Concurrent validity of the Microsoft Kinect for Windows v2 for measuring spatiotemporal gait parameters

https://doi.org/10.1016/j.medengphy.2016.06.015Get rights and content

Highlights

  • We evaluated the performance of the Kinect v2 comparing with the GAITRite system.

  • Kinect v2 can accurately measure spatio-temporal gait comparing with GAITRite.

  • Kinect v2 can measure gait features at usual and fast pace and dual task walking.

  • This study suggests that Kinect v2 is a valid clinical tool for gait measurements.

  • This study offers others to use Kinect v2 for gait assessment in any environment.

Abstract

This paper presents a study to evaluate the concurrent validity of the Microsoft Kinect for Windows v2 for measuring the spatiotemporal parameters of gait. Twenty healthy adults performed several sequences of walks across a GAITRite mat under three different conditions: usual pace, fast pace, and dual task. Each walking sequence was simultaneously captured with two Kinect for Windows v2 and the GAITRite system. An automated algorithm was employed to extract various spatiotemporal features including stance time, step length, step time and gait velocity from the recorded Kinect v2 sequences. Accuracy in terms of reliability, concurrent validity and limits of agreement was examined for each gait feature under different walking conditions. The 95% Bland–Altman limits of agreement were narrow enough for the Kinect v2 to be a valid tool for measuring all reported spatiotemporal parameters of gait in all three conditions. An excellent intraclass correlation coefficient (ICC2, 1) ranging from 0.9 to 0.98 was observed for all gait measures across different walking conditions. The inter trial reliability of all gait parameters were shown to be strong for all walking types (ICC3, 1 > 0.73). The results of this study suggest that the Kinect for Windows v2 has the capacity to measure selected spatiotemporal gait parameters for healthy adults.

Introduction

Gait dysfunction is one of the most commonly reported impairments due to aging or acute events such as stroke or musculoskeletal injury [1], [2]. Gait assessment is a way of examining locomotion and body movements in individuals with gait dysfunction [3]. Despite the fact that most clinicians use observational techniques in their assessments, subjective observational evaluations are shown to be insufficient as a reliable approach in clinical settings [4]. Augmenting subjective observation assessments with objective measurements has been shown to strengthen the reliability and accuracy of the gait assessment [3].

Objective analysis in laboratory settings includes collection of spatiotemporal gait and biomechanical data such as velocity, step length and joint and limb motions. Pressure mats, force plates, and three dimensional motion capture systems are common tools used for gait assessments in laboratories. These measurement tools provide essential information about the functional performance of patients and enable more effective targeted treatment and therapeutic interventions [3]. Despite their high accuracy and reliability, the usability of motion capture systems is limited in clinical settings due to high cost, difficulty in the interpretation of captured data [5], [6], and elaborate set up requirements [7]. Pressure mats and/or force plates are more affordable and easier to use in clinics compared with motion capture systems; however, they only measure gait parameters based on foot placements and/or ground contact forces and cannot track other body parts, joint angles, or limb movements. Moreover, pressure mats: (1) limit the subjects to walk within the narrow width of the mat (88 cm) which is challenging for populations with vision impairment, stroke or brain injuries (since they show walking deviation and have difficulty avoiding the sensors on the mat or stepping off the mat [8]); (2) require the clinicians supervising the patients during the walking task to stay off the mat throughout the gait test; and (3) are highly sensitive to the amount of force applied by the subjects to the mat for outputting complete footfall imprints [9].

In the face of above mentioned challenges, the Kinect sensor (Microsoft Corporation, Redmond, WA) provides an opportunity to reconstruct both spatiotemporal gait and joint motion kinematics at a low cost (less than $200) and with minimal setup requirements. By processing captured depth information, the Kinect tracks people and their full body poses within its field of view at the real-time rate of 30 frames per second. Moreover, the Kinect sensor offers additional advantages over the existing gait assessment tools, including the capacity to provide longitudinal and quantitative analysis of gait in private homes. Furthermore, it does not require markers to track joint motions, and consequently will overcome the limitations associated with wearable sensors and 3D motion capture systems. Consequently, researchers have explored using the Kinect sensor for objective gait analysis [10], [11], [12], [13], [14], [15].

The validity and reliability of the older version of the sensor (Kinect for Windows v1) has been thoroughly analyzed and compared against 3D motion capture systems [10], [11], [13], [16], vision-based tracking algorithms [14], [15], and wearable sensors [17] by several research groups.

Recently, a new version of Kinect for Windows (K4W) v2 has been released that provides an expanded field of view, 70.6 × 60 degrees vs. 58.5 × 46.6 degrees in Kinect v1. The new sensor captures images of significantly higher resolution, 6.75 times more color pixels and 2.82 times more depth pixels than the older version, and is capable of tracking 25 body joints of up to 6 people within its field of view, as compared with 20 body joints of up to 2 people in the older version. As a result, it is of great clinical interest to establish the concurrent validity of the K4W v2 to make objective gait measurements and study the sensor’s accuracy and limitations.

