Force platform analysis

2.4.1 Parameters used to assess centre of pressure excursion

Since the early 1970s, force platforms have been used to acquire quantitative measures for analysis of postural control (Palmieri, Ingersoll, Stone & Krause, 2002). Quantitative measures are gained by the force platform by the recording of GRF projected from the body (Browne & O’Hare, 2000). From this CP can be calculated, this reflects the trajectory of centre of mass and the amount of torque applied at ground surface to control body mass (Winter, Patla & Frank, 1990). Parameters that can be used include, mean sway amplitude, maximum sway amplitude, minimum sway amplitude, peak-to-peak amplitude, sway path, sway velocity, RMS amplitude and RMS velocity, this allows the researcher to quantify alterations in balance (Palmieri et al. 2002). Within the game of golf the double-legged stance is applicable (Palmieri et al. 2002). When CP is obtained when both feet are in contact with a single force platform, it carries the term net CP

(Winter et al. 1990). Over two thirds of body weight is balanced, two thirds above the ground when an upright stance is adopted, this is what places the demands on the postural-control system (Browne & O’Hare, 2000).

The maximum peak amplitude is defined as the absolute displacement of the CP from its mean, whereas minimum amplitude is the minimum displacement of the CP from its average point (Palmieri et al. 2002). Within terms of gait, an increase in either maximum or minimum amplitude suggests a decreased ability to maintain an upright stance, and visa versa for a decrease in either variable (Palmieri et al. 2002). Palmieri et al. (2002) questions the use of these parameters, as they are one dimensional, allowing for the assessment of postural control in both anterior/posterior and medial/lateral, but may not accurately reflect balance. The lack in accuracy may lie in the fact that it is a maximum and minimum measure, which only requires one point to be examined and therefore there can be great variability between trials and subjects (Palmieri et al. 2002). Peak-to- peak amplitude represents the difference between the maximum and minimum amplitudes of CP (Geurts, Nienhuis & Mulder, 1993). Again Palmieri et al. (2002) questions the accuracy in this parameter, due to the large variability. It was concluded in Palmieri et al. (2002) that maximum and minimum amplitude values and peak-to-peak amplitude, are likely to cause misinterpretation of alterations in balance and should not be used to evaluate postural control.

Mean amplitude of CP is an average value over all data points collected in a trial and is a more representative measure of postural control (Palmieri et al. 2002). Increased values in mean CP amplitude suggest decreased postural control, whereas a decrease is thought to represent increased postural stability (Baloh, Jacobson, Beykirch & Honrubia, 1998; Le Clair & Riach, 1996). The mean amplitude of CP does however have limitations, as it is susceptible to noise, Palmieri et al. (2002) suggests using an average of multiple trials to resolve this potential problem. Palmieri et al. (2002) states the importance of when using mean amplitude is defining where the mean amplitude is calculated from, whether it be the centre of the force platform or is dependent on stance location. Mean CP has been shown to fluctuate with different degrees of stance width, stance length and foot angle (Kirby, Price & MacLeod, 1987).

A theme discussed in Palmieri et al. (2002) is total CP excursion and CP velocity. Total excursion of the CP is defined as the total distance traveled by the CP over the course of the trial duration (Palmieri et al. 2002). In terms of postural control literature suggests that increases in total CP excursion represents a decreased ability in the postural-control system to maintain balance (Holme et al. 1999; Ekdahl, Jarnlo & Andersson, 1989; Uimonen, Sorri, Laitakari & Jamsa, 1996). Palmieri et al. (2002) identified a potential limitation in the interpretation of total CP excursion, as it

is possible to see a large total CP excursion during a stable stance or a small total CP excursion representing an unstable stance. A large total CP excursion may suggest that the CP needs to make sizeable excursions to maintain a stable stance (Palmieri et al. 2002). CP velocity represents the total distance traveled by the CP over time (Palmieri et al. 2002). CP velocity has been shown to be reliable between sessions when a double-legged stance is employed (Le Clair & Riach, 1996). An increase in CP velocity is thought to represent a decreased ability in postural control, whereas a decrease in CP velocity suggests a greater ability to control posture (Baloh, Jacobson, Beykirch & Honrubia, 1998; Le Clair & Riach, 1996). Again Palmieri et al. (2002) questions the use of this parameter when assessing postural control, as how the variable is represented in the literature may not be accurate. A limitation of both the total CP excursion and velocity of CP excursion, is both are two-dimensional, representing a combination of AP and ML CP excursion, and therefore Palmieri et al. (2002) claims important directional information can be easily be missed.

