Participants
A total of 10 male volunteers (age 20.4 ± 0.5 years, weight 74.6 ± 5.2 kg, height 176.5 ± 3.7 cm, BMI 23.9 ± 1.5 kg/m2) with no signs of neurological or musculoskeletal impairment participated in the study. Exclusion criteria included: recent major ligament injury, surgery, fracture or muscle injury in the lower limb, abnormal gait pattern, contraindication to exercise, or other health conditions that would adversely impact gait characteristics. Based on previous validation studies with Physilog® [19, 20, 22], the sample size required to calculate the ICC was determined to be 10, which is the number of people required to satisfy the conditions of one examiner, a significance level of 5%, an ICC estimate of 0.8, and a confidence interval of 0.2. All the 10 subjects had the right leg as dominant (the preferred limb used to kick a ball). The protocol for this study was approved by the Institutional Review Board of Hokkaido University (#16-062), and all participants provided their written, informed consent before participating.
Walking Protocol
Subjects wore the Physilog® sensors (Physilog4) on each foot with a Velcro strap and the retro-reflective markers for the OMCS (Motion Analysis, Santa Rosa, USA) (Figure 1). For simultaneous analysis, reflective markers were placed on both feet (first and fifth metatarsal joints, heels) and both ankles (medial and lateral condyles) of the participants with a foot configuration corresponding to the Physilog® sensors. The participants walked 10 round trips at a comfortable pace along a 10-m straight path, and 6 round trips of the middle of the session were used for the analysis. Four 0.5-m-long force plates (AMTI, Watertown, MA, USA) were embedded in the middle of the walkway to identify the time of heel-contact and toe-off. To identify each step on the force plate, each walk was recorded by a digital video camera. The obtained foot data were selected for further comparative analysis when a whole gait cycle on the force plates was properly identified. If slippage of a step over the force plate was found on the video, that gait cycle was excluded from the analysis. All gait parameters of the limb calculated by Physilog® were comparatively analyzed with the same parameters calculated within the Visual3D pipeline (C-motion Inc., Germantown, MD, USA) using the OMCS data. Table 1 provides the definitions of each parameter examined in the present study. To measure the repeatability of the Physilog®, all subjects returned for a second day of testing within two weeks. The protocol remained identical to the first day, and the order of the trials was preserved.
WGAS
The gait analysis system of the Physilog® consisted of two small (50 mm ´ 37 mm ´ 9.2 mm), lightweight (19 g), inertial sensors for each foot, elastic straps to attach the sensors to the dorsum of the foot, and Gait Analyser software version 3.1 (GaitUp) running on a Windows PC (Microsoft, Redmond, WA, USA). Two sensors on both feet can be synchronized wirelessly, and no calibration procedure is required before and during the measurement. The algorithm estimates vertical alignment by detecting the vertical gravity axis from the accelerometer during the standing posture and azimuth alignment by maximizing the pitch angular velocity during walking. The position of the sensor on the foot does not affect the measurement [27]. Signals were sampled at 200 Hz and stored on an internal memory card. The recorded data were converted to left and right spatiotemporal gait parameters (per gait cycle) using the Gait Analyser software. The dedicated algorithms have been described elsewhere [21, 26–28].
OMCS
As a reference for the WGAS, lower extremity kinematic and kinetic data were measured by the system consisting of three-dimensional motion analysis and force platforms. A 12-camera system (Raptor-E, Motion Analysis Corp., Santa Rosa, CA, USA) captured the motion at a sample rate of 200 Hz, and four force platforms (AMTI, Watertown, MA, USA) recorded the ground reaction forces (GRF) at 1000 Hz. Twenty-three, 12.7-mm-diameter, retro-reflective markers were placed on specific locations of the pelvis, thighs, knees, lower legs, ankles, and feet to calculate joint centers and segment positions and to track segment motions, as mentioned above, followed by data processing with custom Visual3D software. The lower body pipeline, which is based the Helen Hayes model, was used to calculate left and right spatiotemporal gait parameters. The data for the marker positions and force were smoothed by a fourth-order, zero phase shift, Butterworth low-pass filter at a cutoff frequency of 6 Hz for the positional data and 18 Hz for the force data. The cutoff frequency was determined by conducting a residual analysis. Initial contact and toe-off events were defined as when the vertical component of the unfiltered GRF exceeded and fell below 10 N, respectively.
As the laboratory coordinate system, camera calibration was conducted just before the motion measurement. The Ylab and Zlab axes corresponded to the posterior-anterior (direction of travel is positive) and inferior-superior (vertical upward direction is positive) directions, respectively. The Xlab axis (with the medial-lateral direction, right side to the direction of travel is considered positive) was calculated from the external product of the Ylab vector and Zlab vectors. Participants were instructed to walk in the positive direction of the Ylab axis. The segmental coordinate system of the foot segment was constructed from the markers affixed to the participants’ feet. For the right foot segment, the midpoint of the line segment connecting the medial and lateral malleoli of the ankle joint was set as the joint center of the ankle, and the vector from the medial malleolus to the lateral malleolus was set as the Xfoot axis for the right foot segment (the vector from the lateral malleolus to the medial malleolus was set as the Xfoot axis for the left foot segment). Next, the midpoint of the markers affixed to the distal ends of the first and fifth metatarsals was defined as the toe, and the Zfoot axis was defined as the external product of the Xfoot vector and the vector from the joint center of the ankle to toe. The Yfoot axis was defined as the vector calculated by the external product of the Zfoot and Xfoot vectors. To generate loading, foot-flat, and pushing (Figure 2), the stance phase was separated based on the waveform data of the pitch angular velocity of the foot segment. The events of toe contact and heel off were defined by the pitch angular velocity of the foot segment being above −2 rad/s and below −1 rad/s during the stance, respectively [28].
Statistical analysis
Statistical analysis was performed using SPSS version 23 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism version 8.4.2 (GraphPad Software, San Diego, CA, USA). To compare the data of the same parameter obtained by the Physilog® and the OMCS, the absolute difference and Pearson’s correlation coefficients (r) were calculated. To visualize the amount and tendency of the system deviations of the two systems, a Bland-Altman plot was used. Limits of agreement were calculated as Meandiff_WO ± (1.96 × SDdiff_WO) with Meandiff_WO being the mean difference between the WGAS and OMCS and SDdiff_WO being the standard deviation of the mean difference between the WGAS and OMCS. Lin’s concordance correlation coefficient (LCC), an index of how a new test reproduces a gold standard test, evaluates the degree to which pairs of observations fall on the 45° line through the origin [30]. To determine the repeatability of the WGAS measurements, LCC, r, and intraclass correlations [ICC(3,1)] between the two testing days were calculated. The following criteria were used to determine the strength of agreement for all parameters. The extent of LCC was categorized into 4 groups: Excellent (0.75–1.00), Good (0.60–0.74), Fair (0.40–0.59), and Poor (<0.40).[31] ICC(3,1) is a common statistic for evaluating repeatability [32] and is needed to calculate the minimally detectable change (MDC). SEM is the standard error of measurement, and SD is the standard deviation of the measure. SEM and MDC were calculated by the following formulas.
