Journal of Biomechanics 71 (2018) 281–285

Contents lists available at ScienceDirect

Journal of Biomechanics
journal homepage: www.elsevier.com/locate/jbiomech
www.JBiomech.com

Short communication

Validity of time series kinematical data as measured by a markerless

motion capture system on a flatland for gait assessment
Ryo Tanaka a,⇑, Haruka Takimoto a, Takahiro Yamasaki a, Ariaki Higashi b
a
Department of Rehabilitation, Hiroshima International University, 555-36 Kurose-gakuendai, Higashi-hiroshima, Hiroshima 739-2695, Japan
b
Systemfriend Inc., 1-11-20, Itsukaichiekimae, Saeki, Hiroshima, Hiroshima 731-5125, Japan

a r t i c l e i n f o a b s t r a c t

Article history: As a cost-effective, clinician-friendly gait assessment tool, the Kinect v2 sensor may be effective for
Accepted 28 January 2018 assessing lower extremity joint kinematics. This study aims to examine the validity of time series kine-
matical data as measured by the Kinect v2 on a flatland for gait assessment. In this study, 51 healthy sub-
jects walked on a flatland while kinematic data were extracted concurrently using the Kinect and Vicon
Keywords: systems. The kinematic outcomes comprised the hip and knee joint angles. Parallel translation of Kinect
Microsoft Kinect v2 data obtained throughout the gait cycle was performed to minimize the differences between the Kinect
Kinematics
and Vicon data. The ensemble curves of the hip and knee joint angles were compared to investigate
Validity
Gait
whether the Kinect sensor can consistently and accurately assess lower extremity joint motion through-
out the gait cycle. Relative consistency was assessed using Pearson correlation coefficients. Joint angles
measured by the Kinect v2 followed the trend of the trajectories made by the Vicon data in both the
hip and knee joints in the sagittal plane. The trajectories of the hip and knee joint angles in the frontal
plane differed between the Kinect and Vicon data. We observed moderate to high correlation coefficients
of 20%–60% of the gait cycle, and the largest difference between Kinect and Vicon data was 4.2°. Kinect v2
time series kinematical data obtained on the flatland are validated if the appropriate correction proce-
dures are performed. Future studies are warranted to examine the reproducibility and systematic bias
of the Kinect v2.
Ó 2018 Published by Elsevier Ltd.

1. Introduction
Gait abnormality is the leading determinant of disability in
adults with stroke (Chiou and Burnett, 1985), and gait analysis is
valid for multiple applications (Eltoukhy et al., 2017b). Microsoft’s
Kinect version 1 (v1), released in 2011, is an inexpensive, portable,
and user-friendly markerless motion capture system. Microsoft
then released Kinect v2 in 2014. Both Kinect v1 and v2 have differ-
ent working principles, resolutions, and frame rates. By assessing
the validity of Kinect v2 data obtained during gait, Eltoukhy et al.
(2017b) found that Kinect v2 is an acceptable tool for assessing
sagittal plane knee and hip range of motion (ROM) and joint angles
throughout the gait cycle.
However, they validated data obtained by the Kinect v2 system
during gait on a treadmill. Although Eltoukhy et al. (2017a) primar-
ily focused on validating Kinect v2 use for flatland gait analysis, the
⇑ Corresponding author.
E-mail addresses: r-tanaka@hs.hirokoku-u.ac.jp (R. Tanaka), harukatakimo-
to106@gmail.com (H. Takimoto), t-yama@hs.hirokoku-u.ac.jp (T. Yamasaki), higa-
shi@systemfriend.co.jp (A. Higashi).
validity of Kinect v2 time series kinematic data during gait analysis
has not been demonstrated. To support the practicality of Kinect v2
for gait assessment, Kinect v2 time series kinematic data obtained
during gait on flatland warrants validation because gait analysis in
the clinical setting is performed on such surfaces. We aim to vali-
date time series kinematic data assessed by Kinect v2 on flatland
for gait assessment.
2. Materials and methods
In total, 51 injury-free individuals [mean age: 20.9 (standard
deviation (SD): 0.2) years; mean height: 166.9 (SD: 8.8) cm; mean
mass: 61.1 (SD: 1.0) kg; male: 35] volunteered. The Hiroshima
International University ethics committee approved this study
(15–43), and all participants provided written informed consent.
We used the Microsoft Kinect v2 sensor (Microsoft Corp., Red-
mond, WA) and obtained a skeleton model directly from the Micro-
soft Kinect official Software Development Kit v2. Before data
collection, we placed the Kinect v2 sensor on a tripod 0.8 m above
the floor. The stick figure comprises 25 points as estimates of
subjects’ joint centers (JCs). We used the anatomical landmarks

