G Model
GAIPOS-4063; No. of Pages 5 Gait & Posture xxx (2013) xxx–xxx
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Gait phase varies over velocities Yancheng Liu a,b, Kun Lu a, Songhua Yan a, Ming Sun c, D. Kevin Lester d, Kuan Zhang a,* a
School of Biomedical Engineering, Capital Medical University, Beijing, China Department of Spinal Surgery, Tianjin Hospital, Tianjin, China c Minisun LLC, Fresno, CA, USA d Orthopedic Private Practice, Fresno, CA, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 8 February 2013 Received in revised form 4 July 2013 Accepted 12 October 2013
We sought to characterize the percent (PT) of the phases of a gait cycle (GC) as velocity changes to establish norms for pathological gait characteristics with higher resolution technology. Ninety five healthy subjects (49 males and 46 females with age 34.9 11.8 yrs, body weight 64.0 11.7 kg and BMI 23.5 3.6) were enrolled and walked comfortably on a 10-m walkway at self-selected slower, normal, and faster velocities. Walking was recorded with a high speed camera (250 frames per second) and the eight phases of a GC were determined by examination of individual frames for each subject. The correlation coefficients between the mean PT of the phases of the three velocities gaits and PT defined by previous publications were all greater than 0.99. The correlation coefficient between velocity and PT of gait phases is 0.83 for loading response (LR), 0.75 for mid stance (MSt), and 0.84 for pre-swing (PSw). While the PT of the phases of three velocities from this study are highly correlated with PT described by Dr. Jacquenlin Perry decades ago, actual PT of each phase varied amongst these individuals with the largest coefficient variation of 24.31% for IC with slower velocity. From slower to faster walk, the mean PT of MSt diminished from 35.30% to 25.33%. High resolution recording revealed ambiguity of some gait phase definitions, and these data may benefit GC characterization of normal and pathological gait in clinical practice. The study results indicate that one should consider individual variations and walking velocity when evaluating gaits of subjects using standard gait phase classification. ß 2013 Elsevier B.V. All rights reserved.
Keywords: Gait cycle Gait phase Gait velocity Variation
1. Introduction As a key parameter for gait analysis, a gait cycle (GC), defined as the time from heel strike to the ipsilateral heel strike, is widely used for the evaluation of basic and clinical disorders [1–5]. The gait cycle (GC) is measured from heel strike to heel strike of the same leg. There are two major phases: stance and swing. The subphases of stance are initial contact (IC), loading response (LR), mid stance (MSt), terminal stance (TSt) and, pre-swing (PSw). The subphases of swing are initial swing (ISw), mid swing (MSw), and terminal swing (TSw) [6]. Precise measures of the sub-phases may be useful to characterize normal gait with variable velocities to improve clinical evaluation of altered ambulation evaluation. The sub-phases during walking occur at IC 2%, LR 12%, MSt 31%, Tst 50%, PSw 62%, ISw 75%, MSw 87%, and TSw 100% of the GC. These measures were derived from gait results of more than 400 subjects (61 subjects with age 6–12 yrs, 53 subjects with age 13– 19 years old, 236 subjects with age 20–69 yrs and 70 subjects with
age with 70+ yrs) [6,7], and however, variations of duration of gait sub-phases with different genders, ages, BMI, and velocities are not well defined. Murray et al. [8] reported the high correlation coefficients between the duration of GC and the durations of phase components (0.96 for stance, 0.75 for swing and 0.79 for double-limb stance) as speeds change. By using photography at a rate of 20 exposures per second with 7 subjects, another study [9] showed that as speeds decreased, the durations of heel rise and toe off were increased. Those studies have demonstrated the strong influences of the velocities on the durations of sub-phases. The objective of this study was to characterize the sub-phase changes of the CG as a function of velocity in normal people by using high time resolution videography. This study was intended to enhance the usefulness of objective evaluation for orthopedic intervention techniques and characterize the GC changes associated with age, gender, BMI and variable velocities. 2. Methods
* Corresponding author at: The School of Biomedical Engineering, Capital Medical University, No. 10 Xitoutiao, You An Men Wai, Beijing 100069, China. Tel.: +86 10 83911806: fax: +86 10 83911560. E-mail addresses: [email protected], [email protected] (K. Zhang).
2.1. Subjects Subjects were recruited through the university and neighborhood community. The subjects received a small monetary
0966-6362/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gaitpost.2013.10.009
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 2
Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
Table 1 Subject characteristics [mean SD (range)].
