Question about knee joint

In one of your recent newsletters, I noticed that there have been
updates to your lower limb models with the use of research out of The
Netherlands on “Morphological muscle and joint parameters for
musculoskeletal modelling of the lower extremity.” I have a question
regarding the knee in the lower limb models - are there two contact
points on the tibial plateau (one for the medial condyle of the femur,
and one for the lateral condyle of the femur)? For example, can I
determine the joint forces felt in each compartment? Or is it just the
joint force felt in the “knee,” not divided physiologically?

Hello Sarah,

the knee is still modeled as a hinge joint, therefore you get only one
force.

Best regards,
Sebastian

— In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@…>
wrote:
>
> In one of your recent newsletters, I noticed that there have been
> updates to your lower limb models with the use of research out of The
> Netherlands on “Morphological muscle and joint parameters for
> musculoskeletal modelling of the lower extremity.” I have a question
> regarding the knee in the lower limb models - are there two contact
> points on the tibial plateau (one for the medial condyle of the femur,
> and one for the lateral condyle of the femur)? For example, can I
> determine the joint forces felt in each compartment? Or is it just the
> joint force felt in the “knee,” not divided physiologically?
>

Dear Sarah,

I am currently using anybody to model knee replacement patients during
gait/sit-stand ect. I too am looking to split the force between
compartments and increase DOF in the knee to better simulate
translations across the joint space. The new TLEM is certainly a
forward step in terms of soft tissue scaling anthropometrics but as
AnyBody have said there is no change to the current hinge model. I did
notice there was a patellofemoral joint in the knee, but i’m not sure
how it is modelled and whether or not it translates during flex/ext.

It would be good if we could create a group looking into developing a
knee and work with AnyBody to enhance the current model.

best wishes

Peter Worsley

— In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@…>
wrote:
>
> In one of your recent newsletters, I noticed that there have been
> updates to your lower limb models with the use of research out of The
> Netherlands on “Morphological muscle and joint parameters for
> musculoskeletal modelling of the lower extremity.” I have a question
> regarding the knee in the lower limb models - are there two contact
> points on the tibial plateau (one for the medial condyle of the femur,
> and one for the lateral condyle of the femur)? For example, can I
> determine the joint forces felt in each compartment? Or is it just the
> joint force felt in the “knee,” not divided physiologically?
>

Hello Peter and Sarah,

In fact the new leg also contains a patella segment. It has a revolute joint
between the femur and the patella, according to the given data from Twente,
and a stiff connection to the tibia. Obviously, the knee is an important
joint which should be modelled in much more detail.

Please notice that there is a webcast on the 20th of August concerning the
leg model. Maybe we can force the discussion afterwards.

Best regards,

Sebastian


From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of peter.worsley
Sent: 29. juli 2008 10:53
To: anyscript@yahoogroups.com
Subject: [AnyScript] Re: Question about knee joint

Dear Sarah,

I am currently using anybody to model knee replacement patients during
gait/sit-stand ect. I too am looking to split the force between
compartments and increase DOF in the knee to better simulate
translations across the joint space. The new TLEM is certainly a
forward step in terms of soft tissue scaling anthropometrics but as
AnyBody have said there is no change to the current hinge model. I did
notice there was a patellofemoral joint in the knee, but i’m not sure
how it is modelled and whether or not it translates during flex/ext.

It would be good if we could create a group looking into developing a
knee and work with AnyBody to enhance the current model.

best wishes

Peter Worsley

— In anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com,
“Sarah R. Sullivan” <sarsulli@…>
wrote:
>
> In one of your recent newsletters, I noticed that there have been
> updates to your lower limb models with the use of research out of The
> Netherlands on “Morphological muscle and joint parameters for
> musculoskeletal modelling of the lower extremity.” I have a question
> regarding the knee in the lower limb models - are there two contact
> points on the tibial plateau (one for the medial condyle of the femur,
> and one for the lateral condyle of the femur)? For example, can I
> determine the joint forces felt in each compartment? Or is it just the
> joint force felt in the “knee,” not divided physiologically?
>

[Non-text portions of this message have been removed]

Hi Peter and Sarah,

I am actually considering an application to the Danish Research
Council for funds to improve the knee model. There are features in
AnyBody now that enable us to define contact conditions separately
for the two condyles and I think these can be exploited to our
advantage. In a pilot study, my master student Maje Rose did some
nice work using these new facilities to assess the forces on the ACL
during a lunge movement.

Any input from you concerning the issues that you would like the new
model to address woud be hepful for my application.

Best regards,
John

— In anyscript@yahoogroups.com, “AnyBodySupport” <support@…>
wrote:
>
> Hello Peter and Sarah,
>
>
>
> In fact the new leg also contains a patella segment. It has a
revolute joint
> between the femur and the patella, according to the given data from
Twente,
> and a stiff connection to the tibia. Obviously, the knee is an
important
> joint which should be modelled in much more detail.
>
> Please notice that there is a webcast on the 20th of August
concerning the
> leg model. Maybe we can force the discussion afterwards.
>
>
>
> Best regards,
>
> Sebastian
>
>
>
> _____
>
> From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com]
On Behalf
> Of peter.worsley
> Sent: 29. juli 2008 10:53
> To: anyscript@yahoogroups.com
> Subject: [AnyScript] Re: Question about knee joint
>
>
>
> Dear Sarah,
>
> I am currently using anybody to model knee replacement patients
during
> gait/sit-stand ect. I too am looking to split the force between
> compartments and increase DOF in the knee to better simulate
> translations across the joint space. The new TLEM is certainly a
> forward step in terms of soft tissue scaling anthropometrics but as
> AnyBody have said there is no change to the current hinge model. I
did
> notice there was a patellofemoral joint in the knee, but i’m not
sure
> how it is modelled and whether or not it translates during
flex/ext.
>
> It would be good if we could create a group looking into developing
a
> knee and work with AnyBody to enhance the current model.
>
> best wishes
>
> Peter Worsley
>
> — In anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
ps.com,
> “Sarah R. Sullivan” <sarsulli@>
> wrote:
> >
> > In one of your recent newsletters, I noticed that there have been
> > updates to your lower limb models with the use of research out of
The
> > Netherlands on “Morphological muscle and joint parameters for
> > musculoskeletal modelling of the lower extremity.” I have a
question
> > regarding the knee in the lower limb models - are there two
contact
> > points on the tibial plateau (one for the medial condyle of the
femur,
> > and one for the lateral condyle of the femur)? For example, can I
> > determine the joint forces felt in each compartment? Or is it
just the
> > joint force felt in the “knee,” not divided physiologically?
> >
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

Hi Peter,

Yes, I am definitely interested in expanding on the current AnyBody knee. I
just got a postdoctoral position and right now, I am transitioning from some
shoulder modeling (which I will be continuing) to the knee. I am putting
together a project looking at in-vivo kinematics differences between
ACL-deficient and ACL-intact knees in an upright weight-bearing MRI. My mentor
here has a Matlab program that calculates the contact area from these images. I
am anticipating that I can use some EMG data to validate a model of both an
ACL-deficient and an ACL-intact knee, find the forces in the knee joint (for
both condyles), and then calculate the contact stresses. Contact stresses are
more clinically relevant because they allow a more direct comparison with known
limits for cartilage damage, which is an assumed prognosticator of
osteoarthritis.

Where are you working out of? I am currently a postdoc at the University of
Medicine and Dentistry of New Jersey (New Jersey Med School), and doing my
research at a rehabilitation hospital. I will be watching the webcast at the
end of August, and that is right about when my timeline will begin regarding
really getting into my proposed plan of research. Hopefully we can get this
knee project going!