https://doi.org/10.1016/j.jbiomech.2018.01.035
0021-9290/Ó 2018 Published by Elsevier Ltd.
282 R. Tanaka et al. / Journal of Biomechanics 71 (2018) 281–285

of the ankle, knee, and hip JCs to calculate the knee and hip joint
angles, respectively. The Mobile Motion Visualizer AKIRA (System-
friend Inc., Itsukaichi, Japan) was used to record Kinect data simul-
taneously with RGB video data.
Three-dimensional motion analysis system data were acquired
at 120 Hz using a seven-camera Vicon MX (Vicon Motion Systems,
Oxford, UK). We placed 33 reflective markers, including markers
placed on the acromion process, elbow, radial styloid process, top
of the iliac crest, anterior superior iliac spine, posterior superior
iliac spine, superior aspect of the greater trochanter, medial and
lateral femoral condyles, midpoint between the greater trochanter
and the lateral femoral condyles, medial and lateral malleoli, mid-
point between the lateral knee joint line and the lateral malleolus,
head of the first and fifth metatarsals, and the calcaneal tuberosity.
With these anatomical markers, we constructed coordinate sys-
tems for the pelvis, thigh, shank, and foot segments. JCs of the
hip, knee, and ankle were approximated as described previously
(Andriacchi et al., 1982; Kito et al., 2010; Koyama et al., 2015;
Kurabayashi et al., 2003; Tateuchi et al., 2017). JC of the knee on
the frontal plane was located by identifying the midpoint of a line
linking the medial femoral condyle marker to the lateral femoral
condyle marker. Furthermore, JC of the ankle was located by iden-
tifying the midpoint of a line linking the medial malleolus marker
to the lateral malleolus marker. The positions of these markers
were obtained using BodyBuilder software (Vicon Motion Systems)
and acquiring image data.
Fig. 1. Trajectories of the hip joint angles in both the sagittal and frontal planes.
Participants wore tight-fitting shorts and an upper body gar- Average and 95% confidential interval of joint angle are demonstrated. The X axis
ment that allowed the placement of the reflective markers without shows the gait cycle (%) and the Y axis shows the angle (degree).
shoes. We performed gait trials along a walkway with an embed-
ded force platform (AMTI; Watertown, MA). With the platform,
we identified ground contact and toe-off of the foot during the tri- measured by the Kinect followed the trend of joint trajectories
als. Participants started each gait trial approximately 8 m from the evaluated by Vicon. The minimum correlation coefficient through-
Kinect. All participants performed the gait trials at their own com- out the gait cycle was 0.43, and the most significant difference
fortable pace. Each participated in five gait trials, and we assessed between the data from Kinect and Vicon was 4.2°.
the data of one trial with the typical waveform of the floor reaction Unlike the sagittal plane, the frontal plane’s curve made by
force in the stance phase. Kinect data did not match the curve made by Vicon data. The cor-
Analog signals and lamps were used to synchronize the Vicon relation coefficients were >0.5 from 20% to 60% of the gait cycle,
and Kinect data. Analog signals were recorded with Vicon data. and the most significant difference was 1.4°.
Lamp emitted light simultaneously with the analog signal trans- Fig. 2 shows the knee joint angle trajectories, and Table 2 shows
mission, which was recorded as RGB video data simultaneously the knee joint angles during the gait cycle. Similar to the hip joint
with Kinect data using Mobile Motion Visualizer AKIRA software. angles in the sagittal plane, the knee joint angles measured by the
We aligned the time stamps of the lamp light on Kinect data and Kinect followed the trend of joint trajectories assessed by Vicon.
that of the analog input signal on Vicon data and synchronized Most of the correlation coefficients throughout the gait cycle were
the Kinect and Vicon data. >0.5. Except for 10–20% period of the gait cycle, the maximum joint
We used spline interpolation to resample Vicon data to 30 Hz angle difference between the two systems was 3.4°.
before analysis. We identified the phases of the gait cycle with In the frontal plane, similar to the hip joint angle, the Kinect
the gait event time points of toe-off and ground contact. We per- data’s curve did not match the Vicon data’s curve. However, higher
formed kinematic analysis for the full gait cycle of each participant. correlation coefficients (r > 0.5) were observed throughout the gait
Coordinate data from the Kinect and Vicon systems for the ankle, cycle, and the maximum joint angle difference between the two
knee, and hip JCs were acquired and used to calculate the projec- systems was 1.5°.
tion angle of the knee and hip joints toward the sagittal and frontal
planes (see Appendix). Finally, we performed parallel translation of
the Kinect v2 gait data obtained throughout the gait cycle to min- 4. Discussion
imize the differences between Kinect and Vicon data.
We compared ensemble curves and the hip and knee joint During normal gait, only a single peak of hip extension and flex-
angles throughout the gait cycle to determine whether the Kinect ion occurs in each gait cycle, and peak hip extension arises before
could consistently and accurately measure the lower extremity the swing phase (Perry, 1974); both of these were demonstrated by
joint motion. Pearson correlation coefficients confirmed the valid- our hip joint’s Vicon data in the sagittal plane. Thus, our Vicon data
ity of the Kinect v2 gait data. Statistical analyses were performed could be used as the reference and compared with Kinect data.
using IBM SPSS statistics 22 (IBM SPSS, Tokyo, Japan). Although the waveform of Kinect v2 data was similar to that of
Vicon data, 95% confidence intervals (CIs) of the Kinect v2 means
in each frame did not always overlap with those of the Vicon
3. Results means, and the largest difference was 4.2°. Our results suggest that
the pattern of sequential change in the hip angle in gait measured
Fig. 1 shows the hip joint angle trajectories, and Table 1 shows by Kinect v2 could be used as conclusive findings rather than that
the hip joint angles during the gait cycle. The hip joint angles of the Kinect v1 (Pfister et al., 2014) and that the absolute degree of
 R. Tanaka et al. / Journal of Biomechanics 71 (2018) 281–285 283

