Abstract. A force platform is a technical method of quantitatively assessing balance indirectly. The use of force platforms in physiotherapy departments has become more prominent over the last few years. However, the main drawback in the use of force platforms is the lack of comprehensive calibration procedures, which casts doubt on the results obtained with these systems. Existing calibration tests are limited to testing the spatial accuracy of the force platform. This paper describes a comprehensive quality control test procedure, which was developed. It is proposed that the developed quality control test procedure could be used to test all types of force platform and it includes a description of how the tests should be carried out, the frequency with which they should be carried out and the expected performance for each of the tests as recommended for the most part by the Association Française de Posturologie.

     Force platforms are technical methods of assessing balance and as such should have calibration results to verify the accuracy and precision of their clinical results. A significant deficiency in the use of force platforms is the lack of a comprehensive quality control test procedure which could be used to assess the accuracy and precision of the force platform measurements on a month-to-month or year-to-year basis (Andres and Anderson 1980). The lack of such a quality control procedure casts doubt on the reliability of the various measured indices of balance described below, and also makes it impossible to compare results from month-to-month within the same clinic and between different clinics, even for the same type of force platform. It is very important to perform quality control tests so that it can be established whether the performance of the force platform is within the specifications.
     A force platform is an indirect technical method of measuring balance or sway, where the force exerted by the body on the ground is measured. From this measurement the person's centre of pressure (COP) is determined, and this is then related to their centre of gravity (COG) and ultimately to their balance. This is due to the relationship between these three parameters: the movement of a person's centre of gravity is a measure of their balance, and during quiet standing a persons centre of pressure can be approximated to their centre of gravity (Winter 1995). Force platforms measure the COP using force or pressure transducers, which produce an electrical signal proportional to the applied force.

• (a) Average distance of the COP from the mean position [mm] (Murray et al. 1975, Goldie et al. 1989, Kilbum et al. 1994).
• (b) The total excursion of the COP [mm] (Kilbum et al. 1994, Fernie et al. 1978),
• (c) The mean instantaneous velocity of the COP [mm.s-1] (Kilbum et al. 1994, Fernie et al. 1978),
• (d) The mean frequency of the COP excursions [Hz] (Hufschmidt et al. 1980, Lakes et al. 1981, Guerts et al. 1993, Lucy et al. 1985),
• (e) Root mean square (rms.) body sway velocity [mm.s-1] (Uimonen et al. 1994) and(f) The sway area, which is the extent of the total area travelled by the COP in a given time. This has been found to be useful in quantifying the extent of visual influence on postural stability [mm2] (Hufschmidt et al. 1980, Diener et al. 1984, McClelland 1989, Goldie et al. 1989, Jeong 1994, Uimonen et al. 1994, Hasan et al. 1996).

