Information about project titled 'A model-based image-matching technique for 3D reconstruction - application to ACL injury situations'
A model-based image-matching technique for 3D reconstruction - application to ACL injury situations
|Details about the project - category||Details about the project - value|
|Project manager:||Tron Krosshaug|
|Supervisor(s):||Roald Bahr, Lars Engebretsen|
|Coworker(s):||Arne Ekeland, James Slauterbeck|
Our knowledge of the mechanisms causing ACL injuries in sports is limited. Even if a lot of attention has lately been given non-contact injuries in team sports like basketball, soccer, lacrosse and European team handball, it is still not clear what mechanisms or loads that are causing the rupture. Knowledge of the injury mechanism, as well as the risk factors, is of major importance to design optimal intervention strategies.
In order to gain insight in the injury mechanisms, several different approaches are used in the literature. Motion analysis of laboratory trials can give important information, but for ethical reasons obviously cannot look into injury situations. Cadaver studies are undoubtly of great value in finding relationships between external loads and ligament stress, but have clear limitations by the fact that the effects of muscles and other tissues cannot be replicated similar to the real in-vivo situation.
By utilizing mathematical simulations, more realistic premises can potentially be achieved.
However, the problem with all the mentioned approaches is the fact that it is not possible to verify if the studied situation actually represents the truth.
For this reason, there is a need for objective methods for analysing injury situations. However, methods for analyzing the mechanisms of injuries in sports from video sequences of injury situations are so far limited to a simple visual inspection, which has shown poor accuracy. We have previously developed a model-based image-matching method for extracting kinematics from uncalibrated videos. The purpose of this study was therefore to investigate if a new model-based image-matching technique could be successfully applied to estimate kinematic characteristics of three typical ACL injury situations.
Methods: A four-camera basketball video, a three-camera European team handball video and a single-camera downhill skiing video were imported into the program Poser® 4, where a skeleton model and a model of the surroundings were matched to the background image frame by frame. When the match was considered satisfactory, joint angles as well as velocity and acceleration of the center of mass were calculated using Matlab®.
Results: In the basketball and handball matchings, the skeleton and surrounding models were successfully matched to the background through all frames in all camera angles. Detailed time courses for joint kinematics and ground reaction force were obtained, while less information could be acquired from the single-view skiing accident.
In conclusion, the model-based image matching technique can be used to extract kinematic characteristics from video tapes of actual ACL injuries, and may provide valuable information on the mechanisms for ACL injuries in sports.