Dear Vasanth

plese take a look to an early approach based on kinematic models and kalman filters I studied more than 10 years ago.

Cerveri P., Pedotti A., Ferrigno G., Robust recovery of human motion from video using Kalman filters and virtual humans, Human Movement Science, 22(3):377-404, 2003.

If you are interested, I could provide you some matlab-based source code.

Pietro

Vasanth,

The marker placement strategy can certainly dictate how you determine the underlying skeletal kinematics.  There are, for example, the point cluster method by Andriacchi et al and minimum marker (less than 3 per body segment) method we developed which would use mathematical optimization to enforce the rigidity of underlying linkage.  Hope this helps.

Vasanth, Interesting topic. Is your intention to create a model that auto renders all joint centres from analysis of relative segment motion (i.e. segments marked by clusters). If so, some of my team will be interested in following your progress... 

Just keep the clusters on the segment surface itself and not on a shell.  Shells tend to slide a little during a string of gait trials, especially if high energy athletic maneuvers are involved (impacts + sweat).  I wish there was an alarm that would go off when this happened.  The OCST method form the Berlin group would be a valuable addition, and this provides a numerical constraint to the markers on the segment rather than a physical constraint.  Surely this would aid in accurate segment tracking in 3D space. Heller MO, Kratzenstein S, Ehrig RM, Wassilew G, Duda GN, Taylor WR. The weighted optimal common shape technique improves identification of the hip joint center of rotation in vivo. J Orthop Res 2011;29,1470-5.

One problem is that the very people for whom we need the most accurate data (those with gait pathologies) do not have the motor control or joint range of motion to adequate propagate the joint center optimization routines.  They must be manipulated passively by an investigator through the needed range of motion at all joints.  These functional joint center routines have now been implemented in Vicon Nexus with a dashboard view of the range of motion of the hip, knee and ankle as the individual progresses through a dynamic range of motion trial.  I have not used it yet, but it looks quite practical if they have implemented it correctly.  It should also be noted that doing this for prosthetic feet is extremely difficult since they are generally stiff, hard to "rotate", and may not even have a "joint center" at all.

The only way to truly validate a model is to have a gold standard for the joint center and segment displacement.  This means bone pins, and therefore cadavers.  If you had cadavers with bone pin clusters for each segment and different marker sets on the surface you could (finally) validate a model.

Enforcing the rigidity of the underlying linkage makes perfect sense unless the joint cannot be accurately modeled as a mechanical hinge of some type.  Some pathological joints sublux during gait 2-5 cm (acetabular dysplasia). During limb lengthening procedures, the knee can sublux 3 cm posteriorly.  Some blast injured ankles are stiffer than prosthetic feet.  I wish I got a warning when the individual violated my assumptions of a hinge, saddle, or ball & socket joint.

Most people who develop models find it difficult to grasp the range of pathology that gait analysis must cope with in the clinical setting.  I wish you the best of luck and will follow your work.

Kind regards,

Michael

Thanking you all for your inputs. Jon Gamble,  Absolutely yes. That's the whole idea. I would be more than happy to work with your team on this.

in case, you keep receiving notification of RG, i just want to mention the CGM2.i models, i have been developping for Vicon, it s now included in Nexus 2.9. check out version CGM2.4.