Root-mean-square (RMS) amplitude represents the standard deviation of the displacement of the CP (Palmieri et al. 2002). This parameter measures the average absolute displacement around the mean CP (Palmieri et al. 2002). A decrease in both RMS amplitude and RMS velocity represents an increased ability to preserve an upright stance, an increased value for either variable suggests a decreased ability in postural control (Geurts et al. 1993). Literature suggests that RMS amplitude and velocity are reliable measures to evaluate postural equilibrium (Geurts et al. 1993; Le Clair & Riach, 1996). Geurts et al. (1993) reported that RMS amplitude had a coefficient of variance of 31.75% and RMS velocity had a coefficient of variance of 26.75% show sufficient intra-subject consistency over a 5-week period. In addition to this Le Clair and Rioch, (1996) demonstrated RMS amplitude intersession reliability in the anterior/posterior direction (r= .86) and the medial/lateral direction (r= .81). Palmieri et al. (2002) states intraclass correlation coefficients need to be analysed to support this reliability.

2.4.2 Measurements of centre of pressure within golf literature

Ball and Best, (2007b) analysed centre of pressure patterns within the golf swing, using two AMTI force plates covered in artificial turf, one placed beneath each foot. Ball and Best, (2007b) normalised CP to foot position at address, which is important as discussed by Palmieri et al. (2002). This was achieved by attaching an overhead camera (50Hz) to capture foot position at address relative to the force plate coordinates (Ball & Best, 2007b). Along with this using a Peak Motus system the heel and toe of each foot were digitised four times, with the average of the four points used to indicate the position of the foot for each swing (Ball & Best, 2007a). From the

and toe along the Y-axis) can be calculated and the centre of pressure along the X-axis can be expressed (Ball & Best, 2007a).

Ball & Best, (2007a) decided to analyse CP along the ML axis at swing events in preference to using centre of pressure time curves. Swing events were chosen because; players and coaches easily understand them (Ball & Best, 2007a). Secondly, there is evidence to suggest that using time-normalised data can have significant limitations because of issues of temporal dependency (Forner-Cordero, Koopman & van der Helm, 2006). The problem arises from the assumption that there is no variability in the timing of events between take-away and ball contact and that no rescaling occurs during the percentage conversion (Ball & Best, 2007a). Ball and Best, (2007b) used the parameters Maximum CP velocity (ms-1), time of max velocity relative to ball contact (s), vertical force underneath each foot (Fz%), maximum CP along the ML axis and CP along the ML axis range in metres, from this weight transfer could be analysed.

Force platforms also have been used to test the amount of torque occurring and the shoe-natural grass interface (Worsfold, Smith & Dyson, 2008). Like Ball and Best, (2007a) and Ball and Best, (2007b) two force platforms with a natural turf surface placed on top were used, one placed beneath each foot (Worsfold et al. 2008). Unlike Ball and Best, (2007a) though was the use of a thin strip of clay attached to a plastic sheet, so the turf could be fixed to the force platforms. Results of this Worsfold et al. (2008) study demonstrated considerable force generation at the golf shoe-natural grass surface interface (17-19Nm). Barrentine, Fleisig and Johnson (1994) also looked at the (GRF) using two force platforms placed beneath the feet. It was observed that LH golfers achieved maximum torque with the rear foot earlier in the downswing, which can be related to the greater CHV that was observed for the LH golfer (Barrentine et al. 1994). As Hume, Keogh and Reid, (2005) discuss this GRF is important to maximize distance obtained with the driver and long irons.