We will be in touch,
Sarah

-----Original Message-----

> Date: Tue Jul 29 04:52:52 EDT 2008
> From: “peter.worsley” <peter.worsley@yahoo.co.uk>
> Subject: [AnyScript] Re: Question about knee joint
> To: anyscript@yahoogroups.com
>
> Dear Sarah,
>
> I am currently using anybody to model knee replacement patients during
> gait/sit-stand ect. I too am looking to split the force between
> compartments and increase DOF in the knee to better simulate
> translations across the joint space. The new TLEM is certainly a
> forward step in terms of soft tissue scaling anthropometrics but as
> AnyBody have said there is no change to the current hinge model. I did
> notice there was a patellofemoral joint in the knee, but i’m not sure
> how it is modelled and whether or not it translates during flex/ext.
>
> It would be good if we could create a group looking into developing a
> knee and work with AnyBody to enhance the current model.
>
> best wishes
>
> Peter Worsley
>
> — In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@…>
> wrote:
> >
> > In one of your recent newsletters, I noticed that there have been
> > updates to your lower limb models with the use of research out of The
> > Netherlands on “Morphological muscle and joint parameters for
> > musculoskeletal modelling of the lower extremity.” I have a question
> > regarding the knee in the lower limb models - are there two contact
> > points on the tibial plateau (one for the medial condyle of the femur,
> > and one for the lateral condyle of the femur)? For example, can I
> > determine the joint forces felt in each compartment? Or is it just the
> > joint force felt in the “knee,” not divided physiologically?
> >
>
>

Hi Sarah and Peter (and other interested knee folks) ,

My name is David Wagner and I work at Ozen Engineering, the North American
distributor
for AnyBody. We are based in Sunnyvale, CA.

I did my PhD on simulating human movement (focusing on the lower extremities)
and am
also interested in getting involved in expanding the current AnyBody knee for
use with
implant evaluation and design. We have been using the output from AnyBody for
defining
the boundary conditions of associated FE models and have software that does this
automatically. Most recently I have been focusing more on the Application side
and would
like to get back into the Body modeling.

Lets keep this discussion going. Please feel free to contact me at
dwwagner@ozeninc.com
if you would like to talk further offline.

Best,
David Wagner
Ozen Engineering
dwwagner@ozeninc.com

— In anyscript@yahoogroups.com, “Sarah R. Dubowsky” <sarsulli@…> wrote:
>
> Hi Peter,
>
> Yes, I am definitely interested in expanding on the current AnyBody knee. I
just got a
postdoctoral position and right now, I am transitioning from some shoulder
modeling
(which I will be continuing) to the knee. I am putting together a project
looking at in-vivo
kinematics differences between ACL-deficient and ACL-intact knees in an upright
weight-
bearing MRI. My mentor here has a Matlab program that calculates the contact
area from
these images. I am anticipating that I can use some EMG data to validate a
model of both
an ACL-deficient and an ACL-intact knee, find the forces in the knee joint (for
both
condyles), and then calculate the contact stresses. Contact stresses are more
clinically
relevant because they allow a more direct comparison with known limits for
cartilage
damage, which is an assumed prognosticator of osteoarthritis.
>
> Where are you working out of? I am currently a postdoc at the University of
Medicine
and Dentistry of New Jersey (New Jersey Med School), and doing my research at a
rehabilitation hospital. I will be watching the webcast at the end of August,
and that is
right about when my timeline will begin regarding really getting into my
proposed plan of
research. Hopefully we can get this knee project going!
>
> We will be in touch,
> Sarah
>
>
>
> -----Original Message-----
>
> > Date: Tue Jul 29 04:52:52 EDT 2008
> > From: “peter.worsley” <peter.worsley@…>
> > Subject: [AnyScript] Re: Question about knee joint
> > To: anyscript@yahoogroups.com
> >
> > Dear Sarah,
> >
> > I am currently using anybody to model knee replacement patients during
> > gait/sit-stand ect. I too am looking to split the force between
> > compartments and increase DOF in the knee to better simulate
> > translations across the joint space. The new TLEM is certainly a
> > forward step in terms of soft tissue scaling anthropometrics but as
> > AnyBody have said there is no change to the current hinge model. I did
> > notice there was a patellofemoral joint in the knee, but i’m not sure
> > how it is modelled and whether or not it translates during flex/ext.
> >
> > It would be good if we could create a group looking into developing a
> > knee and work with AnyBody to enhance the current model.
> >
> > best wishes
> >
> > Peter Worsley
> >
> > — In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@>
> > wrote:
> > >
> > > In one of your recent newsletters, I noticed that there have been
> > > updates to your lower limb models with the use of research out of The
> > > Netherlands on “Morphological muscle and joint parameters for
> > > musculoskeletal modelling of the lower extremity.” I have a question
> > > regarding the knee in the lower limb models - are there two contact
> > > points on the tibial plateau (one for the medial condyle of the femur,
> > > and one for the lateral condyle of the femur)? For example, can I
> > > determine the joint forces felt in each compartment? Or is it just the
> > > joint force felt in the “knee,” not divided physiologically?
> > >
> >
> >
>

Hi again,

yahoogroups sure doesn’t want us to share our email addresses. My contact info
is:

David Wagner
408-732-4665
‘dwwagner’ ‘at’ ‘ozeninc.com

— In anyscript@yahoogroups.com, “dww805” <dwwagner@…> wrote:
>
> Hi Sarah and Peter (and other interested knee folks) ,
>
> My name is David Wagner and I work at Ozen Engineering, the North American
distributor
> for AnyBody. We are based in Sunnyvale, CA.
>
> I did my PhD on simulating human movement (focusing on the lower extremities)
and
am
> also interested in getting involved in expanding the current AnyBody knee for
use with
> implant evaluation and design. We have been using the output from AnyBody for
defining
> the boundary conditions of associated FE models and have software that does
this
> automatically. Most recently I have been focusing more on the Application
side and
would
> like to get back into the Body modeling.
>
> Lets keep this discussion going. Please feel free to contact me at
dwwagner@…
> if you would like to talk further offline.
>
> Best,
> David Wagner
> Ozen Engineering
> dwwagner@…
>
>
> — In anyscript@yahoogroups.com, “Sarah R. Dubowsky” <sarsulli@> wrote:
> >
> > Hi Peter,
> >
> > Yes, I am definitely interested in expanding on the current AnyBody knee. I
just got a
> postdoctoral position and right now, I am transitioning from some shoulder
modeling
> (which I will be continuing) to the knee. I am putting together a project
looking at in-
vivo
> kinematics differences between ACL-deficient and ACL-intact knees in an
upright
weight-
> bearing MRI. My mentor here has a Matlab program that calculates the contact
area
from
> these images. I am anticipating that I can use some EMG data to validate a
model of
both
> an ACL-deficient and an ACL-intact knee, find the forces in the knee joint
(for both
> condyles), and then calculate the contact stresses. Contact stresses are more
clinically
> relevant because they allow a more direct comparison with known limits for
cartilage
> damage, which is an assumed prognosticator of osteoarthritis.
> >
> > Where are you working out of? I am currently a postdoc at the University of
Medicine
> and Dentistry of New Jersey (New Jersey Med School), and doing my research at
a
> rehabilitation hospital. I will be watching the webcast at the end of August,
and that is
> right about when my timeline will begin regarding really getting into my
proposed plan
of
> research. Hopefully we can get this knee project going!
> >
> > We will be in touch,
> > Sarah
> >
> >
> >
> > -----Original Message-----
> >
> > > Date: Tue Jul 29 04:52:52 EDT 2008
> > > From: “peter.worsley” <peter.worsley@>
> > > Subject: [AnyScript] Re: Question about knee joint
> > > To: anyscript@yahoogroups.com
> > >
> > > Dear Sarah,
> > >
> > > I am currently using anybody to model knee replacement patients during
> > > gait/sit-stand ect. I too am looking to split the force between
> > > compartments and increase DOF in the knee to better simulate
> > > translations across the joint space. The new TLEM is certainly a
> > > forward step in terms of soft tissue scaling anthropometrics but as
> > > AnyBody have said there is no change to the current hinge model. I did
> > > notice there was a patellofemoral joint in the knee, but i’m not sure
> > > how it is modelled and whether or not it translates during flex/ext.
> > >
> > > It would be good if we could create a group looking into developing a
> > > knee and work with AnyBody to enhance the current model.
> > >
> > > best wishes
> > >
> > > Peter Worsley
> > >
> > > — In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@>
> > > wrote:
> > > >
> > > > In one of your recent newsletters, I noticed that there have been
> > > > updates to your lower limb models with the use of research out of The
> > > > Netherlands on “Morphological muscle and joint parameters for
> > > > musculoskeletal modelling of the lower extremity.” I have a question
> > > > regarding the knee in the lower limb models - are there two contact
> > > > points on the tibial plateau (one for the medial condyle of the femur,
> > > > and one for the lateral condyle of the femur)? For example, can I
> > > > determine the joint forces felt in each compartment? Or is it just the
> > > > joint force felt in the “knee,” not divided physiologically?
> > > >
> > >
> > >
> >
>

Guys,

Let me tell a little about my ideas for the knee. We are of course
aiming to create a detailed model containing all the important
elements:

  • The tibial plateau, possibly including the menisci. The plateau is
    often considered plane in models, but I am not sure this is what we
    want to do here. Some references indicate that the cruciate ligaments
    are not loaded much during normal use of the knee. This implies that
    the necessary shear constraints may come from the menisci or the
    congruency between the shape of the tibial plateau and the condyles.

