Non-conforming shoulder

Hi Søren,

What do you mean by “some contact taking place” ? Do you refer to my model with the permanent acromial contact ? If yes, sorry if I forgot to mention you that I do not use this model right now.

What I’m trying to do here is to compute humeral translations with the intact shoulder model (from the repository) and some translational stiffness coefficients at the GH-joint and see if there is any difference with the “no-translation” model.

But from what I read in the literature, the two constraints that keep the reaction force inside the glenoid fossa and prevent any humeral translation (GH-Reactions) appear to underestimate the rotator cuff activity. This would be related to the fact that the reaction force can points anywhere inside the glenoid cavity, which does not guarantee the joint stability and because the trajectory of the joint reaction force does not show its actual application point on the glenoid cavity (which is normal since no humeral translation is simulated).

At first, my idea was to narrow the constraining area of the glenoid, in order to force the joint reaction force to points more towards the midst of the glenoid. By doing so, I would expect to obtain only negligible translational forces. However, it does not work, since the humeral head is still translating and the rotator cuff activation does not increase. Could it be due to the fact that the Force Dependant Kinematics algorithm does not account for velocity and acceleration ? Or maybe its because we have to avoid the use of the “quadratic criterion” that only favours large moment arm muscles ?!?

Thank you

Pierre

What do you think about all that ?

Hi Pierre,

I think i need to ask some question to know more about your model.

Is the following correct:

[ol]
[li]The reaction is created in the traditional way with a set of force vectors from gh center to glenoid edge
[/li][li]You have added stiffness on the joint translation which is determined by force dependency
[/li][/ol]

If this is correct i think what is missing is a proper contact definition between the humerus head and the glenoid, you can use one the new measures AnyKinPointSurface for this and create a sphere to point cloud contact. So you will need to fit a sphere to the humerus head and distribute a number of point on the cavity.

Once this is done you need add springs acting on each of the measures output, if you have 10 points it will give 10 numbers… So you then need to add a spring to these measures which is zero when there is no contact and grows when there is penetration.

Best regards
Søren

One more thing please watch Michael Skipper’s webcast tonight it is very relevant for this type of work :wink:

Hi Søren,

You are correct.

I tried adding these elements but when I plot the AnyKinPointSurface measure, I get a weird measure (see the image). Is that normal ? I would think that these green lines would converge towards the center of the surface sphere !?! What do they represent exactly ?

In order to make the spring elements work in the compression direction, I turned off the “GH-Reactions.any” and added the third humeral translation in the mediolateral direction. The humerus was then compressed in the glenoid and the spring elements apply reaction forces in various directions perpendicular to the glenoid surface. I used an hyperelastic cartilage deformation theory for the spring elements stiffnesses.

I think that this contact simulation is a big step forward, because it may compute the real application point of the joint reaction force. What do you think about these changes ?

Thank you

Pierre

Hi Pierre,

Do not worry about the drawing of the kinematic measure, the green lines shows that the measure is measuring between points and the sphere, it is a “conceptual standard view”

It sounds correct what you did, the gh reactions should be removed and replaced by spring contact between humerus head and glenoid cavtity and all three linear constraints in the gh joint should be force dependent.

How does the results look like?

Best regards
Søren

Hi Søren,

I got it for the AnyKinPointSurface drawing. I just wanted to be sure :slight_smile:

Simulation time is much higher (36 mins) than without the translation, but I think that this is normal, right ?

The results seem plausible. Please have a look at the “GHLin” measure. The displacement of the humeral head in the tangential direction is maybe too high (< 4 mm), but still look good. The compression (blue curve) and the superior translation of the humeral head (green curve) look good and mostly occur during the early phase of elevation, which seems logical.

For now, I only have spring elements on the glenoid, but maybe I should add some on the humerus. The equation that I used was giving the stress (Force / Area) in term of stretch (final length / initial length). To obtain the force, I thus divided the stress by the approximative area of the glenoid and by the number of elements.

Moreover, I used a linear stiffness for the translation of the humeral head in the tangential directions. Maybe I should seek for a non-linear stiffness.

What do you think ? If you have better ideas, please feed me :smiley:

Best regards.

Pierre

Hi Pierre,

Sounds great but I did not see the attachment :slight_smile:

Best regards
Søren

Hi Soren,

Do you mean that the attachment does not appear ? I re-verified and it is there.

Just tell me if you can see it.

Thank you

Pierre

Hi Pierre,

I see the attachment now, strange i did not see it on friday :wink:

It is difficult to judge, the curves could be alright, i am not fully aware of the geometry you are using.

I think nonlinear spring may be better.

For the knee models Michael showed on the previous webcast he used the AnyCamera class for visualizing the motion in the knee. It helps a lot to see it move in a close up, and with the camera mounted on one of the two bones.

The easy way to do this is to include this file in your study:
http://forge.anyscript.org/gf/snippet/?action=SnippetBrowse&id=18

Best regards
Søren