See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/12072668

Evaluation of a sensor for low interface pressure

applications

ARTICLE in MEDICAL ENGINEERING & PHYSICS DECEMBER 2000

Impact Factor: 1.83 DOI: 10.1016/S1350-4533(00)00080-1 Source: PubMed

CITATIONS READS

57 17

3 AUTHORS, INCLUDING:

Martin Ferguson-Pell
University of Alberta
63 PUBLICATIONS 969 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Martin Ferguson-Pell
letting you access and read them immediately. Retrieved on: 21 December 2015
 Medical Engineering & Physics 22 (2000) 657663
www.elsevier.com/locate/medengphy

Technical note
Evaluation of a sensor for low interface pressure applications
a,*
Martin Ferguson-Pell , Satsue Hagisawa b, Duncan Bain a

a
Centre for Disability Research and Innovation, Institute of Orthopaedics and Musculo-Skeletal Sciences, Royal Free and University College
Medical School, Brockley Hill, Stanmore, Middlesex HA7 4LP, UK
b
School of Nursing, Oita Medical University, Hasama, Oita 879-5593, Japan

Received 16 August 2000; received in revised form 21 November 2000; accepted 7 December 2000

Abstract

An ultra-thin, small sensor has recently been developed, FlexiForce (Tekscan, Boston, MA, USA), which may be effective for
the measurement of low interface pressure between the skin, support surfaces and pressure garments.
To evaluate the suitability of the sensor for these applications, drift, repeatability, linearity, hysteresis and curvature effects were
tested under laboratory conditions.
The drift was 1.72.5%/logarithmic time, the repeatability was 2.36.6% and the linearity was 1.99.9% in the range of forces
of 1050 g applied. The hysteresis was 5.4% on average. The output offset of the sensor increased with decreasing radius of
curvature for radii less than 32 mm compared with a flat surface when no pressure was applied. The sensitivity to pressure decreased
with curvature for radii less than 32 mm.
It was found that the sensor had acceptable drift, repeatability, linearity and hysteresis. However, a significant curvature effect
was observed indicating that the sensor is suitable for direct measurement on surfaces with the radii greater than 32 mm under
static conditions. 2001 IPEM. Published by Elsevier Science Ltd. All rights reserved.

Keywords: Low interface pressure; Pressure garments; Pressure sensor calibration; Curvature effect

1. Introduction pressure, and conform to the curvature of the skin sur-

face.
External pressures applied to the surface of the skin The accuracy of interface pressure measurements
are commonly employed to reduce oedema in the relies upon sensor characteristics and measurement pro-
residual limb following amputation [1]. Similarly, press- cedures which may give rise to a variety of errors [3].
ure garments are used to prevent and reduce hyper- Specifications for the ideal sensor to measure the inter-
trophic scars following burn injury [2]. In both these face stresses associated with the skin of the human body
examples it is probable that application of a specific were proposed by Ferguson-Pell [4] and Grant [5]. Sen-
range of pressures can expedite and optimise the out- sors should be:
come. Furthermore, prescription of garment size based
on objective measurements of interface pressure is more (a) small and thin, e.g. 1 mm in thickness, 10 mm
likely to provide a consistent outcome for different in diameter, and highly flexible;
patients. (b) have a continuous output;
In order to measure the interface pressure between (c) be able to measure shear as well as normal
support surfaces, pressure garments and the skin, the forces;
sensor should be small, thin, flexible and sensitive to low (d) be free from error of measurement on curved
surfaces and from the effects of temperature
and moisture;
(e) low cost.
* Corresponding author. Tel.: +44-20-8909-5471; fax: +44-20-
8385-7151. With regard to specification (d), previous studies have
E-mail address: m.ferguson-pell@ucl.ac.uk (M. Ferguson-Pell). emphasized that curvature effects must be taken into