  • The two femoral condyles with their non-spherical shapes.

  • The patella and its joint with the femur.

  • The ligaments.

  • The muscles. Here the model will distinguish itself from other
    models in the literature that are based on forward dynamics
    approaches.

My basic idea is to represent the contact conditions between the
condyles and the tibial plateau by means of a norm measure
(AnyKinMeasureNormComb), which is a fairly new element in AnyScript.
It allows us to define multiple potential contact points and require
just one of them to be active at any time. It should be useful for
modeling the bone-on-bone contact in the knee because we can
represent the condyle shapes as a cloud of points.

As I mentioned earlier, I am preparing a funding application and
cooperation with all users represented here is of course very much
welcomed.

Best regards,
John

— In anyscript@yahoogroups.com, “dww805” <dwwagner@…> wrote:
>
> Hi Sarah and Peter (and other interested knee folks) ,
>
> My name is David Wagner and I work at Ozen Engineering, the North
American distributor
> for AnyBody. We are based in Sunnyvale, CA.
>
> I did my PhD on simulating human movement (focusing on the lower
extremities) and am
> also interested in getting involved in expanding the current
AnyBody knee for use with
> implant evaluation and design. We have been using the output from
AnyBody for defining
> the boundary conditions of associated FE models and have software
that does this
> automatically. Most recently I have been focusing more on the
Application side and would
> like to get back into the Body modeling.
>
> Lets keep this discussion going. Please feel free to contact me at
dwwagner@…
> if you would like to talk further offline.
>
> Best,
> David Wagner
> Ozen Engineering
> dwwagner@…
>

Hi Sarah,

I am working at the University of Southampton (England) in a
BioEngineering group. I have a background in Physical Therapy but I am
now doing my doctorate linking clinical tests and Musculoskeletal
modelling. My project aims to put 80 knee arthroplasty patients
through the gait lab with EMG to validate the model as well. I am also
using real time ultrasound imaging to hopefully improve the patient
specific anthropometrics I am feeding the models. Are you using the
MRI to scale the models?

I too would like to model the contact stresses particularly during
mid-flexion activities. The orthopaedic company I am sponsored by are
obviously keen to see wear/stress patterns on the prostheses. I
believe John Rasmussen and the team are looking to make a contact
constrained driven knee model in the near future. My concerns are the
error figures from the data collection process (skin movement
artefact, marker placement error ect), and the sensitivity of the
models to the error during conversion of data. If a new knee can be
achieved how accurately will the model simulate translations and
contact stresses?

My data collection of my control group starts in a few weeks, then I
should have some knee arthroplasty patients by December/Jan. John has
suggested a trip to Aalburg to work on the knee, but I’m unsure when
I’ll have the time to do it.

best wishes

Peter

— In anyscript@yahoogroups.com, “Sarah R. Dubowsky” <sarsulli@…>
wrote:
>
> Hi Peter,
>
> Yes, I am definitely interested in expanding on the current AnyBody
knee. I just got a postdoctoral position and right now, I am
transitioning from some shoulder modeling (which I will be continuing)
to the knee. I am putting together a project looking at in-vivo
kinematics differences between ACL-deficient and ACL-intact knees in
an upright weight-bearing MRI. My mentor here has a Matlab program
that calculates the contact area from these images. I am anticipating
that I can use some EMG data to validate a model of both an
ACL-deficient and an ACL-intact knee, find the forces in the knee
joint (for both condyles), and then calculate the contact stresses.
Contact stresses are more clinically relevant because they allow a
more direct comparison with known limits for cartilage damage, which
is an assumed prognosticator of osteoarthritis.
>
> Where are you working out of? I am currently a postdoc at the
University of Medicine and Dentistry of New Jersey (New Jersey Med
School), and doing my research at a rehabilitation hospital. I will
be watching the webcast at the end of August, and that is right about
when my timeline will begin regarding really getting into my proposed
plan of research. Hopefully we can get this knee project going!
>
> We will be in touch,
> Sarah
>
>
>
> -----Original Message-----
>
> > Date: Tue Jul 29 04:52:52 EDT 2008
> > From: “peter.worsley” <peter.worsley@…>
> > Subject: [AnyScript] Re: Question about knee joint
> > To: anyscript@yahoogroups.com
> >
> > Dear Sarah,
> >
> > I am currently using anybody to model knee replacement patients during
> > gait/sit-stand ect. I too am looking to split the force between
> > compartments and increase DOF in the knee to better simulate
> > translations across the joint space. The new TLEM is certainly a
> > forward step in terms of soft tissue scaling anthropometrics but as
> > AnyBody have said there is no change to the current hinge model. I did
> > notice there was a patellofemoral joint in the knee, but i’m not sure
> > how it is modelled and whether or not it translates during flex/ext.
> >
> > It would be good if we could create a group looking into developing a
> > knee and work with AnyBody to enhance the current model.
> >
> > best wishes
> >
> > Peter Worsley
> >
> > — In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@>
> > wrote:
> > >
> > > In one of your recent newsletters, I noticed that there have been
> > > updates to your lower limb models with the use of research out
of The
> > > Netherlands on “Morphological muscle and joint parameters for
> > > musculoskeletal modelling of the lower extremity.” I have a
question
> > > regarding the knee in the lower limb models - are there two contact
> > > points on the tibial plateau (one for the medial condyle of the
femur,
> > > and one for the lateral condyle of the femur)? For example, can I
> > > determine the joint forces felt in each compartment? Or is it
just the
> > > joint force felt in the “knee,” not divided physiologically?
> > >
> >
> >
>

Hi guys,

thanks for your response John.

The AnyKinMeasureNormComb seems like a good way to increase the DOF in the knee
whilst constraining the model accurately. Will the model be flexible enough to
start removing structures for example crutiate ligaments? Like you said
ligaments would have less bearing on gait, but for activities such as
sit-stand and stairs it may be more influential.

It is my understanding that the main constraint for the knee would indeed come
tibial plateau and soft tissue structures but accurately simulating structures
such as the patella retinaculum and the menisci will not be easy. Also are you
aware of the Freeman/Johal papers indicating the point of importance for
tibiofemoral movement was the posterior articular surfaces of both the femoral
condyles (called the flexion facet centres or FFC’c). Here they suggest the
shape of the femoral condyles influence AP translation and therefore IE
rotation at the knee joint.

Best of luck for the proposal!

Peter

— On Wed, 30/7/08, John Rasmussen <jr@ime.aau.dk> wrote:
From: John Rasmussen <jr@ime.aau.dk>
Subject: [AnyScript] Re: Question about knee joint
To: anyscript@yahoogroups.com
Date: Wednesday, 30 July, 2008, 9:51 AM

Guys,

Let me tell a little about my ideas for the knee. We are of course
aiming to create a detailed model containing all the important
elements:

  • The tibial plateau, possibly including the menisci. The plateau is
    often considered plane in models, but I am not sure this is what we
    want to do here. Some references indicate that the cruciate ligaments
    are not loaded much during normal use of the knee. This implies that
    the necessary shear constraints may come from the menisci or the
    congruency between the shape of the tibial plateau and the condyles.

  • The two femoral condyles with their non-spherical shapes.

  • The patella and its joint with the femur.

  • The ligaments.

  • The muscles. Here the model will distinguish itself from other
    models in the literature that are based on forward dynamics
    approaches.

My basic idea is to represent the contact conditions between the
condyles and the tibial plateau by means of a norm measure
(AnyKinMeasureNormComb), which is a fairly new element in AnyScript.
It allows us to define multiple potential contact points and require
just one of them to be active at any time. It should be useful for
modeling the bone-on-bone contact in the knee because we can
represent the condyle shapes as a cloud of points.

As I mentioned earlier, I am preparing a funding application and
cooperation with all users represented here is of course very much
welcomed.

