View Full Version : Functional methods for the ankle complex

Steve Piazza
07-12-2007, 06:52 AM
Ton raises a good point, that without active muscle forces to compress the joints, the joints of the ankle complex might behave less like revolutes. In our cadaver tests we applied motions with an axial load of over 200 N across the ankle in an effort to reduce joint "looseness". Still, there might be something different in the character of these two loading methods (foot in air with motion driven by muscles vs. foot on ground with applied tibia motion) that results in more hinge-like behavior in one case. Another possibility suggested by my colleague Bob Sainburg is that with the foot in the air, motions may be repeatable because the muscle activations/loadings are repeatable and the parameters that define the joints may still not be well defined.

In Ton's in vivo study, he and his coworkers found group average axis orientations that were within 5 to 10 degrees of the group average orientations reported in a previous cadaver study. We still think that it's important to show that these axes are accurate on a subject-by-subject (or specimen-by-specimen basis), especially given the wide range of joint axis orientations that we know exists across subjects in these joints. Repeatable axes that don't represent the actual subtalar and talocrural axes may still be valuable for modeling the ankle complex, but we should know whether or not these joints represent the actual bony articulations.

Returning to the possible ill-posedness of this problem, we made a short video clip to illustrate this point:

The video shows the simulated motions of two sets of marker clusters. One motion was generated using an idealized universal joint with revolute axes oriented in the AP and ML directions; the second was generated using the same joint rotated in the transverse plane by 30 degrees (the "talocrural" and "subtalar" joint angles for each case are different). It's evident that the markers in one set deviate from their counterparts in the other set only by small amounts and only at the ends of the range of motion. With stereophotogrammetric errors, skin movement errors, and (we think most important) deviation from revolute behavior, it is not surprising that an optimization algorithm would have difficulty distinguishing between these two joints.

Steve Piazza
Greg Lewis
Penn State University