Movement assessment tools are widely used in practice to monitor injury risk factors in athletes. However, an issue with these tools is the trade-off between the reliability and validity on one hand and the practicality on the other hand. The Windows Kinect has been proposed as an addition to current movement assessment tools because it has depth-sensing technology and it can collect 3-dimensional kinematic data of anatomical landmarks during dynamic movements via markerless tracking. Therefore, this thesis aimed to show the development of a reliable, valid, and practical movement assessment tool (athletic movement analysis tool, [AMAT]) that makes use of depth-sensing technology and a laptop (i5 processor or higher, Windows operating system) to collect kinematic data of athletes during dynamic movements. To that purpose, the first study discussed the strengths and weaknesses of the Windows Kinect. Moreover, it described the development of algorithms to improve the kinematic data collection with the Kinect. Foot markers with retroreflective markers were developed and an algorithm was developed to track these markers with the Kinect. Moreover, an algorithm was developed to calculate the position of the centre of mass and an algorithm was developed to determine the frame of initial contact. In the next study, the foot markers and normal retroreflective markers were placed on 18 different positions within the camera view to determine the static reliability and validity of the foot marker tracking algorithm when compared to Vicon. The technological error of the AMAT ranged from 1.36 millimetres on the positions closest to the camera to 3.30 millimetres on the positions furthest away from the camera. The mean difference between the marker positions as measured with AMAT and Vicon system ranged from -4.51 to 16.23 millimetres. These outcomes imply that the foot marker tracking is reliable and valid to track the markers in static situations, but it was shown that the AMAT is less reliable when collecting data on positions further away from the Kinect. In the third study, the foot marker tracking algorithm and a tape measure were used to measure jump distances, to determine whether the AMAT is able to track the foot markers during dynamic situations. The mean differences between the AMAT and manual measurements were trivial (-1.69 to 2.41 millimetres). This implies that the foot marker tracking algorithm is valid to track the foot markers during dynamic movements. In the fourth study, the algorithm that calculates the centre of mass position (centre of mass algorithm) was validated by comparing the outcomes of this algorithm with the centre of mass position collected with Vicon during horizontal jumps. The correlations were moderate to extremely high in the medio-lateral axis (0.65 to 1.00), extremely high in the posterior-anterior axis (0.99 to 1.00) and trivial to extremely high in the superior-inferior axis (-0.08 to 0.98). In the last study of this thesis, the standing broad jump performance and the ability to maintain balance of adolescent female soccer players were monitored during a full season with the AMAT. Here it was found that the within-subject variability of the jump performance was small to moderate and that a substantial improvement in jump performance was found over the season. The within-subject variability of the ability to maintain balance ranged from moderate to extremely large. This implies that the jump performance can be reliably collected with the AMAT, whereas the ability to maintain balance cannot be collected in a reliable manner. In total, this thesis showed that the AMAT can reliably and validly track the foot markers and determine the position of the centre of mass. However, more research on other algorithms remains necessary to determine whether kinematic data of other anatomical landmarks can also be collected in a reliable and valid manner.
|Date of Award||Apr 2019|
|Supervisor||Iain Spears (Supervisor), Matthew Weston (Supervisor) & Matthew Portas (Supervisor)|