Best regards,
John

— In anyscript@yahoogroups.com, “dww805” <dwwagner@…>
wrote:
>
> Hi Sarah and Peter (and other interested knee folks) ,
>
> My name is David Wagner and I work at Ozen Engineering, the North
American distributor
> for AnyBody. We are based in Sunnyvale, CA.
>
> I did my PhD on simulating human movement (focusing on the lower
extremities) and am
> also interested in getting involved in expanding the current
AnyBody knee for use with
> implant evaluation and design. We have been using the output from
AnyBody for defining
> the boundary conditions of associated FE models and have software
that does this
> automatically. Most recently I have been focusing more on the
Application side and would
> like to get back into the Body modeling.
>
> Lets keep this discussion going. Please feel free to contact me at
dwwagner@…
> if you would like to talk further offline.
>
> Best,
> David Wagner
> Ozen Engineering
> dwwagner@…
>


Users of the AnyBody Modeling System help each other create biomechanical
models in the AnyScript language.Yahoo! Groups Links

   __________________________________________________________

Not happy with your email address?.
Get the one you really want - millions of new email addresses available now at
Yahoo! http://uk.docs.yahoo.com/ymail/new.html

[Non-text portions of this message have been removed]

Hi David,

looks like we are doing similar projects. I believe some of my data
from Anybody TKA simulation may also end up being converted to Ansys
at some point. What are your thoughts on the error associated with
inverse models and the cross over to dynamic simulations of knee
translations/ contraint reactions?

peter

— In anyscript@yahoogroups.com, “dww805” <dwwagner@…> wrote:
>
> Hi Sarah and Peter (and other interested knee folks) ,
>
> My name is David Wagner and I work at Ozen Engineering, the North
American distributor
> for AnyBody. We are based in Sunnyvale, CA.
>
> I did my PhD on simulating human movement (focusing on the lower
extremities) and am
> also interested in getting involved in expanding the current AnyBody
knee for use with
> implant evaluation and design. We have been using the output from
AnyBody for defining
> the boundary conditions of associated FE models and have software
that does this
> automatically. Most recently I have been focusing more on the
Application side and would
> like to get back into the Body modeling.
>
> Lets keep this discussion going. Please feel free to contact me at
dwwagner@…
> if you would like to talk further offline.
>
> Best,
> David Wagner
> Ozen Engineering
> dwwagner@…
>
>
> — In anyscript@yahoogroups.com, “Sarah R. Dubowsky” <sarsulli@> wrote:
> >
> > Hi Peter,
> >
> > Yes, I am definitely interested in expanding on the current
AnyBody knee. I just got a
> postdoctoral position and right now, I am transitioning from some
shoulder modeling
> (which I will be continuing) to the knee. I am putting together a
project looking at in-vivo
> kinematics differences between ACL-deficient and ACL-intact knees in
an upright weight-
> bearing MRI. My mentor here has a Matlab program that calculates
the contact area from
> these images. I am anticipating that I can use some EMG data to
validate a model of both
> an ACL-deficient and an ACL-intact knee, find the forces in the knee
joint (for both
> condyles), and then calculate the contact stresses. Contact
stresses are more clinically
> relevant because they allow a more direct comparison with known
limits for cartilage
> damage, which is an assumed prognosticator of osteoarthritis.
> >
> > Where are you working out of? I am currently a postdoc at the
University of Medicine
> and Dentistry of New Jersey (New Jersey Med School), and doing my
research at a
> rehabilitation hospital. I will be watching the webcast at the end
of August, and that is
> right about when my timeline will begin regarding really getting
into my proposed plan of
> research. Hopefully we can get this knee project going!
> >
> > We will be in touch,
> > Sarah
> >
> >
> >
> > -----Original Message-----
> >
> > > Date: Tue Jul 29 04:52:52 EDT 2008
> > > From: “peter.worsley” <peter.worsley@>
> > > Subject: [AnyScript] Re: Question about knee joint
> > > To: anyscript@yahoogroups.com
> > >
> > > Dear Sarah,
> > >
> > > I am currently using anybody to model knee replacement patients
during
> > > gait/sit-stand ect. I too am looking to split the force between
> > > compartments and increase DOF in the knee to better simulate
> > > translations across the joint space. The new TLEM is certainly a
> > > forward step in terms of soft tissue scaling anthropometrics but as
> > > AnyBody have said there is no change to the current hinge model.
I did
> > > notice there was a patellofemoral joint in the knee, but i’m not
sure
> > > how it is modelled and whether or not it translates during
flex/ext.
> > >
> > > It would be good if we could create a group looking into
developing a
> > > knee and work with AnyBody to enhance the current model.
> > >
> > > best wishes
> > >
> > > Peter Worsley
> > >
> > > — In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@>
> > > wrote:
> > > >
> > > > In one of your recent newsletters, I noticed that there have been
> > > > updates to your lower limb models with the use of research out
of The
> > > > Netherlands on “Morphological muscle and joint parameters for
> > > > musculoskeletal modelling of the lower extremity.” I have a
question
> > > > regarding the knee in the lower limb models - are there two
contact
> > > > points on the tibial plateau (one for the medial condyle of
the femur,
> > > > and one for the lateral condyle of the femur)? For example,
can I
> > > > determine the joint forces felt in each compartment? Or is it
just the
> > > > joint force felt in the “knee,” not divided physiologically?
> > > >
> > >
> > >
> >
>

Hi Peter,

> The AnyKinMeasureNormComb seems like a good way to
> increase the DOF in the knee whilst constraining
> the model accurately. Will the model be flexible
> enough to start removing structures for example
> crutiate ligaments? Like you said ligaments would
> have less bearing on gait, but for activities such as
> sit-stand and stairs it may be more influential.

I think this can be done much like we do it with the shoulder
stability criterion. The two fossas in the tibial plateau and the
menisci in combination can deliver reaction forces only within their
own convexities. As long as there is no need for other supports,
which I suppose will be the case for gait, then the system will not
recruit other reaction forces and the model should be stable even if
you remove the ACL or PCL. However, if the external forces are
different such as with stairclimbing, the model may need the
ligaments and will show instability as a failure to solve the
equilibrium equations if you omit the ligaments. If they are included
the model will resolve the necessary ligament forces.

> It is my understanding that the main constraint for the
> knee would indeed come tibial plateau and soft tissue
> structures but accurately simulating structures
> such as the patella retinaculum and the menisci will
> not be easy.

Yes, I think I have ideas that could solve this problem. They will
neglect much of the elastic deformation in the cartilage, though.
This is the limitation of rigid body models.

> Also are you aware of the Freeman/Johal papers indicating
> the point of importance for tibiofemoral movement was
> the posterior articular surfaces of both the femoral
> condyles (called the flexion facet centres or FFC’c).
> Here they suggest the shape of the femoral condyles
> influence AP translation and therefore IE
> rotation at the knee joint.

That is exactly my idea. If we model the contacting surfaces of both
condyles then their mutual shapes will drive the knee kinematics and
if the shapes are different they will cause tibial rotation as well.
If the observed tibial rotation is consistent with the simulated
rotation driven by the condyle shapes then we have a good indication
that this is the right ay to drive the model.

I’d be most grateful for your further thoughts.

Thanks,
John

Hi Peter,

I agree. It looks like we are trying to tackle (or at least thinking) about the
same types of
problems. I have actually been focusing my time on coming up with a robust
solution to
your question (but not specifically for the knee).

Because the input to AnyBody is kinematics and the output are the forces
necessary to
realize those kinematics, it is fairly easy to then define a transient dynamic
(i.e. forward
dynamics approach) model using those exact same set of forces (plus some initial
conditions) that essentially re-solves for the kinematics. I have replicated
the AnyBody
models in Ansys (using equivalent rigid bodies) using different procedures and
solvers and
the numerical error that propogates through these model causes the kinematics to
drift
from their original (Anybody defined) values. Once this happens, the model
essentially
blows up (because the muscle forces are not driving the same motion as they were
defined
for). There are ways to constrain the kinematics as well => but then we are
defining a
complete system and not really solving for anything. However, with all that
being said, I
am not sure if all the kinematics and/or forces that come from AnyBody would be
used as
input to a forward dynamics simulation of the knee. What are your thoughts on
this?.

In particular, I was just looking at something John just wrote (in reply to
something you
wrote):

(From Peter)
> It is my understanding that the main constraint for the
> knee would indeed come tibial plateau and soft tissue
> structures but accurately simulating structures
> such as the patella retinaculum and the menisci will
> not be easy.