Number of subjects Age (yr) Height (m) Body weight (kg) BMI (kg/m2)
Total
Female
Male
95 34.9 11.8 (20–67) 1.65 0.09 (1.47–1.84) 64.0 11.7 (42.4–103.5) 23.5 3.6 (17.2–35.8)
46 36.27 10.62 1.58 0.06 59.34 10.93 23.72 3.90
49 33.86 12.66 1.71 0.06 68.06 10.84 23.26 3.32
compensation for their participation. Ninety-five subjects were enrolled in the study (Table 1). Based on the cutoff of BMI for Chinese population [10], there are 54 normal subjects with BMI< 24 (21.00 1.89), 41 overweight/obese subjects with BMI 24 (26.74 2.50). All subjects were free of any impairment of the locomotors system. Subjects wore comfortable tight clothes to facilitate viewing the lower extremities. The study was approved by the Institutional Review Board at Capital Medical University. All subjects signed written consent to participate. 2.2. Devices A High-Speed Video Camera (FASTCAM-ultima 1024, Japan) with a 16 mm lens perpendicular to the walkway was used to record walking at a rate of 250 frames per second (fps). The distance from the camera to the walkway was 4.2 m The height of camera lens was 50 cm. The camera was activated by a manual trigger to capture at least one full GC. The trigger is a small device and researchers can hold it and tap the button to activate the camera when subject walked into the camera view. The sequence of measure was normal, slower and then faster velocities with a 30 s rest period between examinations. 2.3. Experimental protocols In order to eliminate the effect of acceleration and deceleration at the beginning and end of the examination, the 6-m middle part of a 10-m walkway was designated for data collection. The walkway was covered with wear resistant paper marked by lines at intervals of 2 cm. A standard 1 m scale was placed along the walkway as the reference. Before the test, each subject walked barefoot on the walkway at a comfortable velocity to familiarize themselves with the laboratory environment. Totally, three trials with three velocities were recorded and each trial contained one full gait cycle at least. First, the subjects walked at their self-selected comfortable velocity (normal velocity) on the walkway, and one researcher activated high speed camera with a trigger when the subject step into the camera view. After a short break for saving video, with the same protocols, the subject was asked to walk freely at a slower velocity than normal, and then at a faster velocity than normal back and forth for gait recording. 3. Data analysis Phtron FASTCAM Viewer software was used to analyze the video images. All video images were analyzed frame by frame by two research staff to divide all gait cycles into eight sub-phases independently. Then two researchers examined their results together until agreed. It took two months to analyze all video images for sub-phases analysis. Intraclass correlation coefficient (ICC) was used to evaluate reliability of measurements of PT of gait sub-phases. The definitions of eight sub-phases were based on the classical gait analysis methods of Perry [6,7]. IC (0–2%): the foot contacts the ground and a subtle heel soft tissue deformation can be seen;
LR (2–12%): a very rapid ankle plantar flexion, until the frame when the opposite limb has preswing; MSt (12–31%): begins as the opposite foot enters preswing; TSt (31–50%): begins with heel rise and the opposite foot heel strike; PSw (50–62%): begins with IC of the opposite limb and ends with ipsilateral; ISw (62–75%): begins as the foot is lifted from the floor and ends when the swing foot is opposite the stance foot; MSw (75–87%): begins as the swing forefoot is opposite the stance limb and ends when the tibia is vertical with the foot parallel to the floor; TSw (87–100%): begins with a vertical tibia and ends when the foot strikes the floor again. The durations of sub-phases were calculated by (the number of frames of sub-phases) 4 ms. The cumulative PT of each subphase was calculated by [(the number of frames from the ending frame of the sub-phase to first frame of IC)/the total number of frames of the GC)] 100%. The walkway was covered with wear resistant paper marked by lines at intervals of 2 cm. Average of stride length in this study was 100 cm, so error could be controlled within 2%. A standard 1 m scale plate was placed along the walkway as the reference. The stride length is calculated by using the ShiXun motion analysis software (Beijing Sport University, Beijing, China) based on pixel ratio of the scale plate. Velocity was calculated as stride length/GC duration. If the difference between stride lengths measured by two researchers is more than 1 cm, then two researchers had to measure the stride length again until the difference is less than 1 cm. Data were analyzed by SPSS 17.0. The distribution and homogeneity of variance of age, height, BMI, PT of gait sub-phases, and other kinematic parameters were calculated. One way ANOVA was performed to compare PT of each sub-phase and the kinematic parameters between different velocities. Correlations between PT of gait sub-phases and age, height, body weight, BMI, gait velocities were performed with Pearson’s test respectively. Significance is defined as p < 0.05. 4. Results Two hundred eighty five cycles from 95 subjects were analyzed. Mean percents of ending periods of the eight sub-phases at normal, slower and faster velocities are given in Table 2. IC initiated at 0% of GC and TSw ended at 100% of GC. Correlation coefficients between the mean PT at normal, faster and slower gait velocities and the PT defined by Perry and Burnfield [6] are 0.9978, 0.9983 and 0.9991 respectively (Fig. 1). The mean PT of IC (1.78, 1.81 and 1.89%) for normal, faster and slower velocities are smaller than the result (2%) from Perry and Burnfield [6]. Overall, the mean PT of eight subphases at the normal velocity is closer to Perry’s results compared with faster and slower velocities. Intraclass correlation coefficients between measurements of PT of eight sub-phases by two researchers are 0.953, 0971, 0.945, 0. 983, 0.978, 0.986, 0.956 and 1.00 respectively. PT of the stance sub-phases changes largely with velocities. With normal velocity, the mean PT of IC is 1.81% with the range of 1.08–2.59%, 12.89% (8.33–15.73%) for LR and 31.56% (19.05– 40.47%) for MSt; with faster velocity, the mean PT of IC is 1.78% with the range of 0.85–2.55%, 10.32% (5.39–13.77%) for LR and 25.39% (15.48–34.78%) for MSt; with slower velocity the mean PT of IC for is 1.89% with the range of 0.67–2.95%, 14.55% (10.73– 19.29%) for LR and 35.30% (18.75–44.48%) for MSt. When the gait velocity is increased from slower to normal, then to faster, the duration of TSt is increased from 14.94% to 18.67%, then to 24.81% of a GC, the duration of MSt is decreased from 20.75% to 18.66%, then to 15.01% of a GC, and the duration of LR is decreased from 12.66% to 11.08%, then to 8.54% of a GC. Correlation coefficients between velocities and PT of gait phases are given in Table 3. PT of LR, MSt and PSw are highly correlated
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
3
Table 2 PT of gait sub-phases and comparisons with Perry’s results (%) [6,7]. Gait velocity/PT
IC (2)
LR (12)
MSt (31)
TSt (50)
PSw (62)
ISw (75)
MSw (87)
TSw (100)
Normal velocity Mean SD CV Max Min Mean/Perry
1.81 0.32 17.42 2.59 1.08 90.55
12.89 1.41 10.93 15.73 8.33 107.45
31.56 4.35 13.78 40.47 19.05 101.80
50.23 1.06 2.12 53.81 47.69 100.46
62.87 1.53 2.44 67.86 59.41 101.40
75.75 1.16 1.53 78.38 71.53 101.00
88.51 1.69 1.90 92.68 83.77 101.73
100 0 100 100 1
Fast velocity Mean SD CV Max Min Mean/Perry
1.78 0.37 20.67 2.55 0.85 88.94
10.32 1.67 16.13 13.77 5.39 86.02
25.33 4.64 18.32 34.78 15.48 81.72
50.14 0.99 1.98 53.18 47.52 100.29
60.21 1.77 2.95 63.91 53.85 97.11
75.04 1.10 1.46 78.18 71.89 100.06
87.88 1.63 1.86 91.44 83.27 101.01
100 0 100 100 1
Slow velocity Mean SD CV Max Min Mean/Perry
1.89 0.46 24.31 2.95 0.67 94.51
14.55 1.65 11.32 19.29 10.73 121.27
35.30 4.68 13.25 44.48 18.75 113.87
50.24 1.19 2.37 54.36 47.54 100.49
64.69 1.79 2.77 69.06 59.19 104.34
76.11 1.20 1.58 79.62 73.02 101.48
88.72 1.74 1.96 93.08 83.49 101.98
100 0 100 100 1
PT – percent; CV – coefficient of variation; Mean/Perry – mean PT spent in gait phases divided by Perry’s results.
