fiber and tendon

lengths, mass, pennation angle…

Bony constraints Warping points and bony contours

42


Example

Klein Horsman dataset University of Twente, The Netherland [Klein Horsman, Koopman, Van der Helm, Poliacu

Prosé, Veeger. Morphological muscle and joint parameters for musculoskeletal modelling of the lower limb, Clinical Biomechanics (22), pp 239-247, 2007]

Measurements performed on a right lower extremity of a male cadaver (age 77, height 1.74m, weight 105kg)

43


Datasets

44


Datasets

21 markers 4 body parts pelvis, femur, tibia, foot

58 muscles from 163 action lines 5 joints hip, knee, femur-patella, ankle

subtalar 2 wrapping constraints Gastrocnemius around femur condyle Iliopsoas around the pelvis

104 via points

45


Validation of the model

For musculoskeletal simulation use Technical part of formatting the data Compare simulation results with

same motion and previous models experimental data (EMG)

46


INTERACTION WITH MEDICAL IMAGING

47


Interaction with medical imaging

Benefit from the intensive use of medical images to create and validate models

48

DT-MRI + fiber tracking

High resolution of joints

Cross sectional long-leg

Dynamic MRI



Benefit from the intensive use of medical images to create and validate models and simulations

49

DT-MRI + fiber tracking

High resolution of joints

Cross sectional long-leg

Dynamic MRI

Fiber directions in model

Attachment points andFE simulations

Attachment points and scaling validation

validation of kinematics



MRI viewer in OpenSim Alignment using common markers Comparisons between tendon areas and action lines

extremities

50


TOWARDSSCIENTIFIC VISUALIZATIONS

51


Towards more visualizations

To help estimating results and tuning settings Scale

Variation in factors, displacements in second inner step

IK Error over time or time-independent

CMC Magnitude of activation, reserve or residual forces

Validation Difference between activation and EMG patterns

To integrate external results Nodal displacements or pressure from FE simulations

52


Towards more visualizations

53


Thank you for your attention

References

3DAH Marie Curie Project http://3dah.miralab.unige.ch EPFL – VRLAB http://vrlab.epfl.ch Aalborg University – SMI http://www.smi.hst.aau.dk OpenSim https://simtk.org/home/opensim

[email protected] [email protected]

54

mailto:[email protected]�mailto:[email protected]�mailto:[email protected]�mailto:[email protected]�

Diapositive numéro 1Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Musculoskeletal simulation ?Outlines of the tutorialContextDiapositive numéro 17Acquisition of the data ( 1 / 3 )Acquisition of the data ( 1 / 3 )Acquisition of the data ( 2 / 3 )Acquisition of the data ( 3 / 3 )Diapositive numéro 22Diapositive numéro 23Loading a modelScaling the model – Step 1Scaling the model – Step 2Scaling the modelDiapositive numéro 28Inverse KinematicsInverse KinematicsDiapositive numéro 31Muscular activation estimationEffect of reducing residualsMuscular activation estimationMuscular activation estimationDiapositive numéro 36Validation of the simulationPost processing of EMGPost processing of EMGSimulation vs. EMGDiapositive numéro 41How to create a model ?ExampleDatasetsDatasetsValidation of the modelDiapositive numéro 47Interaction with medical imagingInteraction with medical imagingInteraction with medical imagingDiapositive numéro 51Towards more visualizationsTowards more visualizationsThank you for your attention

Documents

MUSCULOSKELETAL SIMULATION FROM MOTION CAPTURE … · M. USCULOSKELETAL SIMULATION TUTORIAL – 3DPH W. ORKSHOP – 2009.11.29. Musculoskeletal simulation ? What is it ? Musculo