Table 1
Hip joint angles during the gait cycle in both the sagittal and frontal planes.

Hip (sagittal)
Gait cycle Kinect System Vicon System r Difference
Mean 95% CI Mean 95% CI Mean 95% CI
0% 15.8 (17.3, 14.3) 11.7 (12.7, 10.7) 0.58** 4.1 (5.4, 2.8)
10% 10.8 (12.2, 9.4) 8.4 (9.4, 7.4) 0.68** 2. (3.4, 1.4)
20% 0.2 (0.9, 1.3) 0.3 (1.3, 0.7) 0.78** 0.6 (0.2, 1.3)
30% 11.8 ( 10.8, 12.8) 9.0 ( 8.2, 9.8) 0.61** 2.7 ( 2.0, 3.6)
40% 21.4 ( 20.3, 22.5) 17.2 ( 16.4, 18.0) 0.43** 4.2 ( 3.2, 5.2)
50% 25.6 ( 24.4, 26.8) 21.5 ( 20.8, 22.2) 0.49** 4.1 ( 3.0, 5.2)
60% 17.9 ( 16.4, 19.4) 16.0 ( 14.9, 17.1) 0.74** 2.0 ( 1.0, 2.9)
70% 1.8 ( 0.3, 3.3) 1.7 ( 0.4, 3.0) 0.75** 0.1 (1.0, 1.1)
80% 11.4 (12.8, 10.0) 10.2 (11.2, 9.2) 0.63** 1.2 (2.3, 0.1)
90% 16.5 (17.8, 15.2) 14.2 (14.9, 13.5) 0.54** 2.3 (3.4, 1.1)
100% 17.1 (18.4, 15.8) 14.1 (14.8, 13.4) 0.47** 3.0 (4.1, 1.8)
Hip (frontal)
Gait cycle Kinect System Vicon System r Difference
Mean 95% CI Mean 95% CI Mean 95% CI
0% 1.6 ( 0.9, 2.3) 1.0 ( 0.4, 1.6) 0.16 0.6 (0.3, 1.5)
10% 0.7 ( 0.1, 1.3) 0.3 (0.3, 0.9) 0.31* 0.4 (0.4, 1.1)
20% 0.4 (1.0, 0.2) 0.4 (1.0, 0.2) 0.58** 0.0 (0.5, 0.6)
30% 0.4 (1.2, 0.4) 1.2 (1.8, 0.6) 0.71** 0.8 ( 0.3, 1.4)
40% 0.1 (1.1, 0.9) 1.5 (2.1, 0.9) 0.68** 1.4 ( 0.6, 2.1)
50% 0.4 (0.7, 1.5) 0.6 (1.2, 0.0) 0.62** 1.0 ( 0.2, 1.9)
60% 1.2 ( 0.4, 2.0) 1.2 ( 0.6, 1.8) 0.63** 0.1 (0.7, 0.6)
70% 1.4 ( 0.7, 2.1) 2.3 ( 1.7, 2.9) 0.51** 1.0 (1.7, 0.3)
80% 1.1 ( 0.3, 1.9) 2.5 ( 1.8, 3.2) 0.31* 1.4 (2.3, 0.5)
90% 1.1 ( 0.3, 1.9) 2.3 ( 1.6, 3.0) 0.18 1.3 (2.2, 0.3)
100% 1.1 ( 0.3, 1.9) 2.1 ( 1.4, 2.8) 0.09 1.0 (2.0, 0.0)