100 Perry
90
Normal
Fast
Slow
80
Percentage
70 60 50 40 30 20 10 0 IC
LR
MSt
TSt
PSw
Isw
MSw
TSw
Gait phases Fig. 1. Comparison of the percent of the sub-phases of a gait cycle between the results from this study with three velocities and the results from Dr. Jacquenlin Perry [6,7].
with velocities. The largest variation of coefficient is 24.31% for IC with slower velocity, and the smallest variation of coefficient is 1.46% for ISw with faster velocity. PT of gait sub-phases is not highly correlated with age, height, and BMI (all correlation coefficients are smaller than 0.2) for three gait velocities. PT of gait sub-phases is also not highly correlated with velocities for both normal and overweight/obese subjects (all correlation coefficients are smaller than 0.26 with slower velocities, smaller than 0.20 with normal velocities and smaller than 0.26 with slower velocities). Kinematic parameters such as velocity, stride length, cadence, a GC duration, PT of stance phase, initial DLS, SLS, terminal DLS, and
percentage of swing phase at different velocities are significantly different between pairwise combination testing (p < 0.001) (Table 4). As velocity is increased, GC duration, percentage of initial DLS, percentage of terminal DLS and percentage of stance phase are decreased while percentage of swing phase, percentage of SLS and stride length increased. Except GC duration (p = 0.622) and cadence (p = 0.767), other parameters are significantly different between different genders, with p = 0.003 for velocity, p < 0.001 for stride length, p = 0.002 for percentage of stance, p = 0.003 for percentage of initial DLS, p = 0.041 for percentage of SLS, p = 0.015 for percentage of terminal DLS, p = 0.002 for percentage of swing phase.
Table 3 Correlation coefficients between velocities and PT spent in gait phases. IC Overall velocities Normal velocities Fast velocities Slow velocities
0.06 0.07 0.08 0.04
LR 0.83 0.52 0.67 0.53
MSt 0.75 0.33 0.48 0.59
TSt 0.02 0.02 0.01 0.07
PSw 0.84 0.56 0.67 0.56
ISw 0.36 0.06 0.22 0.00
MSw 0.22 0.06 0.28 0.13
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
4 Table 4 Kinematic parameters at different velocities (Mean SD). Variables
Normal
Faster
Slower
Female (all velocities)
Male (all velocities)
p value
Velocity (m/s) Stride length (cm) Cadence (steps/s) Gait cycle duration (ms) Stance phase (%) Initial DLS (%) SLS (%) Terminal DLS (%) Swing phase (%)
1.17 0.14 123.33 11.29 113.60 8.58 1062.02 76.38 62.87 1.53 12.89 1.41 37.34 17.16 12.64 1.42 37.13 1.53
1.55 0.20 139.86 14.09 132.74 12.37 911.28 79.65 60.21 1.77 10.32 1.67 39.82 1.84 10.06 1.74 39.79 1.77
0.94 0.15 110.48 11.24 101.53 9.73 1193.14 119.66 64.69 1.79 14.55 1.65 35.70 2.08 14.45 1.79 35.31 1.79
1.16 0.28 118.64 15.46 116.45 17.31 1052.78 155.11 63.05 2.42 13.01 2.26 37.30 2.46 12.74 2.28 36.95 2.42
1.27 0.31 130.11 16.89 115.49 15.74 1058.01 142.68 62.15 2.52 12.20 2.37 37.91 2.58 12.04 2.56 37.85 2.52
.00* .00* .77 .62 .00* .00* .04* .02* .00*
DLS – double limb stance; SLS – single limb stance; p value – difference between female and males with all velocities. * Significant at p < .05.
5. Discussion The GC phases at the different velocity measures were highly correlated with other standards [6,7] (Fig. 1) while there were small differences for IC and MSw phases. It may be because that we used high speed camera and previous studies were derived from gait analyzer and footswitch. Second, ambiguity of the phase definition could also be responsible for variation. Heel rise determined the end of MSt and the start of TSt, however it was difficult to determine the exact moment of heel rise, as the soft tissue of the heel may sag below the actual bony heel rise on video. Instrumented devices like footswitches are commonly used to detect heel rise and other gait events [11,12], but they have limited accuracy depending on the measuring device itself and on how they attach to the body. Consequently, the end of MSw and the start of TSw were defined under two conditions: tibia was vertical and the opposite foot was parallel to the ground [6,7]. However occasionally it is difficult to determine the moment when the tibia was vertical because of its shape. Often the two conditions could not be satisfied in the same video frame. When conditions ‘Tibia is vertical and the foot is parallel to the ground’ for the end of MSw phase and the start of TSw could not been seen at the same frame, then the condition ‘the foot is parallel to the ground’ was considered as a key condition for the corresponding phase. ICC analysis has demonstrated that the measurement of PT of gait subphase by the camera is reliable. Toe off was used to identify the end of LR and PSw, and begins as the foot was lifted for floor clearance. There may be three toe off patterns: first and the most common pattern is that great toe rotates about 60–908 with the great toe on the ground (toe rocker); second, the great toe is rotating about 30–608 without a obvious toe rocker; third, the great toe sometimes did not rotate till the toe was lifted to initiate swing with little or no toe rocker. We defined the video frame as toe off when the great toe starts moving forward horizontally; however, similar to IC and heel rise, it was difficult to identify the contact frame because of the soft tissue of the feet and the optical center of the camera could not absolutely opposite to the lowest gap between foot and ground. To eliminate the visual illusion of early contact, we defined the frame when a subtle heel soft tissue deformation can be seen as the start of IC. Another possible reason may be ethnicity difference: all subjects are Chinese in ethnicity that could have different gait phase patterns compared with Caucasians and perhaps many other ethnic groups. These differentiations may seem arbitrary or too obscure to be of any possible clinical value. However, in recent clinical studies mere milliseconds of difference in GC determinations have lead to important clinical conclusions [4]. In fact, for valuable interpretation of GC data, it appears that high resolution data will be necessary to determine real clinical normalcy after intervention or surgeries such as total knee replacement. In general, the PT spent in each phase from our study matches well with the previous standard, especially for normal velocity