     Balance disorders can result from a loss of vestibular information, Menier's disease or vertigo; from damage to the CNS or the cerebellum, stroke or Parkinson's disease and also from damage to the musculoskeletal system. Different indices of balance are sensitive to the presence of different types of balance disorder (Guerts et al. 1993, Kilbum et al. 1994, Hufschmidt et al. 1980). The changes in the indices of balance due to impaired balance are not very large, 5 mm for a healthy volunteer and 7.1 mm for a patient with moderate Parkinson's disease, a difference of only 2.1 mm (Browne 1999). Therefore, the force platform measurements of the indices of balance need to be both accurate and precise.
     Numerous investigators have carried out calibration tests on force platforms; however, these calibration tests have been mostly limited to assessing only the spatial accuracy of the force platform and the sources of error causing the inaccuracy (Bobbert and Schamhardt 1990, Mita et al. 1993, Granat et al. 1990, Hall et al. l996, Starck et al. 1993). Each of the different investigators identified different sources of error for this spatial inaccuracy: excessive nonlinearity of the transducer response; different offset voltages of each of the transducers and electronic noise. Hall et al. (1996) suggested that a large source of error in the spatial accuracy of the force platform was the installation of the force platform and recommended that a calibration be performed at installation to minimise these errors. This calibration was limited to only determining the output and the crosstalk of each of the transducers. However, despite a knowledge of the errors which cause spatial inaccuracy no calibration procedure bas been developed to include the testing of the different parameters which contribute to the errors in spatial accuracy, such as the level of nonlinearity of the transducers or electronic noise in the analogue-to-digital converter signal of the force platform. Therefore, a quality control test procedure for characterising force platforms was developed. This quality control test procedure includes tests for all of the important aspects of the performance of force platforms, and the recommended performance criteria are for the most part those recommended by the Association Française de Posturologie to standardise the construction of force platforms (Bizzo et al. 1985). The Association Française de Posturologie recognised the need for standard construction specifications for force platforms to be developed and, therefore, outlined specifications for construction of a force platform with a spatial accuracy of 1 mm and a natural frequency greater than 20 Hz. Despite its obvious importance, no recommendation for precision was made by the Association Française de Posturologie and, also, it has not been investigated by other investigators.
     This quality control test procedure investigates the performance of force platforms from a static and a dynamic perspective. The static performance of the force platform is governed by its spatial accuracy and precision while the dynamic performance is governed by its frequency response. The spatial accuracy of the force platform is affected by the electronic noise and hysteresis of the system, the nonlinearity and different offset voltages of the transducers as mentioned earlier, while the precision of force platform is affected by the repeatability and the temporal stability of the system. Therefore, in order for the static performance of the force platform to be fully investigated linearity, hysteresis, noise, repeatability and temporal stability will be investigated as well as spatial accuracy and uniformity. In order for the dynamic performance of the force platform to be investigated the natural frequency and the frequency response, modulated transfer function (MTF) will be investigated. In this paper a quality control (QC) procedure for all types of force platform is proposed.

2.1. Quality control test procedure
     This QC procedure includes QC tests which may be carried out on force platforms which either have access only to the COP measurements or have access to both the COP measurements and to the voltage output from the analogue to digital converter for each of the transducers. The part of the QC procedure which uses the COP measurements will only provide general information about the force platform's accuracy and precision and will only allow problems with nonlinearity, hysteresis, noise, repeatability and stability of the force platform to be identified, whereas the part of the QC procedure which uses the output voltage from the analogue-to-digital converter can pinpoint which transducer has excessive nonlinearity, hysteresis, non-repeatability and instability.
All measurements should be made at room temperature (22°C) and the temperature recorded at 30-min intervals while the QC procedure is being carried out.

2.1.1. Static tests of the QC procedure.
     Spatial accuracy is one of the most important parameters of a force platform's characteristics as it is essential for a patient's treatment management that small changes over time can be accurately detected (Berg 1989). Spatial accuracy is affected by nonlinearity of the transducer's response to applied load; hysteresis in the transducers; different offset voltages of the individual transducers and electronic noise in the individual components of the force platform. For example if the transducers have a nonlinear response to applied load, then the COP may be underestimated or overestimated. The point of force application is measured using the force platform's own processing and analysis program; therefore the same test is used regardless of whether or not the output voltage from the analogue-to-digital converter is accessible. This is one of the more difficult tests to perform in the QC procedure because of the difficulty with positioning the calibrated loads on the top plate of the system. If an accurate co-ordinate grid is placed on the top plate of the force platform, it will facilitate more accurate positioning of the loads on the top plate.
     A number of different measurement techniques have been developed.
     The spatial accuracy was measured by Barrett et al. (1987) and Mita et al. (1993) by placing different magnitudes of calibrated loads on the co-ordinate of interest. The loads were then removed and reapplied on the next co-ordinate of interest. This was repeated for up to 40 co-ordinates, which gave an adequate representation of the co-ordinates of the top plate for the range of loads between 7 and 30 kg. However, this method is time consuming and subject to large sources of inaccuracies in stacking the loads.
     The spatial accuracy was measured by Starck et al. (1993), Bobbert and Schamhardt (1990), Hall et al. (1996) and Gill and O'Connor (1997) using dedicated rigs for applying loads of different magnitudes to specific locations, which gave an adequate representation of the co-ordinates of the top plate. A description of the construction of one such rig is described by Gill and O'Connor (1997).
     In this investigation a translational movement system was constructed, which translated a range of calibrated loads between 10 and 80 kg across the top plate of the force platform, without having to keep removing and reapplying the loads (figure 2). The accuracy of the force platform is determined in a relative manner by the translational movement system, the loads are moved by 2 cm and the accuracy with which the force platform can measure this 2-cm change in position is determined. The translational movement system consisted of a moveable plate on which loads were stacked and then easily and accurately moved by 2 cm (an accuracy of ±l mm). The loads were applied to 40 different locations, which gave an adequate representation of the co-ordinate of the top plate. The co-ordinate points measured were those contained within the area ±10 cm from the origin in both the x- and y-direction (Bizzo et al. 1985).
     The accuracy of the COP measurements should be accurate to within 1 mm when a 10 kg load is placed 10 cm out from the mechanical centre of the force platform (Bizzo et al. 1985, Barrett et al. 1987, Mita et al. 1993, Starck et al. 1993, Gill and O'Connor 1997).
     This test should be carried out as part of acceptance testing and at a minimum of six monthly intervals thereafter unless the stability of the system is such that it needs to be carried out more frequently.