1350-4533/01/$ - see front matter 2001 IPEM. Published by Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 0 - 4 5 3 3 ( 0 0 ) 0 0 0 8 0 - 1
658 M. Ferguson-Pell et al. / Medical Engineering & Physics 22 (2000) 657663
account when using the transducers [47]. Bain [8]
examined this effect using a Talley pneumatic transducer
loaded by a hemispherical dome at different radii (19
87 mm) and reported that curvature had a large influence
on the output of the transducer. A similar study by Buis
and Convery [9] reported that spherical curvatures intro-
duced significant reductions in the outputs using force
sensing resistors (#9810, Tekscan, Boston, MA, USA).
In recent years new types of sensor have been
developed for the measurement of interface pressure.
Grant [5], Wytch et al. [6] and Olson [10] summarize
the characteristics of the sensors. However, until now no
sensor has met all the above specifications.
An ultra-thin, small (0.1 mm in thickness, 9.5 mm in
diameter) and flexible sensor, FlexiForce, has recently
been developed by Tekscan. The specifications of this
device indicate that it may be suitable for the measure-
ment of low interface pressure between the skin, support
surfaces and pressure garments.
2. Purpose
The aim of this study was to evaluate the FlexiForce
sensor for the measurement of interface pressure, parti-
cularly in more challenging applications such as beneath
bandages, compression stockings and pressure garments,
where pressure is relatively low.
3. Method
3.1. The FlexiForce sensor
The FlexiForce sensor employs an ink, the electrical
resistance of which varies with applied force. The active
area of the sensor is produced lithographically with con-
ductive strips connecting them to electronic conditioning
circuitry. The resistance change of the sensors varies
inversely with the applied force.
The sensor is constructed from two layers of a poly-
ester film. On each layer, a conductive material (silver)
is applied, followed by a layer of pressure-sensitive ink.
The two layers are laminated together to form the sensor.
In this study ultra-sensitive sensors operating in the
range 00.45 kg were employed. Sensors operating over
higher force ranges (11, 68, 55 kg) are available rou-
tinely but were not tested in this study. Since the inter-
face pressure reported for stump bandages and pressure
garments are relatively low, 2030 mmHg [2,11], a
pressure of 50 mmHg was used as a maximum pressure
for the evaluation. The active area of the sensor is
7.122105 m2, therefore, a force of 50 g corresponds to
a pressure of 51.6 mmHg. In this study ultra-sensitive
sensors operating in the range 00.45 kg were employed.
A small sensor-handle gathers data from the sensor,
processes it, and sends it to the computer through a serial
port. Software is supplied to control the sampling rate,
gain and displays data dynamically. The system is shown
in Fig. 1.
3.2. Test apparatus
Fig. 2 represents a loading rig to test for drift, repeat-
ability, linearity and hysteresis. A low friction tilt table
was used to apply axial load to the sensor. The sensor
was placed in a sandwich between a plastic flat surface
and a disc made from polyurethane (PU), 9 mm in diam-
eter and 3 mm in thickness, attached to another plastic
flat surface. The weight of the table was counterbalanced
using fixed weights, about an edge fulcrum. This con-
strained the applied force to a single degree of freedom,
in an arc closely approximating the vertical. Deviation
from the vertical due to the arc effect during the deflec-
tion of the assembly comprising sensor and disc was cal-
culated as being less than 0.05. Deviation from a
straight line projection was calculated as being less than
1 m. The simple edge fulcrum design provided near-
zero hysteresis due to friction, while effectively guiding
the applied weights.
Conditioning of the sensor was performed by apply-
ing a weight of 60 g, four times prior to testing, as rec-
ommended by the manufacturers. Data samples were
obtained at 8 Hz for 10 s.
Readings were taken after achieving a stable level,
mostly 1 min after loading. Room temperature was kept
at 2425C, which is close to the surface temperature of
skin of the lower limb.
3.2.1. Drift test
Drift is the change in sensor output when a constant
force is applied over a period of time. Its was tested by
applying weights of 50 and 30 g to the sensor for 2 h.
Readings were taken every 5 min for the first 20 min
then every 10 min up to 2 h.
3.2.2. Repeatability test
Repeatability is the ability of the sensor to respond in
the same way to the same applied force for successive
loadings. It was tested by applying weights of 50, 30
and 10 g to the sensor every 5 min and readings were
taken. This procedure was repeated 20 times.
3.2.3. Linearity test
Linearity is a measure of the proportionality of the
sensors response to the applied load over the range of
loading. A weight of 50 g was applied to the sensor and
readings were taken 1 min after the loading. This pro-
cedure was repeated 10 times.
Similarly, tests using weights of 40, 30, 20 and 10 g
were carried out.
 M. Ferguson-Pell et al. / Medical Engineering & Physics 22 (2000) 657663 659

Fig. 1. FlexiForce sensor system.

sensor by inflation. The apparatus used in this test is

provided diagrammatically in Fig. 3.
Readings were taken 1 min after loading with press-
ures of 0, 10, 20, 30, 40 and 50 mmHg. These procedures
were repeated five times for each pipe.

