Information about a piece of news titled Poor accuracy in video analysis based on visual inspection
Poor accuracy in video analysis based on visual inspection
Intoduction
In the latest issue of Gait & Posture, the validity of visual inspection as a method was tested. The results were not impressive.
Several studies on e.g. the mechanisms of non-contact ACL injuries are published were simple visual inspection was used to provide estimates on knee and hip joint angles, weight distribution, velocities etc. In order to determine the accuracy and precision of this method, six international researchers, most of them with extensive experience in such analyses, were invited to participate. Video sequences of running, landing and cutting maneuvers were assessed and compared with the gold standard – measurements of the motions using a eight-camera marker based high speed motion analysis system. First, a pre test was conducted before the researchers went through a training program of structured feedback using videos of similar situations. The original test was then repeated.
The average error for e.g. knee flexion was found to be 19 degrees (underestimation). In addition, the standard deviation between the analysts was found to be high. Only small changes were found between the pre and post test.
In conclusion, it is reason to be cautious when interpreting the results from previous studies using visual inspection. It is possible that knee and hip flexion angles are underestimated with a considerable margin.
This study was part of the PhD thesis of Tron Krosshaug. Co-authors were Atsuo Nakamae, Barry Boden, Lars Engebretsen, Gerald Smith, James R. Slauterbeck, Tim Hewett and Roald Bahr.
Read the article here.