     Uniformity is also an important parameter of the force platform's characteristics. If the force platform's top plate has a non-uniform response, then the measurements of a patient's COP will be affected if the patient stands on different positions of the top plate, or if the area of their COP increases it will not be measured accurately.

     The co-ordinate points should be contained within the area ± l0 cm from the origin in both the x- and y-direction (Bizzo et al. 1985). It is imperative that the load is accurately placed at each of the co-ordinates of interest and this can be achieved by placing a mat with the traced circumference of the base of the load on it and then positioning the load within the trace (figure 3).
     The force platform's top plate should have a uniform response to within 1 mm across the area contained within ± 10 cm from the mechanical centre in both the x- and y-directions (Bizzo et al. 1985).

     Therefore, the linearity of the force platform's transducers should be determined for this range of loads.
The linearity of the force platform can be determined by applying calibrated loads of between 5 and 100 kg to different positions on the top plate of the force platform and recording the COP measurements calculated. The COP measurements should not vary by more than 1 mm. Therefore, the response of the COP measurements should be independent of the load applied (Bizzo et al. 1985).
     Additional information about the linearity of the individual transducers can be obtained if the output voltage from the analogue-to-digital converter (ADC) is accessible. The linearity of the transducers can be determined by applying calibrated loads of between 5 and 100 kg directly over each of the transducers and the output voltage response of each of the transducers recorded for the increasing magnitude of load. The offset voltage of each of the transducers can also be obtained. The offset voltages of each of the transducers should be the same for all of them otherwise the COP measurements will be inaccurate (Beppu et al. 1985, Mita et al. 1993). The transducers should have a nonlinearity < 0.1% full scale (Bizzo et al. 1985).
     This test should be performed as part of acceptance testing and at a minimum of yearly intervals thereafter unless the stability of the system is such that it needs to be carried out more frequently.
     In normal operation the loads applied to each of the transducers will increase and decrease within a few milliseconds as the subject sways to and from the transducers.
     In order to simulate the clinical test situation a calibrated load between 20 and 40 kg should be applied to different positions on the top plate of the force platform contained within the area ±10 cm from the origin in both the x- and y- direction and the COP measurements recorded. Then a smaller calibrated load between 5 and 10 kg should be applied on top of the larger load and again the COP measurement recorded. The smaller load should then be removed and the COP measurement for the original calibrated load recorded. This test should be repeated at least four times at different positions contained within the area ±10 cm from the origin in both the x- and y-direction for an accurate measurement of hysteresis present in the system to be determined. The COP measurements should not vary by more than 1 mm (Bizzo et al. 1985).
     Additional information about the hysteresis of the individual transducers can be obtained if the output voltage from the ADC is accessible. The level of hysteresis in each of the transducers can be determined by applying a calibrated load between 10 and 40 kg directly over each of the transducers individually and the output voltage response of each of the transducers recorded once the output voltage becomes stable. Then a smaller additional calibrated load between 5 and 10 kg should be applied on top of the larger load and again the output voltage of each of the transducers recorded once the output voltage becomes stable. The smaller load should then be removed and the output voltage response of the transducers for the original calibrated load recorded again once the output voltage becomes stable. The above procedure should be repeated at least four times so that the amount of hysteresis present in the transducers can be determined. The transducers should have hysteresis < 0.2% full scale (Bizzo et al. 1985).
     This test should be performed as part of acceptance testing and at a minimum of yearly intervals thereafter unless the stability of the system is such that it needs to be carried out more frequently.

