When we transferred the muscle forces from AnyBody to FEA, AnyBody exports the locations and magnitudes of all the forces applied to a bone segment. I see AnyBody defined a coupling constraint between each force application “node” and the who bone surface. My questions:
how do we know that the force application locations are correct? is it an anatomical related definition (i.e., is the node NS_Leg_Mus_BicepsFemorisCaputLongum1_1 the origin of the bicep femoris muscle?)
How does AnyBody define the size of the “patch” where the muscle force will be distributed on? I see radius ranging from 0.01 to 0.02 m - again, is that an anatomical estimate?
I see all the muscles forces are defined as the “pulling” force between the bone surface and an application node (muscle origin?), how iss wrapping force defined?
All the force application nodes are taken as is from the AnyBody model. So it crucial to morph to the subject-specific dimensions to have consistent FE and AnyBody model.
By default it tries to pull a “circle” of 1 cm radius, embedded in the converter (for Abaqus). And there is a default parameter in the APDL template for ANSYS. As the tutorial suggests - the final users may need to adjust the muscle attachment area to what it should be.
In some future the actual muscle attachment surfaces will be added (we have worked a bit on that), but it is not coming in the foreseeable future.
Wrapping muscles are reported as the contact points in exactly the same manner as everything else.
You can run an analysis with a “sliding” muscle and see how the point of application changes.
Yes, it does, indeed, come from a cadaver dissection study. The dataset for the current implementation was a PhD study by Martijn Klein-Horsman from the university of Twente that resulted in the Twente Lower Extremity Model (TLEM) - I think you can easily find it yourself or browse through a publication list on our website to find related articles.
Regarding the muscle attachment areas - yes, there will be some errors coming from such implementation. You could carry out a small sensitivity study to figure out how much effect it has on your results - similar to the mesh density tests. I would not expect a dramatic change. It is, of course, best to use the actual anatomic areas.
I am still a little confused about TLEM and AnyWomanExtPercentile.any: if TLEM was based on a male cadaver dataset, does AnyWomanExtPercentile.any morphed TLEM (male) to make the model to represent a female model? or does AnyWomanExtPercentile represent an actual female mode (i.e., based on a female cadaver?)
No, the percentile scaling only defines the size of the bone and mass distributions. If one wants to go into minor differences such as relative changes in male/female muscle attachment location, the scaled model needs to be adjusted according to specific needs.
I personally believe that the bone shape/size + mass distribution will cover the most of the difference. For example, if you incorporate the pelvic bone of a female - the moment arms will change and together with the mass differences will result in a very much female-like joint reaction forces in case of, say, elbow flexion. It might be different if we model special motion that men and women do very different and employ most varying muscles - but that’s up to the modeler to take into account.