INTRODUCTION
METHODS
RESULTS
DISCUSSION
ACKNOWLEDGEMENT
REFERENCES
 

VARIABLE OUTPUT FOOT SENSORS TO PROVIDE PRESSURE DISTRIBUTION ON THE FOOT DURING GAIT
Jerrold S. Petrofsky, Ph.D., J.D.; Salamah Bweir, MPT MPH

From the Department of Physical Therapy, Loma Linda University, USA. Address reprint requests to: Dr. Jerrold S. Petrofsky, Professor and Director of Research, Department of Physical Therapy, School of Allied Health Professions, Loma Linda University, Loma Linda, CA 92350, United States of America, Tel: (909) 558-7274, Fax: +1 (909) 558-0481, E-mail: jerry-petrofsky@sahp.llu.edu

A new type of foot sensor was developed to analyze pressure on the foot during gait. The sensor consists of a layer of pressure sensors constructed from conductive rubber sandwiched between two layers of thin brass. Each cell, when compressed, changes its electrical resistance with pressure. These cells were placed in key positions on the bottom of the foot and used to analyze pressure distribution during the gait cycle. The hysteresis and load-resistance characteristics of the sensors were measured and were acceptable for this application. Creep was not seen as a major problem when tested over a three week period. By building shoe inserts for four different sizes for adult men, four sizes for adult women and two sizes for male and female children, a wide variety of subjects can be analyzed for pressure distribution during gait in normals and people with neurological pathologies.

Key Words: Gait, Walking, Foot, Weight Distribution, Ground Reaction

INTRODUCTION :
          Analysis of gait becomes increasingly important as the baby boomers age. As the number of people in the age range of 60-80 years old increases over the next ten years, more and more gait abnormalities will develop. These gait abnormalities can be subtle abnormalities that may not be seen easily when walking in a straight line or, may result in an inability to maintain a proper center of mass over the legs during turns. Changes from even to uneven surfaces during walking may result in falls^1,2,3. Due to osteoporosis (both disuse in geriatric) these falls can often result in hip fractures⁴. In the elderly, hip fractures can result in death if the fracture does not heal in the osteoporotic bone.
            In addition, with an ever-increasing diabetic population in the world, the influence of diabetes on gait is also becoming of great importance. Diabetes is associated with the loss of the vestibular function, vision and proprioception^5,6,7. Since all three are involved in gait, any loss of either system, let alone all three, can result in gait abnormalities. It has been suspected that these gait abnormalities cause uneven pressure distribution on the foot during gait and, secondarily, cause diabeticulcers^8,9. Diabetic foot ulcers are the most common cause of death in diabetics. Once they begin, they are nearly impossible to heal and frequently result in amputation.
          Given the above, proper analysis of gait has become increasingly important. Unfortunately, most gait systems measure gait by looking at the timing of the stride from step to step. This is accomplished by using a platform that the subject walks across or by using a shoe insert that provides foot contact on the front and back of the foot. The problem with the gait platform is that steps are usually small, and the individual has to time their steps so that they step on specific places of platform. Since turns cannot be accomplished on the platform and the platform requires proper timing, steps are usually not the same as steps the individual would normally use in gait. Therefore, this type of system generates an artificial step. Systems that have contact sensors in the shoe, do not provide pressure distribution but only timing of when the foot hits the ground and goes through each stage of gait^10,11,12. This data, while useful, may not show the true problems associated with gait in diabetic and other patients.
          Therefore, the purpose of the present investigation was to develop a shoe insert usable by children and adults that provides not only timing, but also the exact pressure distribution on the foot during gait. This might be useful for analyzing gait during daily activity¹³. Previous work has shown a significant advantage of knowing both timing and pressure distribution for gait training^14,15.