     The noise present in the force platform's measurements can be determined by applying a calibrated load between 10 and 40 kg to the force platform and the calculated displacement of the COP of the static calibrated load is representative of the amount of noise present in the system. The displacement of the COP measurement should be < 1 mm (Bizzo et al. 1985).
     Additional information about the contribution of the transducers and the ADC can be obtained if the output voltages from the ADC is accessible. The noise in the force platform's measurements can be determined by applying a calibrated load between 10 and 40 kg directly over each of the transducers and measuring the amount of variance in the output voltage signal of the ADC, which corresponds to the noise in the signal. The noise level should contribute to < 1 mm inaccuracy in the spatial accuracy of the force platform COP measurements (Bizzo et al. 1985, Granat et al. 1990).
     This test should be performed as part of acceptance testing and at a minimum of six monthly intervals thereafter unless the stability of the system is such that it needs to be carried out more frequently.

     Repeatability contributes to the precision of the COP measurements: it is important that the COP measurements are precise so that changes over time in the COP measurements can be detected. The repeatability of the force platform can be determined by applying a calibrated load of between 10 and 40 kg to the force platform at least 20 times in the same position and recording the COP measurement obtained each time. It is important that the load is accurately placed in the same position each time: this can be achieved by using a placement mat as described in the uniformity test procedure.
     Additional information about the repeatability of the individual transducers can be obtained if the output voltage from the ADC is accessible. The repeatability of the transducers can be determined by applying a calibrated load between 10 and 40 kg directly over each of transducers individually at least 20 times and measuring the output voltage response of the transducers from the ADC. It is important that the load is accurately placed in the same position each time: this can be achieved by using a placement mat as described in the uniformity test procedure.
     The Association Française de Posturologie made no recommendations for the repeatability of COP measurements; however, upon consideration of the small magnitude of difference, 2 mm, between patients with Parkinson's disease and age-matched healthy subjects it is evident that COP measurements should also be precise to within 1 mm, in order to detect changes in subject's ability to balance (Browne 1999, Berg 1989, Hufschmidt et al. 1980).
     This test should be carried out as part of acceptance testing and at a minimum of weekly intervals thereafter unless the stability of the system is such that it needs to be carried out more frequently.
Drift and temporal stability also contribute to the precision of the COP measurements. The drift in the force platform measurements over a 5-30 minutes time interval can be measured by placing a calibrated load between 10 and 40 kg on the top plate and recording the COP measurements every 5 minutes.
     Additional information about the drift in the individual transducers can be obtained if the output voltage from the ADC is accessible. The drift in the transducers over a 5-30 minutes time interval can be measured by placing a calibrated load between 10 and 40 kg over each of the transducers and recording the output voltage response of the transducers from the ADC every 5 minutes.
     The stability of the force platform over time can be determined by applying a calibrated load between 10 and 40 kg to the force platform and recording the COP measurement obtained each time. It is important that the load is accurately placed in the same position each time: this can be achieved by using a placement mat as described in the uniformity test procedure.

a As recommended by the Association de Française (Bizzo et al. 1985).
b Recommended from experimental evidence of the stability of a prototype force platform evaluated over a year (Browne 1999).

