In partial support of our third hypothesis, 2D-Cam peak frontal plane and 3D dynamic valgus data showed moderate
correlation within subjects for both side step and side jump, and limited association for shuttle run movements.
Considering that intrasubject variations in peak 3D valgus for these movements were typically smaller than intersubject
variations (fig 5), a lower mean r2 intrasubject comparison is intuitive.
It is worth noting, however, that subjects with the
largest between-trial variations in peak 3D valgus also had
the highest
r
2
values (fig 6). Reliable detection of between-
trial differences in peak 3D knee valgus may at least be
possible by the 2D approach for these subjects. This result is
important, as it suggests that a 2D method may still offer
some potential as an evaluation tool within training
programmes aimed at reducing valgus motions. Athletes
exhibiting the greatest 3D knee valgus angles, for example,
are known to be at the greatest risk of suffering an ACL injury
by this mechanism.
11
They will therefore probably require
relatively large reductions in knee valgus to achieve ‘‘safe’’
normative magnitudes.
11 13
If training can reduce peak 3D
valgus motions by magnitudes similar to the largest between-
trial variations observed here for this variable, then reliable
detection of these changes should be possible with a 2D
method. We did not implement any form of training in this
study, nor did we instruct subjects to alter their valgus
patterns in any way during movement trials. Hence, the true
potential for the 2D method as a training evaluation tool
remains speculative. More work is necessary to determine
whether long term modifications to valgus motions are
possible, and, if so, by what magnitude they can realistically
be changed, before inclusion of 2D video as an evaluation can
seriously be considered.
One reason for conducting this study was to determine
whether a 2D video analysis tool could be successfully
implemented within large scale neuromuscular intervention
programmes attempting to reduce ACL injury rates. These
results suggest that reliable screening of ‘‘at risk’’ people for
specific movement tasks may certainly be possible with a 2D
method. Although it may provide a cost effective alternative
to current 3D motion analysis technologies, the relatively
large processing requirements inherent in the 2D approach,
particularly the manual identification of joint centres, does
not lend itself to the already labour intensive requirements of
intervention training programmes.
18
Automated marker
tracking software is available,
30
and markerless methods are
being developed,
31
which would significantly expedite the
processing of 2D video data obtained with low cost cameras.
Such technologies should at the very least be considered for
large scale intervention studies aimed at prevention of knee
injuries related to dynamic valgus. However, it is crucial to
first evaluate the reliability of a 2D video approach for the
specific population/s and movement/s to be tested.
As noted, state of the art 3D motion analysis technologies
and processing computational algorithms appear to provide a
‘‘gold standard’’ in terms of both accurate quantification of
lower limb joint motions and detection of instances when
these motions become large enough to cause injury.
21 32
We
have shown here that comparable success in terms o