Affordable gait analysis using augmented reality markers

• Published in: PloS one Vol. 14; no. 2; p. e0212319
• Main Authors: Nagymáté, Gergely, Kiss, Rita M.
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science 14.02.2019; Public Library of Science (PLoS)
• Subjects: Accuracy; Adolescent; Adult; Aged; Aged, 80 and over; Augmented reality; Biology and Life Sciences; Biomechanical Phenomena; Biomechanics; Biomedical engineering; Calibration; Cameras; Contact angle; Coordinate transformations; Engineering and Technology; Error analysis; Exercise Test; Fitness equipment; Gait; Gait Analysis; Gait recognition; Health aspects; Hip; Hip - physiology; Humans; Image processing; Image processing software; Image Processing, Computer-Assisted - methods; Informatics; Kinematics; Knee; Knee - physiology; Laboratories; Markers; Mathematical analysis; Mechanical engineering; Medical examination; Medical research; Medicine and Health Sciences; Middle Aged; Motion capture; Optics; Parameters; Pelvis; Pelvis - physiology; Physical Sciences; Posture; Range of Motion, Articular; Reference systems; Sensors; Software; Spatial analysis; Walking; Young Adult; Hungary; New York; United States > US
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0212319

A typical optical based gait analysis laboratory uses expensive stereophotogrammetric motion capture systems. The study aims to propose and validate an affordable gait analysis method using augmented reality (AR) markers with a single action camera. Image processing software calculates the position and orientation of the AR markers. Anatomical landmark calibration is applied on the subject to calibrate each of the anatomical points with respect to their corresponding AR markers. This way, anatomical points are tracked through AR markers using homogeneous coordinate transformations, and the further processing of gait analysis is identical with conventional solutions. The proposed system was validated on nine participants of varying age using a conventional motion capture system on simultaneously measured treadmill gait trials on 2, 3 and 4.5 km/h walking speeds. Coordinates of the virtual anatomical points were compared using the Bland-Altman analysis. Spatial-temporal gait parameters (step length, stride length, walking base, cadence, pelvis range of motion) and angular gait parameters (range of motion of knee, hip and pelvis angles) were compared between measurement systems by RMS error and Bland-Altman analysis. The proposed method shows some differences in the raw coordinates of virtually tracked anatomical landmarks and gait parameters compared to the reference system. RMS errors of spatial parameters were below 23 mm, while the angular range of motion RMS errors varies from 2.55° to 6.73°. Some of these differences (e.g. knee angle range of motion) is comparable to previously reported differences between commercial motion capture systems and gait variability. The proposed method can be a very cheap gait analysis solution, but precision is not guaranteed for every aspect of gait analysis using the currently exemplified implementation of the AR marker tracking approach.