gaits. However, the PT spent in each phase varies from person to person with the largest coefficient variation of 24.31%. The range of PT is much larger than the mean PT spent in IC phase, up to 120% [(Max Min)/Mean]. Our data has also shown that the PT of LR, MSt and PSw are significantly negatively correlated ( 0.83 for LR, 0.75 for MSt, 0.84 for PSw) with gait velocities, which indicates that velocities strongly influences PT of individual GC phases, especially during stance sub-phases. These data are consistent with those reported previously [9,13]. When velocity is increased from slower to faster velocities, PT of MSt is decreased almost 10% of GC, from 35.30% to 25.33%, and from 14.55% and 64.69% to 10.32% and 60.21% for LR and PSw respectively. Mean/Perry (mean PT spent in gait phases divided by Perry’s results) was calculated for the direct comparison. The results of these data have shown that Mean/Perry for LR are 121.27%, 107.45% and 86.02% for slower, normal and faster velocities, and Mean/Perry for MSt are 113.87%, 101.80% and 81.72% for slower, normal and faster velocities. In future clinical studies, patients could improve velocities, cadence, stride length and distance etc. with the changes of PT of sub-phases, so evidence from this study indicates that velocity and individual variation should be taken into consideration for better evaluation of gait function when using eight sub-phases classification from Perry [6,7]. As velocities are increased, GC durations, percentages of Initial DLS, percentages of terminal DLS, and percentages of stance phase are decreased, while percentage of swing phase, percentage of SLS and stride length increased, which matches with results from other studies [8,9,14], which indicates that heel rise and toe off are earlier with fast velocity. It could be the result from fast moving of the center of mass and center of pressure [13,15]. These observations are likely to result in adjustments of interpretations in patients with altered gait after knee or hip surgery. There are certain limitations in the study. First, all subjects are Chinese and it is not clear if ethnicity can influence the results; second, the mean age was fairly young and may influence the results and finally, although there was a significant difference between slower and normal velocities, the absolute difference between the slower and normal velocities was smaller than the difference between the faster and normal velocities. Consequently a larger sample size study may be necessary to reveal significant differences. Future study may focus on joints angle deviations and electromyographic changes at different velocities of the eight subphases of the GC. Acknowledgements This study was partially supported by National Natural Science Foundation of China (Grant 31170900), Beijing Natural Science Foundation Program and Scientific Research Key Program of Beijing Municipal Commission of Education (Grant KZ201310025010) and Research Fund for the Doctoral Program of Higher Education of China (Grant 20121107110018).
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
Conflict of interest The authors have no financial and personal relationships with other people or organisations that could inappropriately influence (bias) their work. References [1] Powers CM, Rao S, Perry J. Knee kinetics in trans-tibial amputee gait. Gait Posture 1998;8(1):1–7. [2] Yan S, Zhang K, Tan G, Yang J., Liu Z. Effects of obesity on dynamic plantar pressure distribution in Chinese prepubescent children during walking. Gait Posture 2013;37(1):37–42. [3] Aminian K, Trevisan C, Najafi B, Dejnabadi H, Frigo C, Pavan E, et al. Evaluation of an ambulatory system for gait analysis in hip osteoarthritis and after total hip replacement. Gait Posture 2004;20(1):102–7. [4] Lester DK, Zhang K. Gait analysis of knee arthritis treated with hyaluronic acid. J Arthroplasty 2010;25(8):1290–4. [5] Chevalier TL, Chockalingam N. Effects of foot orthoses: how important is the practitioner? Gait Posture 2012;35(3):383–8. [6] Perry J, Burnfield JM. Gait analysis: normal and pathological function. 2nd ed. SLACK, Inc.: Thorofare, NJ; 2010.
5
[7] Perry J. Gait analysis: normal pathological function. SLACK, Inc.: Thorofare, NJ; 1992. [8] Murray MP, Kory RC, Clarkson BH, Sepic SB. Comparison of free and fast speed walking patterns of normal men. Am J Phys Med 1966;45(1):8–23. [9] Murray MP, Mollinger LA, Gardner GM, Sepic SB. Kinematic and EMG patterns during slow, free, and fast walking. J Orthop Res 1984;2(3): 272–80. [10] Chinese Medical Association Subsection of Cardiovascular Disease, Chinese Journal of Cardiology Editorial Board. Highlights of the Second National Conference on Dyslipidemia. Chin J Cardiol 2002;30:643–6. [11] Perry J, Burnfield JM, Gronley JK, Mulroy SJ. Toe walking: muscular demands at the ankle and knee. Arch Phys Med Rehabil 2003;84(1):7–16. [12] Abu-Faraj ZO, Harris GF, Abler JH, Wertsch JJ. A Holter-type, microprocessorbased, rehabilitation instrument for acquisition and storage of plantar pressure data. J Rehabil Res Dev 1997;34(2):187–94. [13] Chiu MC, Wu HC, Chang LY. Gait speed and gender effects on center of pressure progression during normal walking. Gait Posture 2013;37(1):43–8. [14] Jordan K, Challis JH, Newell KM. Walking speed influences on gait cycle variability. Gait Posture 2007;26(1):128–34. [15] Orendurff MS, Segal AD, Klute GK, Berge JS, Rohr ES, Kadel NJ. The effect of walking speed on center of mass displacement. J Rehabil Res Dev 2004; 41(6A):829–34.
