A study of accuracy for a single Time of Flight camera capturing knee flexion movement

Methods with markers give some accurate results but inconvenients can appear with some of them: patients have to be instrumented (markers on the skin), many cameras have to be used. New motion analysis methods appeared in the recent years, using markerless acquisitions. One of them is particularly interesting: Time-of-Flight (ToF) cameras. In this study we decided to estimate the precision of measures of a ToF camera during the acquisition of a knee flexion movement on an isolated cadaveric leg which is posed on a table. The leg is articulated with the table at the femoral head by a ball and socket joint and set in motion by a manipulator. A ToF camera, fixed around 1.2 m above the leg, was filming the anterior part of the leg during movement. Four intracortical pins were implanted in bones: two near the proximal and distal ends of the femur and the tibia. Those pins, instrumented with reflective markers, allow us to follow bones movements using 9 Vicon cameras. Flexion angle varied from 10 degrees (initial position) to 90 degrees (flexion max obtained on our cadaver leg). As a consequence, we were able to measure by two means the flexion angle during the movement. We used the Vicon system as our gold standard (accuracy around 1 mm) and we have compared the two measures. Major differences appears for an angle superior to 70 degrees, but the mean of difference between the two systems for angles between 10 degrees to 70 degrees is less than 2.5 degrees with a standard deviation of 2.02 degrees and a root mean square around 8 degrees. Results from two methods used were relatively close and showed the capability of a ToF camera for measuring the flexion movement during gait (same order of range). The advantages of markerless solution are discussed relative to the objective of the measurement.