(From John)
Yes, I think I have ideas that could solve this problem. They will
neglect much of the elastic deformation in the cartilage, though.
This is the limitation of rigid body models.

(Back to David)
It occurs to me that depending on if John’s solution simulates the kinematics of
the
deformable bodies (or instead if his solution directly affects the reaction
forces provided
by those simulated structures, it may be possible to ‘validate’ whichever of
those
conditions by removing those forces when they are brought into a FE package and
replace
them with flexible bodies and associated contact conditions.

— In anyscript@yahoogroups.com, “peter.worsley” <peter.worsley@…> wrote:
>
> Hi David,
>
> looks like we are doing similar projects. I believe some of my data
> from Anybody TKA simulation may also end up being converted to Ansys
> at some point. What are your thoughts on the error associated with
> inverse models and the cross over to dynamic simulations of knee
> translations/ contraint reactions?
>
> peter
>
> — In anyscript@yahoogroups.com, “dww805” <dwwagner@> wrote:
> >
> > Hi Sarah and Peter (and other interested knee folks) ,
> >
> > My name is David Wagner and I work at Ozen Engineering, the North
> American distributor
> > for AnyBody. We are based in Sunnyvale, CA.
> >
> > I did my PhD on simulating human movement (focusing on the lower
> extremities) and am
> > also interested in getting involved in expanding the current AnyBody
> knee for use with
> > implant evaluation and design. We have been using the output from
> AnyBody for defining
> > the boundary conditions of associated FE models and have software
> that does this
> > automatically. Most recently I have been focusing more on the
> Application side and would
> > like to get back into the Body modeling.
> >
> > Lets keep this discussion going. Please feel free to contact me at
> dwwagner@
> > if you would like to talk further offline.
> >
> > Best,
> > David Wagner
> > Ozen Engineering
> > dwwagner@
> >
> >
> > — In anyscript@yahoogroups.com, “Sarah R. Dubowsky” <sarsulli@> wrote:
> > >
> > > Hi Peter,
> > >
> > > Yes, I am definitely interested in expanding on the current
> AnyBody knee. I just got a
> > postdoctoral position and right now, I am transitioning from some
> shoulder modeling
> > (which I will be continuing) to the knee. I am putting together a
> project looking at in-vivo
> > kinematics differences between ACL-deficient and ACL-intact knees in
> an upright weight-
> > bearing MRI. My mentor here has a Matlab program that calculates
> the contact area from
> > these images. I am anticipating that I can use some EMG data to
> validate a model of both
> > an ACL-deficient and an ACL-intact knee, find the forces in the knee
> joint (for both
> > condyles), and then calculate the contact stresses. Contact
> stresses are more clinically
> > relevant because they allow a more direct comparison with known
> limits for cartilage
> > damage, which is an assumed prognosticator of osteoarthritis.
> > >
> > > Where are you working out of? I am currently a postdoc at the
> University of Medicine
> > and Dentistry of New Jersey (New Jersey Med School), and doing my
> research at a
> > rehabilitation hospital. I will be watching the webcast at the end
> of August, and that is
> > right about when my timeline will begin regarding really getting
> into my proposed plan of
> > research. Hopefully we can get this knee project going!
> > >
> > > We will be in touch,
> > > Sarah
> > >
> > >
> > >
> > > -----Original Message-----
> > >
> > > > Date: Tue Jul 29 04:52:52 EDT 2008
> > > > From: “peter.worsley” <peter.worsley@>
> > > > Subject: [AnyScript] Re: Question about knee joint
> > > > To: anyscript@yahoogroups.com
> > > >
> > > > Dear Sarah,
> > > >
> > > > I am currently using anybody to model knee replacement patients
> during
> > > > gait/sit-stand ect. I too am looking to split the force between
> > > > compartments and increase DOF in the knee to better simulate
> > > > translations across the joint space. The new TLEM is certainly a
> > > > forward step in terms of soft tissue scaling anthropometrics but as
> > > > AnyBody have said there is no change to the current hinge model.
> I did
> > > > notice there was a patellofemoral joint in the knee, but i’m not
> sure
> > > > how it is modelled and whether or not it translates during
> flex/ext.
> > > >
> > > > It would be good if we could create a group looking into
> developing a
> > > > knee and work with AnyBody to enhance the current model.
> > > >
> > > > best wishes
> > > >
> > > > Peter Worsley
> > > >
> > > > — In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@>
> > > > wrote:
> > > > >
> > > > > In one of your recent newsletters, I noticed that there have been
> > > > > updates to your lower limb models with the use of research out
> of The
> > > > > Netherlands on “Morphological muscle and joint parameters for
> > > > > musculoskeletal modelling of the lower extremity.” I have a
> question
> > > > > regarding the knee in the lower limb models - are there two
> contact
> > > > > points on the tibial plateau (one for the medial condyle of
> the femur,
> > > > > and one for the lateral condyle of the femur)? For example,
> can I
> > > > > determine the joint forces felt in each compartment? Or is it
> just the
> > > > > joint force felt in the “knee,” not divided physiologically?
> > > > >
> > > >
> > > >
> > >
> >
>

— In anyscript@yahoogroups.com, “dww805” <dwwagner@…> wrote:

> (Back to David)
> It occurs to me that depending on if John’s solution
> simulates the kinematics of the deformable bodies
> (or instead if his solution directly affects the
> reaction forces provided by those simulated structures,
> it may be possible to ‘validate’ whichever of those
> conditions by removing those forces when they are
> brought into a FE package and replace them with
> flexible bodies and associated contact conditions.

Hi David,

The idea I have is that the two fossas in the tibial plateau together
with the menisci create two congruent joints with the two condyles.
This is opposed to the normal perception of the condyles “sliding” on
a planar tibial plateau. I believe the observed sliding may be due to
the condyles’ non-circular shapes.

If I am right, we can use this information to model the knee joint.
The menisci are not rigid really bodies as we shall have to presume.
They are flexible and this flexibility will be ignored in the model.

So, if we subsequently bring the model into a finite element tool and
replace the computed reaction forces of the tibial plateau and the
minisci with contact conditions and the relevant material properties
then we shall get elastic deformation of the tissues until the
deformation force balances the femur. This should correspond to the
force of the original AnyBody solution if the two models are in
agreement. If not, something is wrong with one of the models.

So I am uncertain about the value of a finite element model for
validation of the forces. I do think, however, that it would be a
good way of validating the rigid body presumption. If it turns out
that the elastic deformations are large compared to the nominal joint
movement then the rigid body model is faulty. If the elastic
deformations are small, which I think they will be, then we are home
safe.

Best regards,
John

Dear all,

I have recently been modelling some data sets from squat and stair activities.
One of the main problems with the output constraint reactions is the recruitment
solvers inability to model co-contractions. This has lead to high quadriceps
activity and no hamstring activity, giving very high PA constraint knee
reaction. If a new 6DOF knee was to be designed this is a factor that will have
a large affect on the translation across the joint space.

Any thoughts on similar problems/ efforts to include modelling co-contractions
in the recruitment solver.

best wishes

Peter

— On Fri, 1/8/08, John Rasmussen <jr@ime.aau.dk> wrote:
From: John Rasmussen <jr@ime.aau.dk>
Subject: [AnyScript] Re: Question about knee joint
To: anyscript@yahoogroups.com
Date: Friday, 1 August, 2008, 8:04 AM

— In anyscript@yahoogroups.com, “dww805” <dwwagner@…> wrote:

> (Back to David)
> It occurs to me that depending on if John’s solution
> simulates the kinematics of the deformable bodies
> (or instead if his solution directly affects the
> reaction forces provided by those simulated structures,
> it may be possible to ‘validate’ whichever of those
> conditions by removing those forces when they are
> brought into a FE package and replace them with
> flexible bodies and associated contact conditions.

Hi David,

The idea I have is that the two fossas in the tibial plateau together
with the menisci create two congruent joints with the two condyles.
This is opposed to the normal perception of the condyles “sliding” on

a planar tibial plateau. I believe the observed sliding may be due to
the condyles’ non-circular shapes.

If I am right, we can use this information to model the knee joint.
The menisci are not rigid really bodies as we shall have to presume.
They are flexible and this flexibility will be ignored in the model.

So, if we subsequently bring the model into a finite element tool and
replace the computed reaction forces of the tibial plateau and the
minisci with contact conditions and the relevant material properties
then we shall get elastic deformation of the tissues until the
deformation force balances the femur. This should correspond to the
force of the original AnyBody solution if the two models are in
agreement. If not, something is wrong with one of the models.