METHOD :
          A ground reaction system was built into a shoe insert. This shoe insert was a series of load cells. Each load cell was composed of two thin layers of brass (0.4 mm) separated by a layer of conductive rubber (1.0 mm). (As shown in Figure 1) The conductive rubber is made by Zoplex Corporation (ZF60). This particular conductive rubber has variable resistance from 200 meg ohms to 1 ohm when pressure is applied to the rubber. By sandwiching the rubber between two pieces of conductive brass, pressure on the brass plates causes an electrical resistance change in the conductor. While these types of schemes have been tried in the past, one major problem
 

Foot plate with 6 sensors     Single pressure sensor Figure 1. A single sensor and a foot plate with a 6 sensor array are illustrated here

 

Resistance vrs Load Figure 2. The relationship between load and electrical resistance of a typical cell with load is applied (positive adding of load) and removed (negative removal of load)

associated with most conductive rubber is reliability, repeatability and hysteresis in the resistance when load is applied and removed.
          Hysteresis is defined as the differences between the resistance of a sensor at a given load when pressure is applied compared to when a load is lightened. Most conductive rubbers presently available have an extremely large hysteresis. Testing results of this particular rubber is shown in Figure 2. This figure shows that when pressure is applied from 0 to 130 grams across the appropriate load cells, as described above, there is very little hysteresis during placing weight on the cell versus taking weight off the cell. Moreover, another problem with most conductive rubbers is that they have very little response until high loads are applied across the rubber. Also, as shown in Figure 2, there is a wide range of resistive output as a function of pressure.
          The resistive rubber that was chosen was cut to a size of 15 by 15 millimeters. The brass sheets were cut to the same size. The sandwich, consisting of the two brass sheets and the conductive rubber was held together by masking tape. Each cell, with the appropriate wires from the cell, was then placed in a foot pattern as shown in Figure 3. Six foot sensors were placed in the foot insert as follows; one under the big toe and distal base of the first metatarsal, one under the proximal base of the first metatarsal, one under the distal base of the fifth metatarsal, one under the medial tarsal bones, one under the lateral tarsal bones, and one under the heel. Experimental analysis and testing showed this to be best position to show supination, pronation, foot contact, heel contact and toe off during gait. The transducers were sealed in another layer of masking tape. The entire array, as shown in Figure 3 is less than 4 millimeters thick. One set of sensors was manufactured for each of four
 

Figure 3. Foot sensor array used here

different adult sizes for both men and women. Four sets of sensors were manufactured for each male and female child size. The total cost of system is trivial. The rubber, masking tape, electrical wiring and all materials together cost less than $10 per set of sensors.
          The sensor is part of a Wheatstone bridge where the opposing resistor was a 2000-ohm resistor. The output Wheatstone bridge was an amplifier consisting of operational amplifier signal conditioners and the output was also smoothed through a filter.

RESULTS :
          The foot sensors have been tested on 70 subjects to see how well they worked during the gait cycle. Typical results are shown in Table 1 and in Figure 4. Illustrated, here, is the weight distribution on the foot. Table 1 shows the actual measured heel-to-heel time for left and right steps, heel to toe time and the time in stance versus swing for a typical subject. The numbers listed for the six sensors are the actual weight in grams generated as peak weight during gait. As can be seen from the table and graph, this particular subject had a tendency to spend more time in the swing phase on the left leg than the right leg. The right leg, for example, had a heel-to-heel contact time of 1 second where as on the left leg, it averaged 1.2 seconds. The heel-to-toe time (the time that the foot was in contact with the floor) was 0.64 seconds on the right leg and 0.82 seconds on the left leg, giving stance to swing ratios of 32 and 36% for the left and right lower extremity respectively. The interesting phenomena on this particular subject was that not only did the left leg move slower in swing phase but the additional time spent in the stance phase on the right leg was associated with higher absolute weight on the foot and higher rates of rise in weight. For example, the maximum weight on the right heel during initial contact phase was 115.1 grams
 

Table 1. Weight distribution on the foot during gait in male subjects

on the right leg with only 85.3 grams on the left leg. The absolute pressure was not the telling point. The rate in rise of pressure was substantially higher on the right and left leg increasing to 1111.3 grams per second on the right heel where it was only 595.3 grams per second on the left heel. The same was seen for other measures as well.
 

