Scaling pelvis width: ASIS markers or hip joint centers?

[John Davis]

Hi all,

I've been scaling lower body models to a lot of subjects recently. For the pelvis, I've been following recommendations from the scaling webinar, which suggests scaling the pelvis width using the distance between the left and right hip joint centers (HJCs) determined from a functional movement trial.

Sometimes, though, scaling this way results in models whose ASIS markers are several centimeters off the experimental locations if I set the ASIS markers as fixed: usually the ASIS markers on the scaled model are wider apart than the experimental markers. But I have the opposite problem if I scale pelvis width based on ASIS distance: the HJCs aren't in the right place. Is the ASIS/HJC mismatch just a function of the ilium "flaring out" differently for different subjects, which isn't capture by linear scaling of the pelvis?

I looked at some of the public data on SimTK and found a range of approaches for scaling pelvis width:

1) In the walking example in Rajagopal et al., the pelvis width is scaled using the average of the left and right ASIS distance, and the left and right HJC distance
2) In the running example in Rajagopal et al., the pelvis width is scaled using the average of the left and right ASIS distance, and the left and right PSIS distance
3) The scaling tutorials using the Gait 2354 and Gait 2392 models both scale the entire pelvis uniformly (XYZ axes) proportional to the left and right ASIS distance
4) Hamner and Delp use the left and right HJC distance, and let the ASIS markers move during marker registration (they are not "fixed" in the model)

I've been doing #4 for now, since I also use the ASIS markers as tracking markers during movement trials. It feels a little weird to let them move, but I don't need them for scaling the height or depth of the pelvis after I've created virtual mid-ASIS and mid-pelvis markers (which do stay fixed).

I'm curious what the OpenSim community uses to scale pelvis width: inter-ASIS distance, inter-HJC distance, or something else?