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
GAIPOS-4063; No. of Pages 5 Gait & Posture xxx (2013) xxx–xxx
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Gait phase varies over velocities Yancheng Liu a,b, Kun Lu a, Songhua Yan a, Ming Sun c, D. Kevin Lester d, Kuan Zhang a,* a
School of Biomedical Engineering, Capital Medical University, Beijing, China Department of Spinal Surgery, Tianjin Hospital, Tianjin, China c Minisun LLC, Fresno, CA, USA d Orthopedic Private Practice, Fresno, CA, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 8 February 2013 Received in revised form 4 July 2013 Accepted 12 October 2013
We sought to characterize the percent (PT) of the phases of a gait cycle (GC) as velocity changes to establish norms for pathological gait characteristics with higher resolution technology. Ninety five healthy subjects (49 males and 46 females with age 34.9 11.8 yrs, body weight 64.0 11.7 kg and BMI 23.5 3.6) were enrolled and walked comfortably on a 10-m walkway at self-selected slower, normal, and faster velocities. Walking was recorded with a high speed camera (250 frames per second) and the eight phases of a GC were determined by examination of individual frames for each subject. The correlation coefficients between the mean PT of the phases of the three velocities gaits and PT defined by previous publications were all greater than 0.99. The correlation coefficient between velocity and PT of gait phases is 0.83 for loading response (LR), 0.75 for mid stance (MSt), and 0.84 for pre-swing (PSw). While the PT of the phases of three velocities from this study are highly correlated with PT described by Dr. Jacquenlin Perry decades ago, actual PT of each phase varied amongst these individuals with the largest coefficient variation of 24.31% for IC with slower velocity. From slower to faster walk, the mean PT of MSt diminished from 35.30% to 25.33%. High resolution recording revealed ambiguity of some gait phase definitions, and these data may benefit GC characterization of normal and pathological gait in clinical practice. The study results indicate that one should consider individual variations and walking velocity when evaluating gaits of subjects using standard gait phase classification. ß 2013 Elsevier B.V. All rights reserved.
Keywords: Gait cycle Gait phase Gait velocity Variation
1. Introduction As a key parameter for gait analysis, a gait cycle (GC), defined as the time from heel strike to the ipsilateral heel strike, is widely used for the evaluation of basic and clinical disorders [1–5]. The gait cycle (GC) is measured from heel strike to heel strike of the same leg. There are two major phases: stance and swing. The subphases of stance are initial contact (IC), loading response (LR), mid stance (MSt), terminal stance (TSt) and, pre-swing (PSw). The subphases of swing are initial swing (ISw), mid swing (MSw), and terminal swing (TSw) [6]. Precise measures of the sub-phases may be useful to characterize normal gait with variable velocities to improve clinical evaluation of altered ambulation evaluation. The sub-phases during walking occur at IC 2%, LR 12%, MSt 31%, Tst 50%, PSw 62%, ISw 75%, MSw 87%, and TSw 100% of the GC. These measures were derived from gait results of more than 400 subjects (61 subjects with age 6–12 yrs, 53 subjects with age 13– 19 years old, 236 subjects with age 20–69 yrs and 70 subjects with
age with 70+ yrs) [6,7], and however, variations of duration of gait sub-phases with different genders, ages, BMI, and velocities are not well defined. Murray et al. [8] reported the high correlation coefficients between the duration of GC and the durations of phase components (0.96 for stance, 0.75 for swing and 0.79 for double-limb stance) as speeds change. By using photography at a rate of 20 exposures per second with 7 subjects, another study [9] showed that as speeds decreased, the durations of heel rise and toe off were increased. Those studies have demonstrated the strong influences of the velocities on the durations of sub-phases. The objective of this study was to characterize the sub-phase changes of the CG as a function of velocity in normal people by using high time resolution videography. This study was intended to enhance the usefulness of objective evaluation for orthopedic intervention techniques and characterize the GC changes associated with age, gender, BMI and variable velocities. 2. Methods
* Corresponding author at: The School of Biomedical Engineering, Capital Medical University, No. 10 Xitoutiao, You An Men Wai, Beijing 100069, China. Tel.: +86 10 83911806: fax: +86 10 83911560. E-mail addresses: [email protected], [email protected] (K. Zhang).
2.1. Subjects Subjects were recruited through the university and neighborhood community. The subjects received a small monetary
0966-6362/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gaitpost.2013.10.009
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 2
Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
Table 1 Subject characteristics [mean SD (range)].
Number of subjects Age (yr) Height (m) Body weight (kg) BMI (kg/m2)
Total
Female
Male
95 34.9 11.8 (20–67) 1.65 0.09 (1.47–1.84) 64.0 11.7 (42.4–103.5) 23.5 3.6 (17.2–35.8)
46 36.27 10.62 1.58 0.06 59.34 10.93 23.72 3.90
49 33.86 12.66 1.71 0.06 68.06 10.84 23.26 3.32
compensation for their participation. Ninety-five subjects were enrolled in the study (Table 1). Based on the cutoff of BMI for Chinese population [10], there are 54 normal subjects with BMI< 24 (21.00 1.89), 41 overweight/obese subjects with BMI 24 (26.74 2.50). All subjects were free of any impairment of the locomotors system. Subjects wore comfortable tight clothes to facilitate viewing the lower extremities. The study was approved by the Institutional Review Board at Capital Medical University. All subjects signed written consent to participate. 2.2. Devices A High-Speed Video Camera (FASTCAM-ultima 1024, Japan) with a 16 mm lens perpendicular to the walkway was used to record walking at a rate of 250 frames per second (fps). The distance from the camera to the walkway was 4.2 m The height of camera lens was 50 cm. The camera was activated by a manual trigger to capture at least one full GC. The trigger is a small device and researchers can hold it and tap the button to activate the camera when subject walked into the camera view. The sequence of measure was normal, slower and then faster velocities with a 30 s rest period between examinations. 