The current paper focuses on establishing the criterion validity and test retest reliability of the K4W v2 on gait parameters. The specific objective of this study was to establish the concurrent validity of K4W v2 against the GAITRite mat for measuring the kinematic parameters of gait in healthy subjects, including stance time, step time, step length and gait velocity. The GAITRite mat (CIR systems Inc., Sparta, NJ) is one of the most commonly used pressure sensing walkways, and its reliability in observing and extracting spatiotemporal gait parameters has been established in both younger and older subjects [18]. The presented results could allow researchers in the gait and balance rehabilitation field to use K4W v2 as a valid clinical tool and also to employ it in longitudinal gait analysis studies in the home environment.

Section snippets

Study description

Twenty (20) healthy young adults with no history of medical, mobility or chronic musculoskeletal disorders (mean ± standard deviations; age: 28.8 ± 7.1 years, height: 171.2 ± 10.5 cm and weight: 66 ± 16 kg) were recruited to participate in this study. Ten (10) participants (50%) were woman. Data collection took place at the Balance Mobility and Falls Clinic (BMFClinic) in the Toronto Rehabilitation Institute-UHN University Centre. This study was approved by the Research Ethics Boards at the

Results

Since no asymmetries in walking were apparent in our healthy participants, spatiotemporal parameters from all sequences captured by the two Kinect sensors for each participant were combined and averaged for the left and right sides. Average spatiotemporal data were used for comparison analysis. The comparison results between the K4W v2 and GAITRite in terms of the group mean differences, Bland–Altman 95% limit of agreement and correlation coefficients for three types of walking (usual pace,

Discussion and limitations

This study presents the first attempt to evaluate the accuracy of the Kinect for Windows v2 to measure spatiotemporal gait features in comparison with the GAITRite mat. The K4W v2 is a cost effective and easy to use sensor. Along with the algorithm used in this study for measuring gait parameters, the K4W v2 is a good candidate for making objective full gait assessments to provide more effective targeted treatments and therapeutic interventions in gait. It is therefore very important to study

Conflict of interest

No conflict of intrest.

Acknowledgment

The authors would like to thank Ms. Bing Ye for her assistance in acquiring the data presented. The authors would also like to thank the Balance, Falls and Mobility Clinic at the Toronto Rehabilitation Institute for providing us with required equipment and facilities throughout the study. Study funding was provided by the Toronto Rehabilitation Institute-University Health Network. The authors would also like to thank Marge Coahran for proofreading the article.

References (31)

  • L. Sudarsky

    Gait disorders: prevalence, morbidity, and etiology

    Adv Neurol

    (2001)
  • B. Toro et al.

    A review of observational gait assessment in clinical practice

    Phys Theory Pract

    (2003)
  • G.K. Rose

    Clinical gait assessment: A personal view

    J Med Eng Amp Technol

    (1983)
  • E. Stone et al.

    Evaluation of an inexpensive depth camera for passive in-home fall risk assessment

    Proceedings of the 5th international conference on pervasive computing technologies for healthcare (PervasiveHealth).

    (2011)
  • B. Taati et al.

    Towards a real-time, configurable, and affordable system for inducing sensory conflicts in a virtual environment for post-stroke mobility rehabilitation: vision-based categorization of motion impairments

    Proceedings of the 9th international conference on disability, virtual reality and associated technologies

    (2012)
  • Cited by (84)

    • Concurrent validity of evaluating knee kinematics using Kinect system during rehabilitation exercise

      2021, Medicine in Novel Technology and Devices
      Citation Excerpt :

      Therefore, many studies have used the Kinect v2 to replace traditional 3D motion capture systems in several conditions. For gait analysis, the Kinect is an acceptable alternative system for assessing spatiotemporal gait parameters and lower limb kinematics during stair climbing in some clinical situations [16–18]. For measuring upper limb motion, it has been used in clinical settings as a reliable and valid device for measuring active shoulder motion and upper limb kinematics during functional movements [12,19–21].

    • Effects of camera viewing angles on tracking kinematic gait patterns using Azure Kinect, Kinect v2 and Orbbec Astra Pro v2

      2021, Gait and Posture
      Citation Excerpt :

      Many researchers have investigated whether these affordable and portable depth sensors can be used effectively for clinical assessment [11], gait analysis [4,5,12], rehabilitation [13], and fall prevention [14,15]. Previous studies reported depth sensors could measure spatiotemporal gait parameters with excellent agreement [1,2,5,16–18], but they could not provide consistent and accurate kinematic measurements to report joint angles [4,5,11,12]. Furthermore, previous researchers had no consensus about the optimal camera viewing angle relative to human body for measuring treadmill walking [4,10,11], as most studies applied the conventional 0° frontal view relative to the depth sensor [2,3,5,10,17,18], while few studies placed the sensor at oblique angles (e.g., 45°) [19].

    • Parkinson's Disease Progression Prediction Using Longitudinal Imaging Data and Grey Wolf Optimizer-Based Feature Selection

      2024, International Journal of Intelligent Systems and Applications in Engineering
    View all citing articles on Scopus
    View full text