Development and validation of a novel marker tracking approach based on the low-cost Microsoft Kinect v2 sensor for assessing lower limb biomechanics during single-leg squat and treadmill gait

Abstract

Pubescent females are twice more likely to suffer a non-contact ACL injury than their male counterparts. This disparity has been correlated with multiple concurrent factors, including biomechanical, anatomical and hormonal changes. ACL ruptures require serious and costly surgical interventions, which could be avoided if subjects at higher risk of injury were more carefully monitored and trained. Three-dimensional motion analysis is required to identify individuals at risk of ACL injury. Multi-camera optical systems are the gold standard for 3D motion capture, but they are very expensive and cumbersome. The aim of this thesis was to make motion analysis more accessible, developing an affordable and compact 3D motion tracking methodology, alternative to conventional multi-camera systems.

A novel tracking approach was developed using Microsoft Kinect v2, employing custom-made coloured markers and computer vision techniques. This methodology was denoted as Kinect coloured marker tracking (KCMT). The accuracy of KCMT relative to a conventional Vicon motion analysis system was measured performing two Bland-Altman analyses of agreement, the first using single-leg squat (SLS) as benchmark task, the second using treadmill locomotion.

The objective of the first study was to determine if KCMT-derived sagittal joint angles of the lower limb were accurate enough to allow discerning individuals at risk of ACL injury from those not at risk. Eleven healthy participants were asked to perform three SLS trials, while three-dimensional marker trajectories were simultaneously recorded using Vicon and KCMT respectively. Joint angles from the two systems were calculated via inverse kinematics using OpenSim. The limits of agreement (LOA) of the joint angles were −16°, 13° for hip flexion, −12°, 0° for knee flexion and −12°, 9° for ankle flexion. These results indicated that the agreement between KCMT and Vicon was joint dependent, and that further work was required for the novel methodology to replace conventional marker-based motion capture systems for the identification of ACL injury risk from SLS data.

In the second study, an improved data collection protocol for the KCMT was used. Twenty participants were recruited, and markers placed on bony prominences near hip, knee and ankle. Three-dimensional coordinates of the markers were recorded during treadmill walking and running. The LOA of marker coordinates were narrower than −10 and 10 mm in most conditions, however a negative relationship between accuracy and treadmill speed was observed along Kinect depth direction. LOA of the knee angles measured in the global coordinate system were within −1.8°, 1.7° for flexion in all conditions and −2.9°, 1.7° for adduction during fast walking, suggesting that KCMT may be capable of discerning between subjects at risk of ACL injury and controls. The proposed methodology exhibited good agreement with a marker-based system over a range of gait speeds and, for this reason, may be useful as low-cost motion analysis tool for selected biomechanical applications.