Reliability of a markerless motion capture system to measure the trunk, hip and knee angle during walking on a flatland and a treadmill.

Markerless motion capture system (MLS) using an infrared sensor such as Microsoft Kinect has been used for gait analysis. Several studies have shown that kinematic measurements of trunk and lower limb joint angles during walking measured by MLS are valid. However, the reproducibility, presence of systematic error, or degree of random error of kinematic measurements during walking using MLS with Kinect v2 were not demonstrated. This study was made to confirm the reliability of kinematic measurements using Kinect v2 during gait. Twenty-two young, injury-free individuals volunteered to participate. Walks were made at 2 miles per hour (mph) on both the flatland and the treadmill. Intra-class correlation coefficients (ICCs) were calculated, systematic errors identified, and minimal detectable changes (MDCs) were estimated to assess the reliability of kinematic measurements of trunk, hip, and knee joint angles during walking. For trunk angles measured on the flatland, ICC was higher than 0.6, systematic error was smaller, and MDC was 2.2° smaller than that in gait on the treadmill (6.6°). In hip joint angles measured on the flatland, although systematic error was slight, ICC was not higher than on the treadmill and MDC exceeded 5°. The results for the knee joint were similar to those of the hip joint. Kinect can detect kinematic abnormalities of the trunk during slow walking on the flatland easier than on the treadmill.