Time / Sec Figure 4. The electrical output from the left (1) and right (2) foot sensors for the toe (bt), heel (heel) medial front foot sensor (bm) and lateral front foot sensors (bl) and 2 back foot medial and later sensors (cm and cl)

For example, on the toe, the absolute pressure was 110 grams on the right toe whereas on the left toe, it was 73.1 grams during gait. This translated to rates of rise in pressure of 343.0 grams per second on the right toe and 76.5 grams per second on the left toe. The anterior medial aspect of the foot also show substantial increasing to 123.3 grams on the left toe at a peak of 169.2 grams on the right foot, translating to 413.9 grams per second on the right leg and 292.0 grams per second change of weight on the left leg. Curiously, the posterior medial channels show very little pressure distribution on that part of the foot. It appeared that the subject not only walked with a slight limp, but he kept most of the pressure on the anterior medial aspect of the foot with reduced pressures on the posterior via aspect and anterior lateral aspect of the foot.

DISCUSSION :
          As described above, foot sensors were developed and have been preliminary tested on children and adults to analysis of the pressure distribution during gait. These sensors have shown, so far, to be reliable. One problem associated with rubber transducers is the fact that they will wear out over time. The manufacturer shows for these transducers a typical curve of 5,000,000 cycles before the rubber fatigues. With up to 5,000,000 gait cycles, the sensors could be used extensively for years without fatigue in the sensors. Over a period of three months, we found very little difference in electrical output of the sensors during repeated calibration.
          A number of different systems have been developed in recent years to measure the pressure distribution on various areas of the body, including the foot, during gait. For example, lesions to the diabetic foot have been studied with the Pedar pressure distribution shoe system¹⁶. The Pedar system divides the foot into six regions, first toe, second to fifth toes, metatarsal region, medial mid-foot, lateral mid-foot and heel. This system has proven quite useful and used by many investigators to look at pressure distribution on the foot. The only other commonly available system is a derivation of the F-scan system¹⁷. While the F-scan system has been proposed and used in several publications, it has come under attack because of creep and hysteresis in the gages¹⁸. In this particular study, the F-scan system was found to have a great deal of unreliability in that, for example, one day the pressure recorded for a given load were as much as 10 to 20 percent different from the next day. Therefore, the F-scan system was believed to be inappropriate unless calibrated on a daily basis.
Another problem with the F-scan system as seen by Sumiya et al¹⁹ is that the sensor was not uniformly even in responsiveness over its size since one area of the sensor responded differently than another area of the sensor. Errors as high as 74 percent in sensing pressure were seen across various sensors and different types of testing. While the F-scan sensor system has been improved with the use of new inks in recent years and now has less creep and less hysteresis, a common flaw of both systems is their price. Both the F-scan and the Pedar systems generally cost between $8,000 and $20,000 for a basicsystem. In spite of limitations in both systems, they have, however, proven quite good in determining pressure distribution of orthopedic shoes²⁰. comparing insole materials for pressure distribution on the foot²¹, and even pressure distribution in the diabetic foot²². Other studies have even used these technologies to determine whether abnormal gait is cause by neuropathies or sensory loss in the feet²³. While the systems seem to work well, their price is a limitation for many laboratories in times of increasing cost of research and decreasing research dollars. Therefore, the present investigation was intended to use a new technology involving a highly conductive rubber to develop much cheaper sensors that provide an electrical output over a wide range that can be sensed by conventional A/D converters on computer systems. In this respect the present system is a success. The rubber for the present system costs less than 10 dollars for a complete set of shoes. The output is fairly predictable, can be calibrated, and the creep and hysteresis is fairly small. Whereas other systems such as Pedar and F-scan do offer other advantages including an increase in the number of sensors to as high as several hundred per shoe, it is our belief from the data presented in the paper that the reliability is not better and a cost analysis does not warrant these systems for most applications. A system such as this one will provide an excellent way to analyze foot pressure distribution during gait in the future.

ACKNOWLEDGEMENT :
          This work was supported under a grant from the National Medical Test bed, Grant #0315-8872-01. This project or effort depicted was sponsored by the Department of the Army, citing the Cooperative Agreement Number (DAMD17-97-2-7016) and that the content of the information does not necessarily reflect the position or the policy of the government or NMTB, and that no official endorsement should be inferred.