So I am uncertain about the value of a finite element model for
validation of the forces. I do think, however, that it would be a
good way of validating the rigid body presumption. If it turns out
that the elastic deformations are large compared to the nominal joint
movement then the rigid body model is faulty. If the elastic
deformations are small, which I think they will be, then we are home
safe.

Best regards,
John


Users of the AnyBody Modeling System help each other create biomechanical
models in the AnyScript language.Yahoo! Groups Links

[Non-text portions of this message have been removed]

Hi Peter,

I have had a recent conversation with my colleagues at the University of
Michigan who
mentioned they were having a similar problem in validating a lower limb exertion
modeled
in AnyBody with EMG recordings. They attributed the problem to AnyBody not
correctly
solving for co-contracting muscles. Their ad-hoc solution was to change the
muscle
strengths so the muscle recruitment better fit the EMG data.

John would probably be better at commenting on this, but as I recall, there is
not much
flexibility in using or altering the recruitment philosophy beyond the
linear/quadratic and
associated penalty factors. I have heard that co-contractions can be, in some
part,
related to joint stability. However, I am not sure if there is such a minimum
requirement
(or if there should be) in the AnyBody recruitment solver.

Would there be any benefit to opening the recruitment optimization code to the
users so
we can try and implement our own constraints or policy…??

-David

— In anyscript@yahoogroups.com, Peter Worsley <peter.worsley@…> wrote:
>
> Dear all,
>
> I have recently been modelling some data sets from squat and stair activities.
One of the
main problems with the output constraint reactions is the recruitment solvers
inability to
model co-contractions. This has lead to high quadriceps activity and no
hamstring
activity, giving very high PA constraint knee reaction. If a new 6DOF knee was
to be
designed this is a factor that will have a large affect on the translation
across the joint
space.
>
> Any thoughts on similar problems/ efforts to include modelling co-contractions
in the
recruitment solver.
>
> best wishes
>
> Peter
>
> — On Fri, 1/8/08, John Rasmussen <jr@…> wrote:
> From: John Rasmussen <jr@…>
> Subject: [AnyScript] Re: Question about knee joint
> To: anyscript@yahoogroups.com
> Date: Friday, 1 August, 2008, 8:04 AM
>
> — In anyscript@yahoogroups.com, “dww805” <dwwagner@> wrote:
>
> > (Back to David)
> > It occurs to me that depending on if John’s solution
> > simulates the kinematics of the deformable bodies
> > (or instead if his solution directly affects the
> > reaction forces provided by those simulated structures,
> > it may be possible to ‘validate’ whichever of those
> > conditions by removing those forces when they are
> > brought into a FE package and replace them with
> > flexible bodies and associated contact conditions.
>
> Hi David,
>
> The idea I have is that the two fossas in the tibial plateau together
> with the menisci create two congruent joints with the two condyles.
> This is opposed to the normal perception of the condyles “sliding” on
>
> a planar tibial plateau. I believe the observed sliding may be due to
> the condyles’ non-circular shapes.
>
> If I am right, we can use this information to model the knee joint.
> The menisci are not rigid really bodies as we shall have to presume.
> They are flexible and this flexibility will be ignored in the model.
>
> So, if we subsequently bring the model into a finite element tool and
> replace the computed reaction forces of the tibial plateau and the
> minisci with contact conditions and the relevant material properties
> then we shall get elastic deformation of the tissues until the
> deformation force balances the femur. This should correspond to the
> force of the original AnyBody solution if the two models are in
> agreement. If not, something is wrong with one of the models.
>
> So I am uncertain about the value of a finite element model for
> validation of the forces. I do think, however, that it would be a
> good way of validating the rigid body presumption. If it turns out
> that the elastic deformations are large compared to the nominal joint
> movement then the rigid body model is faulty. If the elastic
> deformations are small, which I think they will be, then we are home
> safe.
>
> Best regards,
> John
>
>
>
> ------------------------------------
>
> Users of the AnyBody Modeling System help each other create biomechanical
> models in the AnyScript language.Yahoo! Groups Links
>
>
>
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

Hi Guys,

This is an interesting discussion. Please let me offer some input on the
matter:

We know the following:

  1. For situations such as gait and pedaling we can get good
    correspondence between EMG and predictions by the model for the lower
    extremity, so at least in some cases this works well with the current
    muscle recruitment.
  2. It has been shown a number of times that optimization-based
    methods, such as the one AnyBody uses, precludes antagonistic muscle
    actions, i.e. co-contraction. However, in a very interesting recent
    paper (A. Jinha et al. / Journal of Biomechanics 39 (2006) 1145-1152) is
    it shown that this conclusion is somewhat related to the model
    simplifications in the simple models that the conclusions are based on
    and does not hold for more complex models. Indeed, in many simulations
    in AnyBody we do get antagonistic actions. It is true, however, that we
    do not get co-contractions that do not somehow benefit the overall goal
    of the movement. We cannot predict “excessive” co-contraction such as we
    would get when the body braces itself in an uncertain situation.

My experience is that when a model does not fit well with the
experiments the explanation can often be found in the data we are
putting into the system, typically boundary conditions or kinematics. So
if you get poor results with a model I urge you to seek the support of
the group and let us have a look at it. We might be able to find a
solution to the problem.

David proposes opening up for more user-defined recruitment criteria,
and we would indeed like to do that. It is not so easy, however, because
the recruitment is at the heart what the system does and the efficiency
of the entire system is very dependent on the efficiency of the
recruitment. Going from the current hard-coded algorithms to algorithms
that interpret a user-defined specification is hard to combine with
retaining the efficiency. It is not going to keep us from trying,
through.

Best regards,

John


John Rasmussen,
Professor, PhD, The AnyBody Group, Dept. of Mech. Eng.
Aalborg University
www.ime.aau.dk/~jr, jr@ime.aau.dk <mailto:jr@ime.aau.dk>
Mobile: +45 2089 8350. Phone: +45 9940 9307 (New number!)


From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On
Behalf Of dww805
Sent: 11. september 2008 19:07
To: anyscript@yahoogroups.com
Subject: [AnyScript] Re: Question about knee joint

Hi Peter,

I have had a recent conversation with my colleagues at the University of
Michigan who
mentioned they were having a similar problem in validating a lower limb
exertion modeled
in AnyBody with EMG recordings. They attributed the problem to AnyBody
not correctly
solving for co-contracting muscles. Their ad-hoc solution was to change
the muscle
strengths so the muscle recruitment better fit the EMG data.

John would probably be better at commenting on this, but as I recall,
there is not much
flexibility in using or altering the recruitment philosophy beyond the
linear/quadratic and
associated penalty factors. I have heard that co-contractions can be, in
some part,
related to joint stability. However, I am not sure if there is such a
minimum requirement
(or if there should be) in the AnyBody recruitment solver.

Would there be any benefit to opening the recruitment optimization code
to the users so
we can try and implement our own constraints or policy…??