CI, Confidential Interval; SD, Standard deviation.

*
p < 0.05.
**
p < 0.01.

Fig. 2. Trajectories of the knee joint angles in both the sagittal and frontal planes.
Average and 95% confidential interval of joint angle are demonstrated. The X axis
shows the gait cycle (%) and the Y axis shows the angle (degree).
Kinect v2 (<5°) is acceptable for gait analysis (McGinley et al.,
2009).
The angle shift in the hip joint in the frontal plane during the
gait cycle highlights a double peak; our Vicon data illustrates this
shift. Conversely, the angle shift curve made by our Kinect data
does not show a double peak. The Kinect system did not capture
the hip joint in the frontal plane as accurately as the Vicon system.
However, the 95% CIs for Kinect data significantly overlapped those
of Vicon data in the frontal plane than those in the sagittal plane
throughout the gait cycle. Moreover, the largest difference
between the Kinect and Vicon systems during the gait cycle was
<5°. The absolute value of the hip joint angle assessed using Kinect
in the frontal plane might be close to the actual value as assessed
by the Vicon system.
In the sagittal plane, the knee joint displays a double-knee
action during the gait cycle, which was observed in our Vicon
and Kinect data, although it was not distinctly represented in the
latter. We observed apparent differences between the two systems
during the first half of the stance phase, with a maximal difference
of 5.8°. These differences subsequently diminished to <5°, which is
the minimum acceptable error in the clinical setting (McGinley
et al., 2009). Thus, knee joint measures by the Kinect v2 in the
sagittal plane are acceptable, except for measures from the first
half of the stance phase.
In normal gait, from the terminal stance to midswing, the knee
joint angle in the frontal plane changes toward the abduction
direction (Lafortune et al., 1994), also observed in our Vicon data.
Although this change appears to start faster in the Kinect measures
than the Vicon measures, the largest difference between the sys-
tems during gait was 1.5°, which is smaller than the minimum
detectable change of the goniometer (Krause et al., 2015). The dif-
ference observed in our study is considered negligible in the clini-
cal setting (McGinley et al., 2009).
Mentiplay et al. (2015) have assessed Kinect v2 data validity
measured during gait. They used one sensor that was placed 2.5
or 3 m in front of subjects and measured knee and hip flexion dur-
ing gait. Similar to our study, the standard reference in Mentiplay
et al.’s (2015) study was the Vicon system; however, Mentiplay
284 R. Tanaka et al. / Journal of Biomechanics 71 (2018) 281–285

Table 2
Knee joint angles during the gait cycle in both the sagittal and frontal planes.