     Additional information about the stability of the individual transducers can be obtained if the output voltage from the ADC is accessible. The stability over time of the transducers can be determined by applying a calibrated load between 10 and 40 kg directly over each of the transducers and measuring the output voltage response of the transducers from the analogue to digital converter. It is important that the load is accurately placed in the same position each time: this can be achieved by using a placement mat as described in the uniformity test procedure.
     The Association Française de Posturologie made no recommendations for the temporal stability of COP measurements; however, upon consideration of the small magnitude of difference, 2 mm, between patients with Parkinson's disease and age-matched healthy subjects it is evident that COP measurements should also be precise to within 1 mm, in order to detect changes in subject's ability to balance (Browne 1999, Berg 1989, Hufschmidt et al. 1980).
     This test should be carried out at a minimum of monthly intervals unless the stability of the system is such that it needs to be carried out more frequently.

2.1.2. Dynamic test of the QC procedure.
     The frequency response of a force platform can be determined by using a rocking device such as a pendulum, which would simulate the sway of a subject, and measuring the frequency and amplitude of the rocking device, using the force platform's own processing and analysis program. The rocking device frequency and amplitude would need to be controllable and be able to simulate a range of frequencies from 0-20 Hz. Therefore, it could be evaluated whether the force platform reproduces all the input frequencies and at the correct amplitude.
     Additional information about the force platform's frequency response can be obtained if the output voltage from the ADC is accessible. The frequency response of the force platform can be determined by applying an impulse to the force platform (Bizzo et al. 1985, Starck et al. 1993, Hall et al. 1996). An impulse can be applied by dropping a calibrated load between 5 and 10 kg onto the top plate of the force platform. Once an impulse has been applied to the force platform the point spread function (PSF) of the system can be measured from the output voltage of the analogue to digital converter.

     This test should be carried at acceptance testing and at a minimum of yearly intervals thereafter unless the stability of the system is such that it needs to be carried out more frequently.

     A QC procedure has been described which enables users to determine whether a force platform is performing to the required specifications, as provided by the Association Française de Posturologie (Bizzo et al. 1985). The present paper is the first one, to the author's knowledge, which outlines a comprehensive QC procedure for force platforms and which recommends appropriate time intervals for carrying out each of the tests in the QC procedure. However, the time intervals suggested are only guidelines established after a year of testing a prototype force platform and, therefore, further work needs to be carried out in order to establish a baseline of time intervals for force platforms of varying ages and varying complexity.
     From the results of the QC procedure sources of error caused by the electronic components of the force platform can be identified and if the output voltage from the ADC is accessible then the exact source of the error can be identified for example excessive nonlinearity of the transducer response, different offset voltages of the transducers or noise due to the ADC or mains. Corrections can be applied to correct for excessive nonlinearity and different offset voltages in the transducers. Noise due to the ADC can be reduced by using a higher sampling frequency and then using a low pass filter, while noise from the mains can be reduced with better insulating shielding on the force platform wiring.

     In this paper a QC test procedure is outlined to allow the performance of force platforms to be evaluated and monitored over time. This QC test procedure includes details of what QC tests should be carried out; how the QC tests may be carried out; the frequency with which they should be carried out and the typical performance required in order for the force platform's measurements to be regarded as accurate, precise and giving a faithful reproduction of the frequency components of the subject's sway. Without these QC tests being carried out it is difficult for investigators to trust the results obtained and to compare results obtained for a patient undergoing physiotherapy as, the results obtained may not be accurate or precise. Table 1 is a summary of the QC protocol.

References

Andres R and Anderson D 1980 Designing a better postural measurement system Am. J. Otolaryngol. 1 197-206


Barrett R, Hyde S A and Hark W B (1987) The design of a force platform for clinical use: a feasibility study of stabilography in evaluating the effect of orthotic intervention in Duchenne muscular dystrophy J. Med. Eng. Technol. 11, 68-73.