4. Data analysis

The data were saved in ASCII data format and pro-

Fig. 2. Sensor calibration rig used for drift, repeatability, linearity
and hysteresis tests.
3.2.4. Hysteresis test
Hysteresis is the difference in the sensor output
response during increased loading and decreased loading
at the same force. Its was examined by increasing and
decreasing the applied force consecutively, starting from
0 g up to 50 g, with 10-g increments and then decreasing
back to 0 g. The test was repeated 10 times.
cessed using Microsoft EXCEL version 7.0 to calcu-
late the mean and standard deviation (SD) and to rep-
resent the data graphically.
Drift was calculated by dividing the gradient of linear
regression of log-time vs. output by the starting value of
output. For repeatability and linearity, the coefficient of
variation was calculated by dividing the SD by the mean.
For the hysteresis test, the biggest difference in the read-
ings between increasing and decreasing output values at
the same applied force was taken and divided by the
maximum output reading in the range of loading. For

3.2.5. Curvature test

The curvature effect indicates whether the reading
from the sensor is affected by a curved surface. The sur-
face may have cylindrical or spherical curvature,
depending on the intended application. ABS cylindrical
pipes with different radii of curvature (8.0, 10.4, 13.6,
16.1, 32.5 and 51.7 mm) were used for the test. The
sensor was carefully taped inside of the pipe using 3M
Micropore without generating extra tension on the
sensor and a condom connected to a pressure gauge and
pump was placed in the pipe to apply the pressure to the Fig. 3. Sensor calibration device used for curvature test.
660 M. Ferguson-Pell et al. / Medical Engineering & Physics 22 (2000) 657663
the curvature test, the offset, which is the output when
the pressure is 0 mmHg, and the sensitivity to pressure,
which is represented as the gradient of the linear
regression of applied pressure vs. output, were examined
to evaluate how these are influenced by a radius of cur-
vature compared with a flat surface.
5. Results
5.1. Drift
Weights of 50 and 30 g were applied to the sensor
for 2 h. The output reading started to increase immedi-
ately upon application of the load and at 5 min had
reached a level that was 97.3% and 95.3% of the stable
values for 50 and 30 g. At 10 min the readings for 50
and 30 g reached 98.5% and 97.6% of the stable
values, respectively.
The drift for 50 and 30 g was 1.7% and
2.5%/logarithmic time, respectively.
5.2. Repeatability
When weights of 50, 30 and 10 g were applied, the
mean and SD of the readings for 20 measurements were
45.651.03, 26.740.86 and 7.280.48 g, respectively.
The coefficient of variation for 50, 30 and 10 g was
2.3%, 3.2% and 6.6%, respectively.
5.3. Linearity
Fig. 4 indicates the results of the linearity test. The
mean and SD of readings for 50, 40, 30, 20 and 10 g
were 47.190.89, 37.511.45, 27.531.16, 17.711.07
and 7.370.73 g, respectively. The coefficients of vari-
Fig. 4. The test results for linearity. Bars indicate SD.
ation at each force were 1.9%, 3.8%, 4.0%, 6.0% and
9.9%, respectively.
5.4. Hysteresis
The meanSD of the maximum difference in output
for increasing and decreasing loads was 5.42.5%. A
typical example of the data is shown in Fig. 5.
5.5. Curvature
When the pressures were applied to the curved sur-
face, the outputs of the sensor changed not only the off-
set but also the sensitivity, as indicated in Fig. 6.
The offset on a flat surface was 0 g, however, the
offset outputs with different radii increased with decreas-
ing radius. The offset vs. curvature for radii in the range
8.0100.0 mm are presented graphically in Fig. 7.
The sensitivity to pressure on a flat surface was 1.19
g/mmHg. Changes in sensitivity with curvature for radii
in the range 8.051.7 mm are illustrated in Fig. 8.
6. Discussion
According to Tekscans literature, drift is less than
3%/logarithmic time, repeatability is within 5%, linearity
is within 5%, hysteresis is less than 4.5% of full scale
when 50% of the full force range is applied. There is no
information on curvature effects.
When comparing our results with the literature, drift
was 1.72.5%/logarithmic time; repeatability was 2.3
6.6%; linearity was 1.99.9%, however, when the read-
ings were taken 10 min following loading, the coefficient
of variation became much smaller (2.34.3%); hysteresis
was 5.4% on average.
 M. Ferguson-Pell et al. / Medical Engineering & Physics 22 (2000) 657663 661

Fig. 5. Typical data for the hysteresis test. Bars indicate SD.

Fig. 6. The sensor response for surfaces with different radii of curvature. Bars indicate SD.