Knee (sagittal)
Gait cycle Kinect System Vicon System r Difference
Mean 95% CI Mean 95% CI Mean 95% CI
0% 11.0 (12.6, 9.4) 5.2 (6.2, 4.2) 0.54** 5.8 ( 4.5, 7.2)
10% 11.3 (12.6, 10.0) 7.0 (8.1, 5.9) 0.68** 4.3 ( 3.3, 5.3)
20% 7.4 (8.4, 6.4) 6.4 (7.5, 5.3) 0.71** 1.1 ( 0.2, 1.9)
30% 2.8 (3.6, 2.0) 3.6 (4.3, 2.9) 0.67** 0.8 (1.5, 0.2)
40% 0.9 (1.5, 0.3) 2.3 (2.8, 1.8) 0.62** 1.4 (1.9, 0.9)
50% 8.0 (9.3, 6.7) 8.2 (9.4, 7.0) 0.88** 0.2 (0.8, 0.4)
60% 26.9 (28.7, 25.1) 28.3 (30.4, 26.2) 0.83** 1.4 (2.6, 0.2)
70% 42.9 (44.2, 41.6) 46.3 (48.4, 44.2) 0.49** 3.4 (5.3, 1.6)
80% 38.5 (40.5, 36.5) 39.4 (41.9, 36.9) 0.53** 0.9 (3.7, 1.3)
90% 23.1 (25.8, 20.4) 21.2 (23.7, 18.7) 0.73** 1.9 (0.0, 3.9)
100% 10.2 (12.3, 8.1) 9.4 (10.8, 8.0) 0.74** 0.8 (0.6, 2.2)
Knee (frontal)
Gait cycle Kinect System Vicon System r Difference
Mean 95% CI Mean 95% CI Mean 95% CI
0% 4.2 (5.0, 3.4) 4.4 (5.1, 3.7) 0.86** 0.1 (0.6, 0.3)
10% 4.3 (5.1, 3.5) 4.4 (5.1, 3.7) 0.90** 0.1 (0.4, 0.3)
20% 4.4 (5.2, 3.6) 4.6 (5.3, 3.9) 0.88** 0.1 (0.5, 0.2)
30% 5.2 (6.0, 4.4) 4.6 (5.3, 3.9) 0.90** 0.6 ( 0.3, 1.0)
40% 6.0 (6.8, 5.2) 4.4 (5.1, 3.7) 0.90** 1.5 ( 1.2, 1.9)
50% 6.3 (7.0, 5.6) 4.8 (5.5, 4.1) 0.90** 1.5 ( 1.1, 1.8)
60% 6.6 (7.7, 5.5) 6.8 (7.9, 5.7) 0.74** 0.2 (1.00, 0.6)
70% 7.0 (8.9, 5.1) 8.5 (9.9, 7.1) 0.58** 1.4 (3.1, 0.2)
80% 6.5 (8.2, 4.8) 7.4 (8.6, 6.2) 0.52** 0.9 (2.4, 0.6)
90% 5.1 (6.1, 4.1) 5.7 (6.6, 4.8) 0.69** 0.6 (1.3, 0.2)
100% 4.0 (4.7, 3.3) 4.7 (5.5, 3.9) 0.85** 0.6 (1.0, 0.2)

CI, Confidential Interval; SD, Standard deviation.

**
p < 0.01.

et al. (2015) reported poor kinematic validity results. Our results
revealed moderate kinematic validity results; this inconsistency
could be attributed to correction application. Based on the substan-
tial error in magnitudes reported between Kinect and motion
tracking system by Xu et al. (2015), we performed parallel transla-
tion on the individual Kinect v2 gait data to minimize the differ-
ences between the Kinect and Vicon system data before our
analysis. Because this correction brings the Kinect v2 measures
close to the acceptable level in the clinical setting, we believe that
it is necessary to apply a correction to Kinect v2 data before kine-
matic gait analysis.
Eltoukhy et al. (2017a; 2017b) assessed Kinect v2 data validity
measured during gait and confirmed that the Kinect v2 is an
acceptable tool for the assessment of sagittal plane knee and hip
ROM and joint angles throughout the gait cycle. However, Kinect
v2 time series kinematic data during gait have not been validated.
Our study compared time series data of the hip and knee joint
angles during gait between Kinect v2 and Vicon systems. The
results clarified the deviation from the actual value observed in
the Kinect v2 sensor over one gait cycle.
5. Conclusion
Drafting the manuscript: Dr. Tanaka; revising the manuscript
critically for important intellectual content: Dr. Tanaka, Ms. Taki-
moto, Dr. Yamasaki, and Mr. Higashi.
Approval of the version of the manuscript to be published: Dr.
Tanaka, Ms. Takimoto H, Dr. Yamasaki, and Mr. Higashi.
Conflict of interest statement
This study was supported by a grant from Systemfriend Inc.
(Hiroshima, Japan).
Acknowledgments
The authors thank Mr. Toshihide Okamoto, Ms. Noriko Naka-
shima, Mr. Hiromichi Kawanishi, and Mr. Alejandro Diez of Sys-
temfriend Inc. for their technical help. The authors thank Mr.
Hiroyuki Tamura, Mr. Daisuke Kuwahara, Mr. Kosuke Miyazaki,
Ms. Yuki Ishikawa, Mr. Takuya Kubota, Mr. Haruka Shiraishi, Mr.
Tomoki Takeda, Mr. Yoshiki Ishii, and Mr. Shoichi Saito of the
Department of Rehabilitation, and faculty of the Department of
Rehabilitation, Hiroshima International University for their techni-
cal help.