2.3. Experimental protocols In order to eliminate the effect of acceleration and deceleration at the beginning and end of the examination, the 6-m middle part of a 10-m walkway was designated for data collection. The walkway was covered with wear resistant paper marked by lines at intervals of 2 cm. A standard 1 m scale was placed along the walkway as the reference. Before the test, each subject walked barefoot on the walkway at a comfortable velocity to familiarize themselves with the laboratory environment. Totally, three trials with three velocities were recorded and each trial contained one full gait cycle at least. First, the subjects walked at their self-selected comfortable velocity (normal velocity) on the walkway, and one researcher activated high speed camera with a trigger when the subject step into the camera view. After a short break for saving video, with the same protocols, the subject was asked to walk freely at a slower velocity than normal, and then at a faster velocity than normal back and forth for gait recording. 3. Data analysis Phtron FASTCAM Viewer software was used to analyze the video images. All video images were analyzed frame by frame by two research staff to divide all gait cycles into eight sub-phases independently. Then two researchers examined their results together until agreed. It took two months to analyze all video images for sub-phases analysis. Intraclass correlation coefficient (ICC) was used to evaluate reliability of measurements of PT of gait sub-phases. The definitions of eight sub-phases were based on the classical gait analysis methods of Perry [6,7]. IC (0–2%): the foot contacts the ground and a subtle heel soft tissue deformation can be seen;
LR (2–12%): a very rapid ankle plantar flexion, until the frame when the opposite limb has preswing; MSt (12–31%): begins as the opposite foot enters preswing; TSt (31–50%): begins with heel rise and the opposite foot heel strike; PSw (50–62%): begins with IC of the opposite limb and ends with ipsilateral; ISw (62–75%): begins as the foot is lifted from the floor and ends when the swing foot is opposite the stance foot; MSw (75–87%): begins as the swing forefoot is opposite the stance limb and ends when the tibia is vertical with the foot parallel to the floor; TSw (87–100%): begins with a vertical tibia and ends when the foot strikes the floor again. The durations of sub-phases were calculated by (the number of frames of sub-phases) 4 ms. The cumulative PT of each subphase was calculated by [(the number of frames from the ending frame of the sub-phase to first frame of IC)/the total number of frames of the GC)] 100%. The walkway was covered with wear resistant paper marked by lines at intervals of 2 cm. Average of stride length in this study was 100 cm, so error could be controlled within 2%. A standard 1 m scale plate was placed along the walkway as the reference. The stride length is calculated by using the ShiXun motion analysis software (Beijing Sport University, Beijing, China) based on pixel ratio of the scale plate. Velocity was calculated as stride length/GC duration. If the difference between stride lengths measured by two researchers is more than 1 cm, then two researchers had to measure the stride length again until the difference is less than 1 cm. Data were analyzed by SPSS 17.0. The distribution and homogeneity of variance of age, height, BMI, PT of gait sub-phases, and other kinematic parameters were calculated. One way ANOVA was performed to compare PT of each sub-phase and the kinematic parameters between different velocities. Correlations between PT of gait sub-phases and age, height, body weight, BMI, gait velocities were performed with Pearson’s test respectively. Significance is defined as p < 0.05. 4. Results Two hundred eighty five cycles from 95 subjects were analyzed. Mean percents of ending periods of the eight sub-phases at normal, slower and faster velocities are given in Table 2. IC initiated at 0% of GC and TSw ended at 100% of GC. Correlation coefficients between the mean PT at normal, faster and slower gait velocities and the PT defined by Perry and Burnfield [6] are 0.9978, 0.9983 and 0.9991 respectively (Fig. 1). The mean PT of IC (1.78, 1.81 and 1.89%) for normal, faster and slower velocities are smaller than the result (2%) from Perry and Burnfield [6]. Overall, the mean PT of eight subphases at the normal velocity is closer to Perry’s results compared with faster and slower velocities. Intraclass correlation coefficients between measurements of PT of eight sub-phases by two researchers are 0.953, 0971, 0.945, 0. 983, 0.978, 0.986, 0.956 and 1.00 respectively. PT of the stance sub-phases changes largely with velocities. With normal velocity, the mean PT of IC is 1.81% with the range of 1.08–2.59%, 12.89% (8.33–15.73%) for LR and 31.56% (19.05– 40.47%) for MSt; with faster velocity, the mean PT of IC is 1.78% with the range of 0.85–2.55%, 10.32% (5.39–13.77%) for LR and 25.39% (15.48–34.78%) for MSt; with slower velocity the mean PT of IC for is 1.89% with the range of 0.67–2.95%, 14.55% (10.73– 19.29%) for LR and 35.30% (18.75–44.48%) for MSt. When the gait velocity is increased from slower to normal, then to faster, the duration of TSt is increased from 14.94% to 18.67%, then to 24.81% of a GC, the duration of MSt is decreased from 20.75% to 18.66%, then to 15.01% of a GC, and the duration of LR is decreased from 12.66% to 11.08%, then to 8.54% of a GC. Correlation coefficients between velocities and PT of gait phases are given in Table 3. PT of LR, MSt and PSw are highly correlated
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
3
Table 2 PT of gait sub-phases and comparisons with Perry’s results (%) [6,7]. Gait velocity/PT
IC (2)
LR (12)
MSt (31)
TSt (50)
PSw (62)
ISw (75)
MSw (87)
TSw (100)
Normal velocity Mean SD CV Max Min Mean/Perry
1.81 0.32 17.42 2.59 1.08 90.55
12.89 1.41 10.93 15.73 8.33 107.45
31.56 4.35 13.78 40.47 19.05 101.80
50.23 1.06 2.12 53.81 47.69 100.46
62.87 1.53 2.44 67.86 59.41 101.40
75.75 1.16 1.53 78.38 71.53 101.00
88.51 1.69 1.90 92.68 83.77 101.73
100 0 100 100 1
Fast velocity Mean SD CV Max Min Mean/Perry
1.78 0.37 20.67 2.55 0.85 88.94
10.32 1.67 16.13 13.77 5.39 86.02
25.33 4.64 18.32 34.78 15.48 81.72
50.14 0.99 1.98 53.18 47.52 100.29
60.21 1.77 2.95 63.91 53.85 97.11
75.04 1.10 1.46 78.18 71.89 100.06
87.88 1.63 1.86 91.44 83.27 101.01
100 0 100 100 1
Slow velocity Mean SD CV Max Min Mean/Perry
1.89 0.46 24.31 2.95 0.67 94.51
14.55 1.65 11.32 19.29 10.73 121.27
35.30 4.68 13.25 44.48 18.75 113.87
50.24 1.19 2.37 54.36 47.54 100.49
64.69 1.79 2.77 69.06 59.19 104.34
76.11 1.20 1.58 79.62 73.02 101.48
88.72 1.74 1.96 93.08 83.49 101.98
100 0 100 100 1
PT – percent; CV – coefficient of variation; Mean/Perry – mean PT spent in gait phases divided by Perry’s results.
