Optimization algorithms to correct movements

[Juan Dura]

I am looking for literature about algorithms that correct recorded movements and the position of joints. Algorithms that takes in account restrictions. For example some joints are fixed (no movement) and the length of the segments.
I am specially interested in reviews that compare different algorithms.
Thanks in advance

[Ton van den Bogert]

Re: Optimization algorithms to correct movements

Juan,

Speaking as moderator: Please get us started by sharing what you already have found on your own. Information sharing needs to go both ways, otherwise you won't get much help.

I find this an interesting topic and will post my contribution later. I have not followed this topic recently so I hope you and others will post their own knowledge of the literature.

Ton van den Bogert

[Ross Miller]

Re: Optimization algorithms to correct movements

Hi Juan,

If you're referring to "inverse kinematics" methods (i.e. from marker coordinates, calculate joint and segment kinematics), the three main methods I'm familiar with are:

- Direct Pose (e.g. Davis et al., 1991)
- Segment Optimization (e.g. Challis et al., 1995)
- Global Optimization (e.g. Lu & O'Connor, 1999)

The 3D kinematics chapter in the Research Methods in Biomechanics text has a summary and pros/cons for each. Cappozzo's group had a series of review articles that might also cover this topic (not 100% sure of that, I haven't read them recently):

http://www.ncbi.nlm.nih.gov/pubmed/15639398
http://www.ncbi.nlm.nih.gov/pubmed/15639399
http://www.ncbi.nlm.nih.gov/pubmed/15639400
http://www.ncbi.nlm.nih.gov/pubmed/15639401

I like Global Optimization because I'm often doing forward dynamics and it lets me compare my "measured" and "simulated" kinematics using a consistent model, but there are lots of cases where Segment Optimization is a very good approach too. Direct Pose is I think not used much these days. Global Optimization will allow you to do things like define specific joint models, force segment lengths to be constant even if the data say they are not, etc. Direct Pose and Segment Optimization are both 6DoF methods. All three will have the "fundamental problem of inverse dynamics" (Hatze, 2002) if you're also doing kinetic analyses.

Hope this helps,
Ross

References
Challis JH (1995). A procedure for determining rigid body transformation parameters. Journal of Biomechanics 28, 733-737.

Davis RB, Õunpuu S, Tyburski D, and Gage JR (1991). A gait analysis data collection and reduction technique. Human Movement Science 10, 575-587.

Hatze H (2002). The fundamental problem of myoskeletal inverse dynamics and its implications. Journal of Biomechanics 35, 109-115.

Lu TW and O'Connor JJ (1999). Bone position estimation from skin marker co-ordinates using global optimisation with joint constraints. Journal of Biomechanics 32, 129-134.

[Raphael Dumas]

Re: Optimization algorithms to correct movements

Hi Juan, hi all,

Talking about multi-body optimization (/global optimization /inverse kinematics), it seems that the methods is mainly useful as a starting point for further modeling (inverse dynamics, musculoskeletal modeling). However, as far as kinematics is concerned, it is principally the kinematics of the model and not the one of the subject that is obtained.


With that respect, there are few publications that have tried to validate multi-body optimization for lower limb models:

Stagni R, Fantozzi S, Cappello A. Double calibration vs. global optimisation: performance and effectiveness for clinical application. Gait Posture. 2009; 29(1): 119-22.

Andersen MS, Benoit DL, Damsgaard M, Ramsey DK, Rasmussen J. Do kinematic models reduce the effects of soft tissue artefacts in skin marker-based motion analysis? An in vivo study of knee kinematics. J Biomech. 2010; 43(2): 268-73.

Li K, Zheng L, Tashman S, Zhang X. The inaccuracy of surface-measured model-derived tibiofemoral kinematics. J Biomech. 2012; 45(15): 2719-23.

Gasparutto X, Sancisi N, Jacquelin E, Parenti-Castelli V, Dumas R. Validation of a multi-body optimization with knee kinematic models including ligament constraints. J Biomech. 2015; 48(6): 1141-1146.


It is in these papers that you will find comparison between different kinematic constraints as well as between model-based kinematics and reference kinematics obtained using pins or fluoroscopy.

Regards,