Beppu M, Ehara Y, Monura S, Ikeda M and Takahashi S 1985 Calibration of force plate with the method of least squares Bull. Kanagawa Prefectural Rehabil. Center 10 37-46


Berg K 1989 Balance and its measure in the elderly: a review Physiother Can. 41240-5


Bizzo G, Guillet N, Patat A and Gagey P M 1985 Specifications for building a vertical force platform designed for clinical stabilornetry Med. Biol. Eng. Comput. 23 474-6


Bobbert M.F. and Schamhardt H.C. (1990) Accuracy of determining the point of force application with piezoelectric force plates J. Biomech. 23 705-10.


Browne J 1999 The development of a quantitative posturography system and its clinical evaluation MSc Thesis Trinity College Dublin


Diener H C, Dichgans J, Bacher M and Gompf B 1984 Quantification of postural sway in normals and patients with cerebellar diseases Electroencephalogr. Clin. Neurophysiol. 57 134-42


Femie G R and 1-lolliday P J 1978 Postural sway in amputees and normal subjects J. Bone Joint Surg. A 60 895-8


Gill H.S. and O'Connor J.J. (1997) A new testing rig for force platform calibration and accuracy testing. Gait Posture, 5, 228-32.

 

Goldie P A, Bach T M and Evans D M 1989 Force platform measures for evaluating postural control: reliability and validity Arch. Phys. Med. Rehab. 70 510-17


Granat M H, Kirkwood C A and Andrews B J 1990 Problem, with the use of total distance travelled and average speed as measures of postural stability Med. Biol. Eng. Comput. 28 601-2


Guerts A C H, Nienhuis B and Mulder T W 1993 Intrasubject variability of selected force-platform parameters in the quantification of postural control Arch. Phys. Med. Rehabil. 74 1144-50


Hall M.G., Fleming H.E., Dolan M.J., Millbank S.F.D. and Paul J.P. (1996) Static in situ calibration of force plates J. Biomech. 29, 659-65


Hasan S S, Robin D W, Zurkus D S, Ashmead D H, Peterson S W and Shiavi R G 1996 Simultaneous measurement of body centre of pressure and centre of gravity during upright stance: Part 1: methods Gait Posture 4 1-10


Hufschmidt A, Dichgans J, Mauritz K H and Hufschmidt M 1980 Some methods and parameters of body sway quantification and their neurological applications Arch Psychiat. Nervenkr. 228 135-50


Jeong B Y 1994 Contour representation of sway area in posturography and its application Arch. Phys. Med. Rehabil. 75951-5


Kilburn K H, Warshaw R H and Hanson B 1994 Balance measured by head and a force platform in chernically exposed and referent subjects Occupational Environmental Med. 51381-5


Lakes R S, Korttila K, Eltoft D, Derose A and Ghoneira M 1981 Instrumented force platform for postural sway studies IEEE Trans. Biomech. Eng. 28 725-9


Lucy S D et al. 1985 Postural sway profiles: normal subjects and subjects with cerebellar ataxia Physiother. Can. 37 140-8


McClelland G R 1989 Body sway and the effects of psychoactive drugs-a review Human Psychophartnacol. 4 3-14


Mita K., Kataki K.A., Itoh K. and Nogami H. (1993) An investigation of the accuracy in measuring the body centre of pressure in a standing posture with a force plate Frontiers Med. Biol. Eng. 5, 201-13.


Murray M, Seireg A A and Sepic S B 1975 Normal postural stability and steadiness: quantitative assessment J. Bone Joint Surg. A 57 510- 16


Starck J., Pyykkö I., Aalto H. and Pekkarinen J. (1993) Measurements of postural stability: development of a force platform and some excitation systems, Med. Prog. Through Technol. 18, 209-15.


Uimonen S, Laitakari K, Bloigu R, Reinila M and Sorti M 1994 Static posturography and intravenous alcohol J. Vestibular Res. 4 277-83


Winter D A 1990 Biomechanics and Motor Control ofHuman Movement 2nd edn (New York: Wiley-Interscience) 1995 Human balance and posture control during standing and walking Gait Posture 3 193-214