•  
  •  
 

EXPLORING THE TEST-RETEST RELIABILITY OF MARKERLESS MOTION CAPTURE FOR OUTDOOR WALKING

Abstract

BACKGROUND: Markerless motion capture is a rapidly advancing technology and has been tested indoors and proven reliable. However, there are no studies that have tested a portable markerless motion capture system outdoors and out of the lab setting. The purpose of this study is to determine test-retest reliability of knee kinematics using a portable markerless motion capture system during a walking test in healthy adults. METHODS: Twenty-one participants (6M, 15F, age=23.11 ± 1.89 years, height = 167.17 ± 7.89 cm, mass = 69.85 ± 23.11 kg) performed two walking tests each separated by a minimum of five days. Participants were instructed to walk at their comfortable walking pace, which was recorded at the beginning of their trials. Eight trials were recorded within 5% of their recorded walking pace. Videos were recorded using 8 video cameras (OptiTrack Prime Color, Corvallis, OR. Natural Point Inc.) at 60hz. After collection, video data was exported and reduced in a markerless motion capture software (Theia3D, Kingston, Ontario). The maximum and minimum knee flexion and extension and knee abduction and adduction angles during the first 50% of stance phase were collected for data analysis. Test-retest reliability was calculated with intraclass correlation coefficients (ICCs) between testing time 1 and time 2. Reliability was interpreted as excellent (>0.90), good (0.90-0.75), moderate (0.75-0.50), and poor (<0.50). Precision was calculated with the standard error of measurement (SEM). RESULTS: ICC and SEM demonstrated good reliability and precision for peak knee flexion (ICC=0.859 (0.722,0.931), SEM=1.84). Reliability and precision were moderate for knee extension (ICC=0.732 (0.506, 0.864), SEM=1.23). Reliability was moderate for knee abduction (ICC=0.547 (0.547,0.757), SEM=1.47) and knee adduction (ICC=0.576 (0.273, 0.775), SEM=1.62). CONCLUSION: Our results suggest that sagittal plane movements like knee extension and flexion demonstrate good to moderate reliability, however, they can still improve. Movements in the frontal plane are less reliable than the sagittal plane. Possible solutions to improve reliability would include more cameras or positioning in a different pattern enabling a better view of the knee joint. Artificial Intelligence (AI) may need improvements, and future research should continue to investigate reliability and precision as improvements are made.

This document is currently not available here.

Share

COinS