100 Perry
90
Normal
Fast
Slow
80
Percentage
70 60 50 40 30 20 10 0 IC
LR
MSt
TSt
PSw
Isw
MSw
TSw
Gait phases Fig. 1. Comparison of the percent of the sub-phases of a gait cycle between the results from this study with three velocities and the results from Dr. Jacquenlin Perry [6,7].
with velocities. The largest variation of coefficient is 24.31% for IC with slower velocity, and the smallest variation of coefficient is 1.46% for ISw with faster velocity. PT of gait sub-phases is not highly correlated with age, height, and BMI (all correlation coefficients are smaller than 0.2) for three gait velocities. PT of gait sub-phases is also not highly correlated with velocities for both normal and overweight/obese subjects (all correlation coefficients are smaller than 0.26 with slower velocities, smaller than 0.20 with normal velocities and smaller than 0.26 with slower velocities). Kinematic parameters such as velocity, stride length, cadence, a GC duration, PT of stance phase, initial DLS, SLS, terminal DLS, and
percentage of swing phase at different velocities are significantly different between pairwise combination testing (p < 0.001) (Table 4). As velocity is increased, GC duration, percentage of initial DLS, percentage of terminal DLS and percentage of stance phase are decreased while percentage of swing phase, percentage of SLS and stride length increased. Except GC duration (p = 0.622) and cadence (p = 0.767), other parameters are significantly different between different genders, with p = 0.003 for velocity, p < 0.001 for stride length, p = 0.002 for percentage of stance, p = 0.003 for percentage of initial DLS, p = 0.041 for percentage of SLS, p = 0.015 for percentage of terminal DLS, p = 0.002 for percentage of swing phase.
Table 3 Correlation coefficients between velocities and PT spent in gait phases. IC Overall velocities Normal velocities Fast velocities Slow velocities
0.06 0.07 0.08 0.04
LR 0.83 0.52 0.67 0.53
MSt 0.75 0.33 0.48 0.59
TSt 0.02 0.02 0.01 0.07
PSw 0.84 0.56 0.67 0.56
ISw 0.36 0.06 0.22 0.00
MSw 0.22 0.06 0.28 0.13
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
4 Table 4 Kinematic parameters at different velocities (Mean SD). Variables
Normal
Faster
Slower
Female (all velocities)
Male (all velocities)
p value
Velocity (m/s) Stride length (cm) Cadence (steps/s) Gait cycle duration (ms) Stance phase (%) Initial DLS (%) SLS (%) Terminal DLS (%) Swing phase (%)
1.17 0.14 123.33 11.29 113.60 8.58 1062.02 76.38 62.87 1.53 12.89 1.41 37.34 17.16 12.64 1.42 37.13 1.53
1.55 0.20 139.86 14.09 132.74 12.37 911.28 79.65 60.21 1.77 10.32 1.67 39.82 1.84 10.06 1.74 39.79 1.77
0.94 0.15 110.48 11.24 101.53 9.73 1193.14 119.66 64.69 1.79 14.55 1.65 35.70 2.08 14.45 1.79 35.31 1.79
1.16 0.28 118.64 15.46 116.45 17.31 1052.78 155.11 63.05 2.42 13.01 2.26 37.30 2.46 12.74 2.28 36.95 2.42
1.27 0.31 130.11 16.89 115.49 15.74 1058.01 142.68 62.15 2.52 12.20 2.37 37.91 2.58 12.04 2.56 37.85 2.52
.00* .00* .77 .62 .00* .00* .04* .02* .00*
DLS – double limb stance; SLS – single limb stance; p value – difference between female and males with all velocities. * Significant at p < .05.
5. Discussion The GC phases at the different velocity measures were highly correlated with other standards [6,7] (Fig. 1) while there were small differences for IC and MSw phases. It may be because that we used high speed camera and previous studies were derived from gait analyzer and footswitch. Second, ambiguity of the phase definition could also be responsible for variation. Heel rise determined the end of MSt and the start of TSt, however it was difficult to determine the exact moment of heel rise, as the soft tissue of the heel may sag below the actual bony heel rise on video. Instrumented devices like footswitches are commonly used to detect heel rise and other gait events [11,12], but they have limited accuracy depending on the measuring device itself and on how they attach to the body. Consequently, the end of MSw and the start of TSw were defined under two conditions: tibia was vertical and the opposite foot was parallel to the ground [6,7]. However occasionally it is difficult to determine the moment when the tibia was vertical because of its shape. Often the two conditions could not be satisfied in the same video frame. When conditions ‘Tibia is vertical and the foot is parallel to the ground’ for the end of MSw phase and the start of TSw could not been seen at the same frame, then the condition ‘the foot is parallel to the ground’ was considered as a key condition for the corresponding phase. ICC analysis has demonstrated that the measurement of PT of gait subphase by the camera is reliable. Toe off was used to identify the end of LR and PSw, and begins as the foot was lifted for floor clearance. There may be three toe off patterns: first and the most common pattern is that great toe rotates about 60–908 with the great toe on the ground (toe rocker); second, the great toe is rotating about 30–608 without a obvious toe rocker; third, the great toe sometimes did not rotate till the toe was lifted to initiate swing with little or no toe rocker. We defined the video frame as toe off when the great toe starts moving forward horizontally; however, similar to IC and heel rise, it was difficult to identify the contact frame because of the soft tissue of the feet and the optical center of the camera could not absolutely opposite to the lowest gap between foot and ground. To eliminate the visual illusion of early contact, we defined the frame when a subtle heel soft tissue deformation can be seen as the start of IC. Another possible reason may be ethnicity difference: all subjects are Chinese in ethnicity that could have different gait phase patterns compared with Caucasians and perhaps many other ethnic groups. These differentiations may seem arbitrary or too obscure to be of any possible clinical value. However, in recent clinical studies mere milliseconds of difference in GC determinations have lead to important clinical conclusions [4]. In fact, for valuable interpretation of GC data, it appears that high resolution data will be necessary to determine real clinical normalcy after intervention or surgeries such as total knee replacement. In general, the PT spent in each phase from our study matches well with the previous standard, especially for normal velocity