-David

— In anyscript@yahoogroups.com <mailto:anyscript%40yahoogroups.com> ,
Peter Worsley <peter.worsley@…> wrote:
>
> Dear all,
>
> I have recently been modelling some data sets from squat and stair
activities. One of the
main problems with the output constraint reactions is the recruitment
solvers inability to
model co-contractions. This has lead to high quadriceps activity and no
hamstring
activity, giving very high PA constraint knee reaction. If a new 6DOF
knee was to be
designed this is a factor that will have a large affect on the
translation across the joint
space.
>
> Any thoughts on similar problems/ efforts to include modelling
co-contractions in the
recruitment solver.
>
> best wishes
>
> Peter
>
> — On Fri, 1/8/08, John Rasmussen <jr@…> wrote:
> From: John Rasmussen <jr@…>
> Subject: [AnyScript] Re: Question about knee joint
> To: anyscript@yahoogroups.com <mailto:anyscript%40yahoogroups.com>
> Date: Friday, 1 August, 2008, 8:04 AM
>
> — In anyscript@yahoogroups.com <mailto:anyscript%40yahoogroups.com>
, “dww805” <dwwagner@> wrote:
>
> > (Back to David)
> > It occurs to me that depending on if John’s solution
> > simulates the kinematics of the deformable bodies
> > (or instead if his solution directly affects the
> > reaction forces provided by those simulated structures,
> > it may be possible to ‘validate’ whichever of those
> > conditions by removing those forces when they are
> > brought into a FE package and replace them with
> > flexible bodies and associated contact conditions.
>
> Hi David,
>
> The idea I have is that the two fossas in the tibial plateau together
> with the menisci create two congruent joints with the two condyles.
> This is opposed to the normal perception of the condyles “sliding” on
>
> a planar tibial plateau. I believe the observed sliding may be due to
> the condyles’ non-circular shapes.
>
> If I am right, we can use this information to model the knee joint.
> The menisci are not rigid really bodies as we shall have to presume.
> They are flexible and this flexibility will be ignored in the model.
>
> So, if we subsequently bring the model into a finite element tool and
> replace the computed reaction forces of the tibial plateau and the
> minisci with contact conditions and the relevant material properties
> then we shall get elastic deformation of the tissues until the
> deformation force balances the femur. This should correspond to the
> force of the original AnyBody solution if the two models are in
> agreement. If not, something is wrong with one of the models.
>
> So I am uncertain about the value of a finite element model for
> validation of the forces. I do think, however, that it would be a
> good way of validating the rigid body presumption. If it turns out
> that the elastic deformations are large compared to the nominal joint
> movement then the rigid body model is faulty. If the elastic
> deformations are small, which I think they will be, then we are home
> safe.
>
> Best regards,
> John
>
>
>
> ------------------------------------
>
> Users of the AnyBody Modeling System help each other create
biomechanical
> models in the AnyScript language.Yahoo! Groups Links
>
>
>
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

[Non-text portions of this message have been removed]

Hello John, Peter and others.

This is a very interesting and important subject (ie co-contraction
and antagonistic muscles).

I’m not quite sure that I follow you. Does the Jinha paper (A. Jinha
et al. / Journal of Biomechanics 39 (2006) 1145-1152) contradict (no
pun intended) the findings in the “old” paper of Herzog?

Herzog W & Binding P, Cocontraction of pairs of antagonistic muscles:
analytical solution for planar static nonlinear optimization
approaches, Mathematical Biosciences Volume 118, Issue 1 , November
1993, Pages 83-95

/jokke

— In anyscript@yahoogroups.com, “John Rasmussen” <jr@…> wrote:
>
> Hi Guys,
>
>
>
> This is an interesting discussion. Please let me offer some input
on the
> matter:
>
>
>
> We know the following:
>
> 1. For situations such as gait and pedaling we can get good
> correspondence between EMG and predictions by the model for the
lower
> extremity, so at least in some cases this works well with the
current
> muscle recruitment.
> 2. It has been shown a number of times that optimization-based
> methods, such as the one AnyBody uses, precludes antagonistic muscle
> actions, i.e. co-contraction. However, in a very interesting recent
> paper (A. Jinha et al. / Journal of Biomechanics 39 (2006) 1145-
1152) is
> it shown that this conclusion is somewhat related to the model
> simplifications in the simple models that the conclusions are based
on
> and does not hold for more complex models. Indeed, in many
simulations
> in AnyBody we do get antagonistic actions. It is true, however,
that we
> do not get co-contractions that do not somehow benefit the overall
goal
> of the movement. We cannot predict “excessive” co-contraction such
as we
> would get when the body braces itself in an uncertain situation.
>
>
>
> My experience is that when a model does not fit well with the
> experiments the explanation can often be found in the data we are
> putting into the system, typically boundary conditions or
kinematics. So
> if you get poor results with a model I urge you to seek the support
of
> the group and let us have a look at it. We might be able to find a
> solution to the problem.
>
>
>
> David proposes opening up for more user-defined recruitment
criteria,
> and we would indeed like to do that. It is not so easy, however,
because
> the recruitment is at the heart what the system does and the
efficiency
> of the entire system is very dependent on the efficiency of the
> recruitment. Going from the current hard-coded algorithms to
algorithms
> that interpret a user-defined specification is hard to combine with
> retaining the efficiency. It is not going to keep us from trying,
> through.
>
>
>
> Best regards,
>
> John
>
>
>
> __________________________________
> John Rasmussen,
> Professor, PhD, The AnyBody Group, Dept. of Mech. Eng.
> Aalborg University
> www.ime.aau.dk/~jr, jr@… <mailto:jr@…>
> Mobile: +45 2089 8350. Phone: +45 9940 9307 (New number!)
>
>
>
> ________________________________
>
> From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com]
On
> Behalf Of dww805
> Sent: 11. september 2008 19:07
> To: anyscript@yahoogroups.com
> Subject: [AnyScript] Re: Question about knee joint
>
>
>
> Hi Peter,
>
> I have had a recent conversation with my colleagues at the
University of
> Michigan who
> mentioned they were having a similar problem in validating a lower
limb
> exertion modeled
> in AnyBody with EMG recordings. They attributed the problem to
AnyBody
> not correctly
> solving for co-contracting muscles. Their ad-hoc solution was to
change
> the muscle
> strengths so the muscle recruitment better fit the EMG data.
>
> John would probably be better at commenting on this, but as I
recall,
> there is not much
> flexibility in using or altering the recruitment philosophy beyond
the
> linear/quadratic and
> associated penalty factors. I have heard that co-contractions can
be, in
> some part,
> related to joint stability. However, I am not sure if there is such
a
> minimum requirement
> (or if there should be) in the AnyBody recruitment solver.
>
> Would there be any benefit to opening the recruitment optimization
code
> to the users so
> we can try and implement our own constraints or policy…??
>
> -David
>
> — In anyscript@yahoogroups.com <mailto:anyscript%
40yahoogroups.com> ,
> Peter Worsley <peter.worsley@> wrote:
> >
> > Dear all,
> >
> > I have recently been modelling some data sets from squat and stair
> activities. One of the
> main problems with the output constraint reactions is the
recruitment
> solvers inability to
> model co-contractions. This has lead to high quadriceps activity
and no
> hamstring
> activity, giving very high PA constraint knee reaction. If a new
6DOF
> knee was to be
> designed this is a factor that will have a large affect on the
> translation across the joint
> space.
> >
> > Any thoughts on similar problems/ efforts to include modelling
> co-contractions in the
> recruitment solver.
> >
> > best wishes
> >
> > Peter
> >
> > — On Fri, 1/8/08, John Rasmussen <jr@> wrote:
> > From: John Rasmussen <jr@>
> > Subject: [AnyScript] Re: Question about knee joint
> > To: anyscript@yahoogroups.com <mailto:anyscript%
40yahoogroups.com>
> > Date: Friday, 1 August, 2008, 8:04 AM
> >
> > — In anyscript@yahoogroups.com <mailto:anyscript%
40yahoogroups.com>
> , “dww805” <dwwagner@> wrote:
> >
> > > (Back to David)
> > > It occurs to me that depending on if John’s solution
> > > simulates the kinematics of the deformable bodies
> > > (or instead if his solution directly affects the
> > > reaction forces provided by those simulated structures,
> > > it may be possible to ‘validate’ whichever of those
> > > conditions by removing those forces when they are
> > > brought into a FE package and replace them with
> > > flexible bodies and associated contact conditions.
> >
> > Hi David,
> >
> > The idea I have is that the two fossas in the tibial plateau
together
> > with the menisci create two congruent joints with the two
condyles.
> > This is opposed to the normal perception of the
condyles “sliding” on
> >
> > a planar tibial plateau. I believe the observed sliding may be
due to
> > the condyles’ non-circular shapes.
> >
> > If I am right, we can use this information to model the knee
joint.
> > The menisci are not rigid really bodies as we shall have to
presume.
> > They are flexible and this flexibility will be ignored in the
model.
> >
> > So, if we subsequently bring the model into a finite element tool
and
> > replace the computed reaction forces of the tibial plateau and
the
> > minisci with contact conditions and the relevant material
properties
> > then we shall get elastic deformation of the tissues until the
> > deformation force balances the femur. This should correspond to
the
> > force of the original AnyBody solution if the two models are in
> > agreement. If not, something is wrong with one of the models.
> >
> > So I am uncertain about the value of a finite element model for
> > validation of the forces. I do think, however, that it would be a
> > good way of validating the rigid body presumption. If it turns
out
> > that the elastic deformations are large compared to the nominal
joint
> > movement then the rigid body model is faulty. If the elastic
> > deformations are small, which I think they will be, then we are
home
> > safe.
> >
> > Best regards,
> > John
> >
> >
> >
> > ------------------------------------
> >
> > Users of the AnyBody Modeling System help each other create
> biomechanical
> > models in the AnyScript language.Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

Dear all,

glad to spark some conversation on this topic. Thank you everyone for your
input and the papers regarding co-contractions. I agree with john that the
models heavily depend on the input quality and solver penalties, but I feel
there is a evident need for co-contractions. Certain joints (knee for example)
are heavily constrained in the anatomical sense by muscle input and
co-contractions during mid-flexion high loading activities. I’ve attached a
small summary with some literature on the subject.

