Development of a Cost-effective Biomechanical System for the Assessment of Kinematic and Kinetic Parameters in Soccer Players During a Lower Limb Pointing Task

Purpose The main aim of the present study was to describe the development of a non-commercial biomechanical system designed for the simultaneous measurement of kinematic and kinetic parameters during a lower limb pointing task. The task was performed by two groups of soccer players (10 novices and 10 experts, aged 16-19 years) and the differences between the two groups, in terms of these parameters, were also assessed. Methods calibration procedure of the combined The system measured the Centre of (CoP) and body of Mass (CoM) displacements, the of leg. Results

Introduction A lower limb pointing task is mainly related to kicking accuracy, which entails delivering the ball with precision over a desired distance to an intended target [1,2]; precision is achieved by reducing velocity during an accurate kick, which allows players to minimise errors. Reducing velocity results in longer duration of the shooting phase [3]. Duration refers to the time a player takes to carry out the kicking movement, from the moment the supporting leg is in contact with the ground to initial ball impact [3,4,5].
Maintaining postural stability-i.e. keeping the body's centre of mass (CoM) over the base of support by moving the centre of pressure (CoP) of the ground reaction forces (GRF) beneath the feet-during an accurate kick also allows the players to minimise errors and attain kicking accuracy [6,7,8]. To maintain postural stability throughout the phases of the kick, players must minimise CoM and CoP displacements and velocities [2,8,9].
Several methods have been developed to measure the above kinematic and kinetic parameters related to the soccer kick. Velocities and accelerations have been measured using two-and three-dimensional videography [2,10,11,12,13,14] and accelerometry [13,14]. Three-dimensional motion capture can measure kinematic parameters more accurately; however, this method requires several cameras and markers, which is time consuming and can have other problems: for example, the employment of smoothing lters makes it di cult to distinguish between true deceleration due to the impact with the ball and false acceleration signals produced by the noise in the data [15,16,17,18,19]. Accelerometry, on the other hand, measures human movement through sensors directly attached to the body part which is measured without requiring any lters [13,14,19].
External forces (GRF, CoP) can be measured through a force platform which provides electrical signals relative to the applied forces [20,21]. Measuring these forces simultaneously with kinematic data through combined systems can provide a time-effective and comprehensive method [21,22,23]. A study by [22], used a combined system consisting of a force platform, an accelerometer and an electromechanical device to simultaneously measure kinetic and kinematic variables during a controlled kick by comparing the signals from the three devices.
To our knowledge, non-commercial combined systems are scarce. We, therefore present a combined system that consists of three devices for the simultaneous measurement of kinematic and kinetic variables related to a lower limb kicking task. Furthermore, since the players' level of expertise has been shown to in uence these parameters [8, 23,24,25,26], a secondary purpose of the present investigation was to use the system to measure the differences between expert and novice players, in terms of CoP and CoM displacement, kicking accuracy and duration during the performance of the kicking task.

Materials And Methods
Description regarding of the devices and kinematic and kinetic apparatus and calibration procedures can be found in the supplementary material.

Experimental study
An experimental study consisting of a lower limb pointing task was performed by twenty male (10 experienced and 10 novice) soccer players (mean age 18.35 ± 0.58 years old, weight 72.8 ± 5.69 kg, and height 178.36 ± 5.74 cm). None of the participants reported any orthopaedic condition or musculoskeletal disorder during the six months prior to the data collection. The experienced players had to participate in regional level championships (U-routine of three 1-hr training sessions per week. The participants' task was to kick a ball at their own pace with the instep of their right foot (declared as the participants' preferred foot) as accurately as possible, 20), keep a routine of four 1.5-hr training sessions per week, and have at least ve years playing experience. The novice players were amateur college students with a minimum of three years playing experience and had to keep a toward a target (2.0 m x 2.0 m) positioned at a ground level, 6.0 m away from the participants. The right foot was placed out of the platform, close to the ball, while the left foot remained on the force platform throughout the task. Once stabilised through a minimal sway for 3 s, the participants performed the accurate kick, and then returned to the starting position without stopping the movement or touching the oor with their right foot. If a participant stopped the movement or touched the oor the trial was considered invalid and had to be repeated. Each participant performed eight kicks with one min rest interval between kicks.
The three-dimensional coordinates were recorded through two digital cameras which were turned on after the participants had positioned themselves on the FP and turned off after the performance of the kick (Figure 1, supplementary material). These coordinates were expressed as a right-handed orthogonal reference frame xed on the ground. The following sign convention was adopted: x for the AP axis, horizontal and pointed to the centre of the target (positive forward and negative backward), y for the ML axis, perpendicular to x (positive to the right and negative to the left) and z for the craniocaudal axis (positive upward and negative downward).
All signals were captured by an HBM Spider8 conditioner with a CATMAN software, sampled at 100 Hz (HBM Spider8, with 16-bit resolution and 16 channels A/D converter). After the kicks had been recorded, the les were archived and sequenced from 1 to 8 ( rst to last kick). The video of each kick was divided into frames with one image every 0.033 s (30 Hz). To obtain the CoM displacement for each phase of the kick the program "am3DKick" (UNESP -Biomechanics Laboratory) was used. Fourty frames were obtained from each digital camera marking the anthropometric points of the images (total of 80 frames). The data were saved and archived for further analysis.
The FP signals were sampled at 100 Hz (HBM Spider8, with 16-bit resolution and 16 channels A/D converter) and ltered through a zero lag second-order pass Butterworth lter (10 Hz low-pass cut-off frequency with a dual pass to remove phase shift. The accelerometer was synchronised with the force platform through a resampling method of two time series objects using a common time vector in Matlab. The data from both time series were collected through a conditioner of signals (Spider 8) and interpreted by the CATMAN software. The trials were also viewed off-line on a monitor screen and temporally synchronised, according to the rst visible de ection of the right ankle kinematics signal, de ned as the instant the right foot was raised off the ground. This time was considered as "time zero" (T0) for all subsequent analyses.

