Three-dimensional Reconstruction Of Acl Injuries

Introduction
Although the biomechanics of ACL injuries is a popular research topic, precise descriptions of the motion involved is
not available. We have therefore developed a model-based image-matching technique to reconstruct human motion
from uncalibrated video sequences of actual injury situations. The method has been validated against traditional 3D
motion analysis system with good results [1]. The purpose of the present study was to test the applicability of the
method on two different ACL injuries from different sports.

Methods
The first situation was a non-contact plant-and-cut injury with three different camera views (Fig. 1a). In situation two
(Fig. 1b), a downhill skier suffers an ACL injury to the left leg as the right ski looses the grip, causing a wide snowplough
position. The 3D modelling program Poserâ provided the environment for the model matching procedure. A
customized skeleton computer-model was built based on anthropometrical measurements which were obtained from
each injured athlete. A computer model of the handball field was built and the Poser cameras were "calibrated" by
adjusting the translation, orientation and focal length parameters for each time frame, until our field model matched the
background video sequences. This was not possible with the same precision for the downhill video, since the positions
of landmarks in the background were unknown. Finally, a frame-by-frame matching of the skeleton model to the athlete
in the video was completed.

Results
NOTA: O texto com a iconografia está no anexo.

Discussion / Conclusions
The new model-based image-matching method provides a means to reconstruct and analyse injury situations in three
dimensions. Such information can potentially aid our understanding of how injuries occur (e.g. ACL injuries), and form
the basis for preventive measures in sport.

References
[1]. Krosshaug T & Bahr R. J. Biomech. (In revision)
a)

NOTA: O texto com a iconografia está no anexo.

Arquivo