Kinect v2 time series kinematic data obtained during gait in a

flatland are validated by the Vicon system if appropriate correc-
tions are performed in Kinect data. Future studies are warranted
to examine the reproducibility and systematic bias of the Kinect
v2.
Authorship contributions
Conception and design of study: Dr. Tanaka; acquisition of data:
Dr. Tanaka, Ms. Takimoto, Dr. Yamasaki, and Mr. Higashi; analysis
and/or interpretation of data: Dr. Tanaka and Ms. Takimoto.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jbiomech.2018.01.
035.
References
Andriacchi, T.P., Galante, J.O., Fermier, R.W., 1982. The influence of total knee-
replacement design on walking and stair-climbing. J. Bone Joint Surgery 64,
1328–1335.
 R. Tanaka et al. / Journal of Biomechanics 71 (2018) 281–285
Chiou, I.I.L., Burnett, C.N., 1985. Values of activities of daily living. A survey of stroke
patients and their home therapists. Phys. Ther. 65, 901–906.
Eltoukhy, M., Kuenze, C., Oh, J., Jacopetti, M., Wooten, S., Signorile, J., 2017a.
Microsoft Kinect can distinguish differences in over-ground gait between older
persons with and without Parkinson’s disease. Med. Eng. Phys. 44, 1–7.
Eltoukhy, M., Oh, J., Kuenze, C., Signorile, J., 2017b. Improved kinect-based
spatiotemporal and kinematic treadmill gait assessment. Gait Posture 51, 77–
83.
Kito, N., Shinkoda, K., Yamasaki, T., Kanemura, N., Anan, M., Okanishi, N., Ozawa, J.,
Moriyama, H., 2010. Contribution of knee adduction moment impulse to pain
and disability in Japanese women with medial knee osteoarthritis. Clin.
Biomech. 25, 914–919.
Koyama, Y., Tateuchi, H., Nishimura, R., Ji, X., Umegaki, H., Kobayashi, M., Ichihashi,
N., 2015. Relationships between performance and kinematic/kinetic variables of
stair descent in patients with medial knee osteoarthritis: An evaluation of
dynamic stability using an extrapolated center of mass. Clin. Biomech. 30,
1066–1070.
Krause, D.A., Boyd, M.S., Hager, A.N., Smoyer, E.C., Thompson, A.T., Hollman, J.H.,
2015. Reliability and accuracy of a goniometer mobile device application for
video measurement of the functional movement screen deep squat test. Int. J.
Sports Phys. Therapy 10, 37–44.
285
Kurabayashi, J., Mochimaru, M., Kouchi, M., 2003. Validation of the estimation
methods for the hip joint center. J. Soc. Biomech. (Japan) 27, 29–36.
Lafortune, M.A., Cavanagh, P.R., Sommer 3rd, H.J., Kalenak, A., 1994. Foot inversion-
eversion and knee kinematics during walking. J. Orthop. Res. 12, 412–420.
McGinley, J.L., Baker, R., Wolfe, R., Morris, M.E., 2009. The reliability of three-
dimensional kinematic gait measurements: a systematic review. Gait Posture
29, 360–369.
Mentiplay, B.F., Perraton, L.G., Bower, K.J., Pua, Y.H., McGaw, R., Heywood, S., Clark,
R.A., 2015. Gait assessment using the Microsoft Xbox One Kinect: Concurrent
validity and inter-day reliability of spatiotemporal and kinematic variables. J.
Biomech. 48, 2166–2170.
Perry, J., 1974. Kinesiology of lower extremity bracing. Clin. Orthop. Relat. Res. 102,
18–31.
Pfister, A., West, A.M., Bronner, S., Noah, J.A., 2014. Comparative abilities of
Microsoft Kinect and Vicon 3D motion capture for gait analysis. J. Med. Eng.
Technol. 38, 274–280.
Tateuchi, H., Koyama, Y., Akiyama, H., Goto, K., So, K., Kuroda, Y., Ichihashi, N., 2017.
Daily cumulative hip moment is associated with radiographic progression of
secondary hip osteoarthritis. Osteoarthritis Cartilage 25, 1291–1298.
Xu, X., McGorry, R.W., Chou, L.S., Lin, J.H., Chang, C.C., 2015. Accuracy of the
Microsoft Kinect for measuring gait parameters during treadmill walking. Gait
Posture 42, 145–151.