gaits. However, the PT spent in each phase varies from person to person with the largest coefficient variation of 24.31%. The range of PT is much larger than the mean PT spent in IC phase, up to 120% [(Max Min)/Mean]. Our data has also shown that the PT of LR, MSt and PSw are significantly negatively correlated ( 0.83 for LR, 0.75 for MSt, 0.84 for PSw) with gait velocities, which indicates that velocities strongly influences PT of individual GC phases, especially during stance sub-phases. These data are consistent with those reported previously [9,13]. When velocity is increased from slower to faster velocities, PT of MSt is decreased almost 10% of GC, from 35.30% to 25.33%, and from 14.55% and 64.69% to 10.32% and 60.21% for LR and PSw respectively. Mean/Perry (mean PT spent in gait phases divided by Perry’s results) was calculated for the direct comparison. The results of these data have shown that Mean/Perry for LR are 121.27%, 107.45% and 86.02% for slower, normal and faster velocities, and Mean/Perry for MSt are 113.87%, 101.80% and 81.72% for slower, normal and faster velocities. In future clinical studies, patients could improve velocities, cadence, stride length and distance etc. with the changes of PT of sub-phases, so evidence from this study indicates that velocity and individual variation should be taken into consideration for better evaluation of gait function when using eight sub-phases classification from Perry [6,7]. As velocities are increased, GC durations, percentages of Initial DLS, percentages of terminal DLS, and percentages of stance phase are decreased, while percentage of swing phase, percentage of SLS and stride length increased, which matches with results from other studies [8,9,14], which indicates that heel rise and toe off are earlier with fast velocity. It could be the result from fast moving of the center of mass and center of pressure [13,15]. These observations are likely to result in adjustments of interpretations in patients with altered gait after knee or hip surgery. There are certain limitations in the study. First, all subjects are Chinese and it is not clear if ethnicity can influence the results; second, the mean age was fairly young and may influence the results and finally, although there was a significant difference between slower and normal velocities, the absolute difference between the slower and normal velocities was smaller than the difference between the faster and normal velocities. Consequently a larger sample size study may be necessary to reveal significant differences. Future study may focus on joints angle deviations and electromyographic changes at different velocities of the eight subphases of the GC. Acknowledgements This study was partially supported by National Natural Science Foundation of China (Grant 31170900), Beijing Natural Science Foundation Program and Scientific Research Key Program of Beijing Municipal Commission of Education (Grant KZ201310025010) and Research Fund for the Doctoral Program of Higher Education of China (Grant 20121107110018).
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
G Model
GAIPOS-4063; No. of Pages 5 Y. Liu et al. / Gait & Posture xxx (2013) xxx–xxx
Conflict of interest The authors have no financial and personal relationships with other people or organisations that could inappropriately influence (bias) their work. References [1] Powers CM, Rao S, Perry J. Knee kinetics in trans-tibial amputee gait. Gait Posture 1998;8(1):1–7. [2] Yan S, Zhang K, Tan G, Yang J., Liu Z. Effects of obesity on dynamic plantar pressure distribution in Chinese prepubescent children during walking. Gait Posture 2013;37(1):37–42. [3] Aminian K, Trevisan C, Najafi B, Dejnabadi H, Frigo C, Pavan E, et al. Evaluation of an ambulatory system for gait analysis in hip osteoarthritis and after total hip replacement. Gait Posture 2004;20(1):102–7. [4] Lester DK, Zhang K. Gait analysis of knee arthritis treated with hyaluronic acid. J Arthroplasty 2010;25(8):1290–4. [5] Chevalier TL, Chockalingam N. Effects of foot orthoses: how important is the practitioner? Gait Posture 2012;35(3):383–8. [6] Perry J, Burnfield JM. Gait analysis: normal and pathological function. 2nd ed. SLACK, Inc.: Thorofare, NJ; 2010.
5
[7] Perry J. Gait analysis: normal pathological function. SLACK, Inc.: Thorofare, NJ; 1992. [8] Murray MP, Kory RC, Clarkson BH, Sepic SB. Comparison of free and fast speed walking patterns of normal men. Am J Phys Med 1966;45(1):8–23. [9] Murray MP, Mollinger LA, Gardner GM, Sepic SB. Kinematic and EMG patterns during slow, free, and fast walking. J Orthop Res 1984;2(3): 272–80. [10] Chinese Medical Association Subsection of Cardiovascular Disease, Chinese Journal of Cardiology Editorial Board. Highlights of the Second National Conference on Dyslipidemia. Chin J Cardiol 2002;30:643–6. [11] Perry J, Burnfield JM, Gronley JK, Mulroy SJ. Toe walking: muscular demands at the ankle and knee. Arch Phys Med Rehabil 2003;84(1):7–16. [12] Abu-Faraj ZO, Harris GF, Abler JH, Wertsch JJ. A Holter-type, microprocessorbased, rehabilitation instrument for acquisition and storage of plantar pressure data. J Rehabil Res Dev 1997;34(2):187–94. [13] Chiu MC, Wu HC, Chang LY. Gait speed and gender effects on center of pressure progression during normal walking. Gait Posture 2013;37(1):43–8. [14] Jordan K, Challis JH, Newell KM. Walking speed influences on gait cycle variability. Gait Posture 2007;26(1):128–34. [15] Orendurff MS, Segal AD, Klute GK, Berge JS, Rohr ES, Kadel NJ. The effect of walking speed on center of mass displacement. J Rehabil Res Dev 2004; 41(6A):829–34.
Please cite this article in press as: Liu Y, et al. Gait phase varies over velocities. Gait Posture (2013), http://dx.doi.org/10.1016/ j.gaitpost.2013.10.009
