Visual inspection suggests that the errors from different marker placement
are of the
same order as walk to walk and patient to patient variability although we have not
analysed this any more formally. (Any suggestions as to how to extend ANOVA techniques
to the Coefficient of Variation or a similar measure?).
I think it is good practice to separate out left and right sides in
the averaging. We
inspected these visually and decided the differences were small. As the main object of
creating the database was to display in the background of our clinical data it seemed
more sensible to combine data from both sides.
All data was collected with six camera.VICON 370 system. Standard VCM
marker set was
used. All thigh markers were placed in the plane containing the greater trochanter and
knee joint axis, this was judged by eye with the use of a mirror to reduce parallax (i.e.
no use of either KAD or my correction algorithm). VCM was used to analyse data with
ASIS-to-ASIS and ASIS-to-trochanter distances left blank (please note that the term
"standard VCM analysis" can actually cover a range of biomechanical models depending on
which parameters are left blank and which are filled in).
Force plate data was not captured. I have a feeling that by starting
off subjects at
the same point and only recording the walks on which they hit force plates a fixed
distance from the starting point we are automatically selecting the more repeatable
Gait Analysis Service Manager
Musgrave Park Hospital
Tel: +44 (0)1232 669501 ext 2155
Fax: +44 (0)1232 611064