The drift test in this study indicated that there was an
increase in sensor output with time, achieving approxi-
mately 96% and 98% of the stable level at 5 and 10 min
following loading, respectively. Although FlexiForce
provides continuous real-time output, the values
presented immediately after loading are smaller than the
stable values from the results of the drift and linearity
tests described above. This suggests that the sensor is
best suited for measurements under static conditions.
Dynamic measurements are possible in situations where
degradation in accuracy is permissible and the compat-
ible calibration methods are used.
The repeatability, linearity and hysteresis are accept-
able. The hysteresis of the FlexiForce sensor has been
substantially improved compared with earlier Tekscan
sensors in which hysteresis was 21.7% [12]. In the lin-
earity test, the coefficient of variation was higher at
lower forces (9.9% at 10 g) compared with higher forces
(1.9% at 50 g). This may be due to the limited sensitivity
of the sensor at such low forces, suggesting that the out-
662 M. Ferguson-Pell et al. / Medical Engineering & Physics 22 (2000) 657663

Fig. 7. The effect of radius curvature on the offset output. Bars indicate SD.

Fig. 8. The effect of radius of curvature on sensitivity to pressure applied.

put obtained from a force of less than 10 g may not
be accurate.
The curvature test generated significant inaccuracies
in the output of data at small radii of curvature. When
the offset reading was plotted against radius of curva-
ture, an inverse relationship was found (Fig. 7). For cur-
vatures with a radius of less than 32.5 mm the offset
output increased with decreasing radius. From these
results the acceptable range in radius of curvature in
terms of offset would be for radii greater than approxi-
mately 32 mm. The output sensitivity to pressure was
also calculated for radii of curvature in the range of flat
to 8.0 mm and was found to be unaffected by curvatures
down to approximately 32 mm. For radii of curvature
less than approximately 32 mm the sensitivity decreased
substantially. Combining these curvature effects (offset
and sensitivity) it is apparent that the smallest radius of
curvature recommended is approximately 32 mm and
that the error is dominated by the offset effect. However,
if the offset at 0 mmHg is subtracted from all readings,
then the device can be used for radii down to 16 mm
where the sensitivity effect becomes significant.
Therefore, it was concluded that drift, repeatability,
linearity and hysteresis of the sensor used in this study
are acceptable and the sensor is suitable on flat and
curved surfaces with the radii greater than approximately
32 mm under the static conditions.
Further study is needed to evaluate the performance of
the sensor on spherical surfaces and also how the sensor
performs when its total area is not completely loaded.
 M. Ferguson-Pell et al. / Medical Engineering & Physics 22 (2000) 657663
Acknowledgements
We would like to thank Mr Peter Lennon for provid-
ing technical support for this study. This study was
funded in part by the Engineering and Physical Sciences
Research Council, JREI Programme Grant No.
GR/L86708/01, the Royal National Orthopaedic Hospi-
tal NHS Trust and ASPIRE.
References
[1] Lambert A, Johnson J. Stump shrinkers: a survey of their use.
Phys Ther 1995;81(4):2346.
[2] Mann R, Yeong EK, Moore ML, Engrav LH. A new tool to meas-
ure pressure under burn garments. J Brun Care Rehabil
1997;18(2):1603.
[3] Beebe D. Accuracy of pressure and shear measurement. In:
Webster JG, editor. Prevention of Pressure Sores. Bristol: Adams
Hilger, 1991:15574.
[4] Ferguson-Pell MW. Design criteria for the measurement of press-
ure at body/support interfaces. Eng Med 1980;9(4):20914.
663
[5] Grant LJ. Interface pressure measurement between a patient and
a support surface. CARE Br J Rehab Tissue Viability
1985;1(1):79.
[6] Wytch R, Neil G, Kalisse C. Skin-orthosis interface pressure
transducers. CARE Sci Practice 1989;7(4):1004.
[7] Veraart JCJM, Pronk G, Neumann HAM. Pressure difference of
elastic compression stockings at the ankle region. Dermatol Surg
1997;23:9359.
[8] Bain DS. Development of a phantom for the assessment of patient
support systems for the prevention of pressure sores. PhD thesis,
University of Surrey, 1997.
[9] Buis AWP, Convery P. Calibration problems encountered while
monitoring stump/socket interface pressures with force sensing
resistors: techniques adopted to minimise inaccuracies. Prosthet
Orthot Int 1997;21:17982.
[10] Olson J. Conventional pressure sensors. In: Webster JG, editor.
Prevention of pressure sores. Bristol: Adams Hilger, 1991:121
41.
[11] Isherwood PA, Robertson JC, Rossi A. Pressure measurements
beneath below-knee amputation stump bandages: elastic bandag-
ing, the Puddifoot dressing and a pneumatic bandaging technique
compared. Br J Surg 1975;62:9826.
[12] Ferguson-Pell M, Cardi M. Prototype development and compara-
tive evaluation of wheelchair pressure mapping system. Assist
Technol 1993;5:7891.