REFERENCES :

1. Rancho Los Amigos: Observational Gait Analysis. 4 ed. Downey: Los Amigos Research and Education Institute, Inc., 2001

2. Smidt, Gary L: Gait in Rehabilitation. 2nd ed. Churchill Livingston. N Y. 1990

3. Wolfson LI, Whipple R, Amerman P, Kaplan j, Kleinberg A: Gait and balance in the elderly. Two functional capacities that link sensory and motor ability to falls. Clin Geriatr Med 1985;1:649-59

4. Baloh RW, Fife TD, Zweling L, Socotch T, Jacobson D, Bell T, Beykirch K: Comparison of static and dynamic posturography in young and older people. J Am Geriatr Soc 1994;42:405-12

5. Hiltunen LA: Does glucose tolerance affect elderly persons' balance, gait or muscle strength? Cent Eur J PublicHealth 2001;9:22-5

6. Oppenheim U, Kohen-Raz R, Alex D, Kohen-Raz A, Azarya M: Postural characteristics of diabetic neuropathy. Diabetes Care 1999;22:328-32

7. Cavanaugh PR, Derr JA, Ulbrecht JS, Maser RE, Orchard TJ: Problems with gait and posture in neuropathic patients with insulin-dependent diabetes mellitus. Diabet Med 1992;9:469-74

8. Center for Disease Control. Survey of complications of Diabetes. 2000

9. Darlington CL, Erasmus J, Nicholson M, King J, Smith PF: Comparison of visual--vestibular interaction in insulin-dependent and non-insulin-dependent diabetes mellitus. Neuroreport 2000;11:487-90

10. Veltink PH, Bussman HB, de Varies W, Martens WL, Van Lummel RC: Detection of static and dynamic activities using uniaxial accelerometers. IEEE Trans Rehabil Eng 1996;4:375-85

11. Veltink PH: Detection of knee unlock during stance by accelerometry. IEEE Trans Rehabil Eng 1996;4:395-402

12. Petrofsky JS: Microprocessor-based gait analysis system to retrain Trendelenburg gait. Med Biol Eng Comput 2001;39:140-3

13. Luinge HJ, Veltink PH, Baten CT: Estimating orientation with gyroscopes and accelerometers. Technol HealthCare 1999;1;7:455-9

14. Petrofsky JS: The use of biofeedback to reduce trendelenburg gait. Europ. J. Appl. Physiol. 2001;85:491-495

15. Petrofsky JS: Microprocessor-based gait analysis system to retrain Trendelenburg gait. Med Biol Eng Comput 2001;39:140-3

16. Drerup B, Hafkemeyer U, Moller M, Wetz HH: Effect of walking speed on pressure distribution of orthopedic shoe technology. Orthopade. Mar 2001;30(3):169-75. German.

17. Quesada PM, Rash GS: Quantitative assessment of simultaneous capacitive and resistive plantar pressure measurements during walking. Foot Ankle Int. 2000;21(11):928-34.

18. Quesada P, Rash G, Jarboe N: Assessment of pedar and F-Scan revisited. Clin Biomech Bristol,Avon. 1997;12(3):S15.

19. Sumiya T, Suzuki Y, Kasahara T, Ogata H: Sensing stability and dynamic response of the F-Scan in-shoe sensing system: a technical note. J Rehabil Res Dev. 1998;35(2):192-200.

20. Drerup B, Hafkemeyer U, Moller M, Wetz HH: Effect of walking speed on pressure distribution of orthopedic shoe technology. Orthopade. 2001;30(3):169-75. German.

21. Pawelka S, Kopf A, Zwick E, Bhm T, Kranzl A: Comparison of two insole materials using subjective parameters and pedobarography (pedar-system). Clin Biomech (Bristol, Avon). 1997;12(3):S6-S7.

22. Mueller MJ, Salsich GB, Bastian AJ: Differences in the gait characteristics of people with diabetes and transmetatarsal amputation compared with age-matched controls. Gait Posture 1998;7:200-206

23. Dingwell JB, Cavanagh PR: Increased variability of continuous overground walking in neuropathic patients is only indirectly related to sensory loss. Gait Posture. 2001;14(1):1-10.

Previous Article