I have a question regarding the availability of a better knee joint.
A few days ago I was very happy to read in the AnyBody newsletter that a better Knee model is in the works. We are working on a (very basic) knee model which is meant for the determination of influences of external forces on the femoral condyle displacement on the tibial plateau. It would be a great step forward if we could have a better (more complex) knee model provided by AnyBodyTech.
Is there any knee model that could be used already or do we have to wait for the mentioned Knee model example to be released? When will it come out?
I was searching the repositories for published knee models as presented for example in “Biomechanical Analysis of Anterior Cruciate Ligament Injury Mechanisms” by Maja Rose Schmidt, but it seems they are not available to other AnyBody users. Maybe I am missing an available model?
You are right, we are currently working on a new knee model. The new model will be more detailed and not be a revolute joint. However, at the moment, we can not say when it will be ready for release.
Are you still working on this model or it has been released ? Is it the “TKA-KneeBendDemo” model ?
I also work on the knee model and would like to allow an additional rotation about the y-axis (axial rotation). However, I’m not sure what changes to make to the CD3 data (e.g. OptX, OptY, OptZ) in order to obtain a physiological result (i.e. some axial rotation but not too much).
Also, do you think that allowing this axial rotation of the knee would result in a better concordance between the initial and optimized C3D data ?
Yes, you are right. The ‘TKA-KneeBendDemo’ model is a kind of non-conforming knee joint model.
If you want to allow an additional knee axial rotation,
then I think it will be better for you to change the definition of knee joint to allow that rotation instead of changing C3D data.
Of course then you may have to use more markers on your legs.
And also you have to use this kind of non-conforming joint which includes ligaments.
That’s what I did. I changed the joint from a revolute to an universal joint with the Z and Y axes free. However, when I performed the motion optimization again with this new joint configuration, I ended with non-physiological axial rotations. Thus, I though that some minor changes had to be done to the motion optimization process in order to take into account this new DOF. I was thinking about changing the OptX, OptY and OptZ, so that some of the markers would become optimized and some others would become not optimized.
However, as you said, we should use more markers at the knee in order to better measure this axial rotation. With the current set of markers, it seems difficult to obtain good estimations. So I will put axial rotation aside for the moment.
For the ligaments, I was asking if it would be possible to implement their definition in the “LowerExtremityGait” model, since it seems to be the same bones.
Ok great but originally, the ligaments are there to stabilize the knee and help the FDK to converge, right ? Without any translation between the femur and the tibial plateau, do you think that simulating the ligaments will still significantly contribute to the knee reaction force ?
As you all know, validation of model estimations is actually a big issue. Personalization of models based on single subject measurements along with in vivo measurements of contact forces on such subject is a promising avenue to address that issue. This is one of the goals of the “Grand Challenge competition”.
I’m still thinking about the TKA knee bend model, which is based on the dataset of that “Grand Challenge competition”. It seems that you did not use the bones from the dataset, is that right ? Is there a reason for this choice ? Also, did you consider the muscle strength measurements from the Biodex exercises ?
The ideal solution would be to create and validate an AnyBody model using all the available data from the “Grand Challenge” dataset (i.e bone and prosthesis geometry, MoCap, EMG, contact forces, muscle strength tests).
I think that with such personalized model, there would be less uncertainties related to some parameters such as muscle strength, anthropometry, kinematics and contact force estimation. Thus, this would allow to concentrate our effort in finding the best polynomial orders for the muscle recruitment criterion and activity transient to minimize the difference between estimated and measured activations and contact forces, depending on the performed task (knee bend, gait, etc.).
Is this something that you already plan to do in a near future with the TKA Knee Bend model ? Is that model still under development ? Will a new version be released soon ?
First of all, the TKA model in the repository is just a demo model to show how to use FDK and contact. It is not our Grand Challenge model although the implant stl’s come from the first Grand Challenge dataset.
You are absolutely right that it would be a very good idea to develop a model that uses as much of the available data as possible. I do have a FDK-based model based on the first Grand Challenge dataset but not for the following years. However, the first year dataset does not include full CT scans of the leg nor MRI of the knee. These were included in the later datasets.
While I was developing the model for the first year, we were still working on implementing FDK, stl contact force computations and not least nonlinear morphing methods. All the tools are now available to continue the work on the next Grand Challenge datasets, where all three components (FDK, contact and morphing) are going to be needed if all the data should be used. However, developing these extensive models for all the datasets is a big task. It is certainly on my todo list, but since I am also busy with other tasks, progress is not as fast as I would like it to be.
Since you ask for the availability of this model, I was wondering if you are planning on starting some knee modeling? If yes, I suggest that we join forces and develop the model you requested together? Please feel free to contact me directly if you think this is a good idea.
Best regards
Michael Skipper Andersen
Associate Professor
The AnyBody Research Group