Dependent variables
The dependent variables associated the lower limb precision task, included kicking analysis and calculation of mean AP and ML amplitude (mm) and speed (mm/s) of CoP and CoM displacement. Three dependent variables described the kicking movement: kicking accuracy, duration, and right foot acceleration. Kicking accuracy was established as the sum of the different scores across trials, as follows: 1) the ball missed the target, 2) the ball reached the outer target zone 3) the ball reached the central target zone. The higher the score, the higher was the kicking accuracy.
The duration of the accurate kick was examined in connection with the backswing and shooting phases; the latter corresponds to the forward and impact phases presented by other authors [2,27]: 1) The backswing phase: the swing of the kicking leg in preparation for the downward motion towards the ball which started with the raising of the right ankle until the maximal backward position of the limb was reached 2) the shooting phase: kicking towards the target which started from the maximal backward position until the right foot was in contact with the ball toward the target [26,27]. The sum of those phases represented the total duration of the kick. The foot contact with the ball was determined directly through the accelerometer signal, de ned as the time of peak deceleration during the shooting phase.
Four dependent variables mean mediolateral (ML) and anteroposterior (AP) amplitude (mm) and speed (mm.s −1 ) of CoP displacements, described the participants' postural behaviour during the kicking task for the backswing and shooting phase. The amplitude of CoP displacements indicated the mean deviation of CoP in the ML and AP axes. The mean speed of CoP displacements is the sum of the displacement scalars divided by the sampling time, i.e., the duration of each movement phase. It represents the amount of activity required to maintain stability and provides a more functional measure of postural control [28,29].
The amplitude and speed of CoM displacements in the ML and AP axes were obtained through the kinematics method that estimates CoM displacements through the trajectory of the body segments in function of time [17]. To calculate the CoM displacement in the ML and AP axes, a three-dimensional reconstruction of the markers' kinematics data combined with anthropometric measurements was performed through off-line data processing techniques from Winter's model, in which the mass of each segment is given as a function of the total mass and the CoM, and the radius of rotation of these segments is provided as a function of their lengths [30].
To allow the kinematic identi cation of each anatomical segment and ensure reliable between-participant data, 15 mm styrofoam hemi-spheres were attached to the 16 anatomical landmarks in all the participants: left and right ear, acromion, anterior superior iliac spine, knee, lateral malleolus, fth metatarsal, elbow and wrist joints [30].

Statistical analysis
Kicking accuracy was submitted to a one-way analysis of variance (ANOVA). Kicking time, foot acceleration, CoP and CoM dependent variables were submitted to 2 conditions (experienced vs. novices) x 2 movement phases (backswing vs. shooting) ANOVAs with repeated measures for both factors. The level of signi cance was set at p < .05. Whenever the ANOVA showed a signi cant effect, a Tukey's HSD post hoc test was performed for multiple comparisons. All statistical analyses were carried out using SPSS 10.0 software (SPSS Inc., Chicago, IL, USA).

Calibration
Kinematic apparatus Spatial calibration results The equations used to reconstruct the images can be found in the supplementary material (pp 5,6). The margin of error found for calibration through reconstruction of the three-dimensional image was less than 1%.

Accelerometer calibration results
Results measured from the accelerometer signal showed that the maximum voltage indicated by the oscilloscope was + 2.8 V, the medium voltage + 2.64 V and the minimum voltage + 2.48 V, corresponding to +3g, +1g and -1 g respectively.