 Perhaps an option for the recruitment solver would be to include a
function to minimise joint constraint reaction (similar human musculoskeletal
system) so the joints are in better equilibrium during dynamic movement.
Perhaps the computational expense would be warranted by more accurate outputs
from the model.

best wishes

Peter

— On Fri, 12/9/08, joakim_holmberg <joakim@sockertoppen.com> wrote:
From: joakim_holmberg <joakim@sockertoppen.com>
Subject: [AnyScript] Re: Question about knee joint
To: anyscript@yahoogroups.com
Date: Friday, 12 September, 2008, 6:39 AM

Hello John, Peter and others.

This is a very interesting and important subject (ie co-contraction
and antagonistic muscles).

I’m not quite sure that I follow you. Does the Jinha paper (A. Jinha
et al. / Journal of Biomechanics 39 (2006) 1145-1152) contradict (no
pun intended) the findings in the “old” paper of Herzog?

Herzog W & Binding P, Cocontraction of pairs of antagonistic muscles:
analytical solution for planar static nonlinear optimization
approaches, Mathematical Biosciences Volume 118, Issue 1 , November
1993, Pages 83-95

/jokke

— In anyscript@yahoogroups.com, “John Rasmussen” <jr@…>
wrote:
>
> Hi Guys,
>
>
>
> This is an interesting discussion. Please let me offer some input
on the
> matter:
>
>
>
> We know the following:
>
> 1. For situations such as gait and pedaling we can get good
> correspondence between EMG and predictions by the model for the
lower
> extremity, so at least in some cases this works well with the
current
> muscle recruitment.
> 2. It has been shown a number of times that optimization-based
> methods, such as the one AnyBody uses, precludes antagonistic muscle
> actions, i.e. co-contraction. However, in a very interesting recent
> paper (A. Jinha et al. / Journal of Biomechanics 39 (2006) 1145-
1152) is
> it shown that this conclusion is somewhat related to the model
> simplifications in the simple models that the conclusions are based
on
> and does not hold for more complex models. Indeed, in many
simulations
> in AnyBody we do get antagonistic actions. It is true, however,
that we
> do not get co-contractions that do not somehow benefit the overall
goal
> of the movement. We cannot predict “excessive” co-contraction
such
as we
> would get when the body braces itself in an uncertain situation.
>
>
>
> My experience is that when a model does not fit well with the
> experiments the explanation can often be found in the data we are
> putting into the system, typically boundary conditions or
kinematics. So
> if you get poor results with a model I urge you to seek the support
of
> the group and let us have a look at it. We might be able to find a
> solution to the problem.
>
>
>
> David proposes opening up for more user-defined recruitment
criteria,
> and we would indeed like to do that. It is not so easy, however,
because
> the recruitment is at the heart what the system does and the
efficiency
> of the entire system is very dependent on the efficiency of the
> recruitment. Going from the current hard-coded algorithms to
algorithms
> that interpret a user-defined specification is hard to combine with
> retaining the efficiency. It is not going to keep us from trying,
> through.
>
>
>
> Best regards,
>
> John
>
>
>
> __________________________________
> John Rasmussen,
> Professor, PhD, The AnyBody Group, Dept. of Mech. Eng.
> Aalborg University
> www.ime.aau.dk/~jr, jr@… <mailto:jr@…>
> Mobile: +45 2089 8350. Phone: +45 9940 9307 (New number!)
>
>
>
> ________________________________
>
> From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com]
On
> Behalf Of dww805
> Sent: 11. september 2008 19:07
> To: anyscript@yahoogroups.com
> Subject: [AnyScript] Re: Question about knee joint
>
>
>
> Hi Peter,
>
> I have had a recent conversation with my colleagues at the
University of
> Michigan who
> mentioned they were having a similar problem in validating a lower
limb
> exertion modeled
> in AnyBody with EMG recordings. They attributed the problem to
AnyBody
> not correctly
> solving for co-contracting muscles. Their ad-hoc solution was to
change
> the muscle
> strengths so the muscle recruitment better fit the EMG data.
>
> John would probably be better at commenting on this, but as I
recall,
> there is not much
> flexibility in using or altering the recruitment philosophy beyond
the
> linear/quadratic and
> associated penalty factors. I have heard that co-contractions can
be, in
> some part,
> related to joint stability. However, I am not sure if there is such
a
> minimum requirement
> (or if there should be) in the AnyBody recruitment solver.
>
> Would there be any benefit to opening the recruitment optimization
code
> to the users so
> we can try and implement our own constraints or policy…??
>
> -David
>
> — In anyscript@yahoogroups.com <mailto:anyscript%
40yahoogroups.com> ,
> Peter Worsley <peter.worsley@> wrote:
> >
> > Dear all,
> >
> > I have recently been modelling some data sets from squat and stair
> activities. One of the
> main problems with the output constraint reactions is the
recruitment
> solvers inability to
> model co-contractions. This has lead to high quadriceps activity
and no
> hamstring
> activity, giving very high PA constraint knee reaction. If a new
6DOF
> knee was to be
> designed this is a factor that will have a large affect on the
> translation across the joint
> space.
> >
> > Any thoughts on similar problems/ efforts to include modelling
> co-contractions in the
> recruitment solver.
> >
> > best wishes
> >
> > Peter
> >
> > — On Fri, 1/8/08, John Rasmussen <jr@> wrote:
> > From: John Rasmussen <jr@>
> > Subject: [AnyScript] Re: Question about knee joint
> > To: anyscript@yahoogroups.com <mailto:anyscript%
40yahoogroups.com>
> > Date: Friday, 1 August, 2008, 8:04 AM
> >
> > — In anyscript@yahoogroups.com <mailto:anyscript%
40yahoogroups.com>
> , “dww805” <dwwagner@> wrote:
> >
> > > (Back to David)
> > > It occurs to me that depending on if John’s solution
> > > simulates the kinematics of the deformable bodies
> > > (or instead if his solution directly affects the
> > > reaction forces provided by those simulated structures,
> > > it may be possible to ‘validate’ whichever of those
> > > conditions by removing those forces when they are
> > > brought into a FE package and replace them with
> > > flexible bodies and associated contact conditions.
> >
> > Hi David,
> >
> > The idea I have is that the two fossas in the tibial plateau
together
> > with the menisci create two congruent joints with the two
condyles.
> > This is opposed to the normal perception of the
condyles “sliding” on
> >
> > a planar tibial plateau. I believe the observed sliding may be
due to
> > the condyles’ non-circular shapes.
> >
> > If I am right, we can use this information to model the knee
joint.
> > The menisci are not rigid really bodies as we shall have to
presume.
> > They are flexible and this flexibility will be ignored in the
model.
> >
> > So, if we subsequently bring the model into a finite element tool
and
> > replace the computed reaction forces of the tibial plateau and
the
> > minisci with contact conditions and the relevant material
properties
> > then we shall get elastic deformation of the tissues until the
> > deformation force balances the femur. This should correspond to
the
> > force of the original AnyBody solution if the two models are in
> > agreement. If not, something is wrong with one of the models.
> >
> > So I am uncertain about the value of a finite element model for
> > validation of the forces. I do think, however, that it would be a
> > good way of validating the rigid body presumption. If it turns
out
> > that the elastic deformations are large compared to the nominal
joint
> > movement then the rigid body model is faulty. If the elastic
> > deformations are small, which I think they will be, then we are
home
> > safe.
> >
> > Best regards,
> > John
> >
> >
> >
> > ------------------------------------
> >
> > Users of the AnyBody Modeling System help each other create
> biomechanical
> > models in the AnyScript language.Yahoo! Groups Links
> >
> >
> >
> >
> >
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>
>
>
>
>
> [Non-text portions of this message have been removed]
>


Users of the AnyBody Modeling System help each other create biomechanical
models in the AnyScript language.Yahoo! Groups Links

[Non-text portions of this message have been removed]