2.4.3 Reliability of force platforms

As well as hard based force platforms, there are also plantar measurement devices (PMDs) which are often considered a less powerful choice in both clinical and a research context (Giacomozzi, 2010a), although the potential of PMDs is highly recognised (Giacomozzi, 2010b; Putti, Arnold, Cochrane & Abboud, 2008). Current problems include comparisons between different PMDs; current research has not presented absolute pressure values (Alvarez, De Vera, Chhina & Black, 2008; Thijs, Van Tiggelen, Roosen, Declercq & Witvrouw, 2007), absolute pressure values may help in understanding how much comparable different datasets are (Giacomozzi, 2010a).

Rather than absolute pressure values, clinicians and researchers are often more concerned with relative pressure values or relative pressure distribution changes, relative values should be approached with caution when comparisons are being made. This is due to the values being post – processing products, and can be affected by PMD sensor response (Giacomozzi, 2010a). As shown in Table 2.1 the leading PMDs all have different Pressure ranges and resolutions, so comparisons when using different PMDs can be difficult. Giacomozzi (2010a) identified the following parameters for assessment of the reliability of PMDs. Sensor response variability with respect to different pressure levels, dividing the PMD into large sub – areas. Sensor response in terms of absolute value of pressure over a small, uniformly loaded area within the loading ranges. Sensor hysteresis, measured by a loading – unloading frequency not greater than 1 Hz. Sensor response in terms of creep, so to vary the pressure response and platform response in terms of accuracy and repeatability of CP coordinates estimation.

Table 2.1. Main characteristics of five commonly used PMDs (Giacomozzi, 2010b). AM CUBE MEDILO

GIC

NOVEL RSSCAN TEKSCAN

Tested Device

AM3 platform Medilogic platform

EMED-x Rsscan platform Matscan

Technology Capacitive, air-based Resistive Capacitive, elastomer- based Resistive Resistive

Calibration In-factory (up to 900 kPa)

In-factory In-factory In factory plus user calibration In factory plus user calibration Overall Sensor Matrix 64 x 64 32 x 64 64 x 95 64 x 64 44 x 52 Pressure range (kPa) 0-1200 0-640 0-1270 Not Available 0-850 Resolution (sens/cm2) 1.7 1.78 4 2.67 1.4 Active sensors per area 9 9 16 9 9

Giacomozzi (2010c) technically assessed the PMDs using a custom made pneumatic bladder pressure tester (PM) and a pneumatic – force testing device (PTD). The PM was used to uniformly apply pressure over the entire PMD sensor matrix. The PTD consists of a pneumatic testing device with an on off valve, a proportional valve and pressure controls to manipulate pressure ranges from 0 – 600 kPa (Giacomozzi, De Angelis, Paolizzi, Silvestri & Macellari, 2009; Giacomozzi, 2010b). An additional tool was built in order to assess CP coordinates allowing the application of known forces through three pylons; theoretical CP coordinates could then be

were averaged over a 10 second static loading period under six angular positions, from this root means square error could be calculated (Giacomozzi, 2010c).

The RS FootScan pressure range was not reported in this research as an exclusive technical note was agreed for more extensive analysis. The technical assessment of the other PMDs showed good reliability (Giacommozzi, 2010b). Centre of pressure estimation showed high precision for all PMDs, NOVEL accuracy error was always lower than its PMD spatial resolution (0.25 cm); TEKSCAN and AM CUBE error was greater than the spatial resolution (0.35 and 0.39 cm respectively); MEDILOGIC error however was always greater than spatial resolution (0.37 cm) (Giacomozzi, 2010b), therefore for CP analysis the NOVEL would provide the most accurate results. Giacomozzi (2010b) results showed that the capacitive, elastomer – based PMD by NOVEL showed high accuracy and precision in CP estimation, with low variability of all performances over the whole sensor matrix. The resistive PMD by TEKSCAN, demonstrated high accuracy and precision in CP estimation except for one tested area, this is also applicable for the capactive air – based PMD by AM CUBE. In the current study a RS FootScan was available, like the TEKSCAN the RS FootScan uses resistive technology and therefore reliability results may be similar.