Kinetic apparatus Force platform calibration results
The sum of the outputs from the four load cells presented a correlation coe cient R² = 1 in the load range of 0 to 900 N, equivalent to the total capacity that each load cell can support.
The load cells presented a linearity and hysteresis with error below 0.25% of the vertical and 0.31% of the horizontal for a full scale (900 N) (Figure 4 a, b).
The hysteresis and linearity presented the following results: hysteresis (Mx = 4.91%, My = 7.16%, Mz = 6.04%); linearity (My = 0.9872, Z = 0.9973, Mx = 0.9692). The force platform presented a resolution of 0.003685 Nm in Mx, 0.004327 Nm in My and 0.035403 N in the Z axis. To measure the accuracy and precision of the CoP (reproducibility), the test of distributed load was applied, which is the closest to reality regarding the stabilometric platform. Table 2 Table 3.
The post hoc revealed that the duration of the backswing phase was longer in the novices than it was in the experienced players (p < .001) whereas the duration of the shooting phase was similar between the two groups. The foot acceleration of the backswing phase was slower in the novices than in the experienced players (p < .05) whereas the foot acceleration of the shooting phase was similar between the two groups.
No signi cant effect was observed when the kicking duration was analysed from T0 until the time that the foot was in contact with the ball toward the target (p > .05). The results for the mean amplitude and speed of CoP and CoM displacement are shown in Table 4.
The analysis of the mean CoP amplitude showed a main effect of condition, with a larger mean amplitude for the novices than for the experienced players. The main effect of phase and the interaction of Condition x Phase for the AP mean amplitude were not signi cant. The analysis of the mean CoP speed in the AP axis showed a main effect of condition, suggesting a faster CoP speed in the novices than in the

Discussion
The kinematic calibration through reconstruction of the 3D image presented a margin of error less than 1%, in accordance with previous studies [31]. Therefore, the three-dimensional system developed in the present study captured the participants' movement during the kicking task, in different planes/axes. The calibration results showed that the sensor presented a level of oscillation (g) range varying between -1 to +3 g, con rming that this device responded adequately to the acceleration to which it was submitted [32], which is in accordance with previous studies [22,31,32,33]. Furthermore, the platform presented an adequate connectivity, linearity and hysteresis as well as a satisfactory resolution and sensitivity of curves, combined with a structural rigidity and low weight. The errors in the CoP location were less than 1 mm, in accordance with previous studies [21,34].
Contrary to previous studies [34,35], the CoP measurement error found during the distributed load test was greater than the error observed during the application of the centralised load. This could have been caused by an uneven load distribution on the plate not well transferred to the four load cells or imperfections on the plate [22,35]. Though this characteristic appears to be negative, it showed that a system designed in this manner can reduce the hysteresis effects [21].
In addition, the system was used to measure the kinematic and kinetic parameters of the kicking task in expert and novice players since it has been shown that long term training contributes to high level of coordinated movement and balance [2,8,12,35]. The novices presented greater right foot acceleration than the expert players during the kicking action, which resulted in less accuracy [33,37,38]. Expert players in [39,40] achieved the most accurate kicks when they prioritised precision over velocity.
Moreover, the novice players spent more time to perform the backswing movement whereas the duration of the shooting phase was similar between the groups [41,42]. Novice players need to focus on the components of the performed task, resulting in delays in performing the task [43,44,45,46]. In the present study novices spent longer on maintaining stability and controlling the movement of their kicking leg during the backswing phase compared to experts; therefore, this phase lasted longer for the novices.
In addition, CoP and CoM analyses were performed in the AP and ML axes since postural balance during the kick depends mainly on the lateral and frontal oscillations of the supporting foot [20]. Novice players presented greater amplitude and speed of CoP displacement values and, therefore, greater postural instability than expert players. There was greater CoP variability in the AP than in the ML axis, which was more consistent with the backswing phase [27,41,47,48]. Previous studies investigating groups of amateur and elite soccer players, found greater variability in the ML axis and a reduced AP CoP displacement in amateur players [49,50], which apparently contradicts our results.
Novices presented a greater amplitude and speed of CoM displacement than expert players. A signi cant forward CoM displacement during the backswing phase [41] enabled the novices to maintain stability [41,51]. Amateur players in the study by [52] also presented a signi cant forward CoM displacement throughout the phases of an instep kick.

Conclusion
The present study aimed at developing a cost-effective combined system for the simultaneous measurement of kinematic and kinetic parameters during a lower limb pointing task performed by expert and novice players. The results of the calibration and experimental study indicate that the combined system presents a comprehensive method for the assessment of dynamic postural stability and kicking performance.
To ensure generalisability of ndings future research should include athletes of different ages, both male and female.

Declaration of con icting interests
The author(s) declared no potential con icts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The authors did not receive support from any organization for the submitted work.    Marking sequence for 3D calibration (Quintic Player®).