Australas Phys Eng Sci Med (2014) 37:25–35 DOI 10.1007/s13246-013-0236-7

SCIENTIFIC PAPER

Embedded sensor insole for wireless measurement of gait parameters Fernando Martı´nez-Martı´ • Marı´a Sofı´a Martı´nez-Garcı´a Santiago G. Garcı´a-Dı´az • Javier Garcı´a-Jime´nez • Alberto J. Palma • Miguel A. Carvajal



Received: 9 November 2013 / Accepted: 14 December 2013 / Published online: 31 December 2013 Ó Australasian College of Physical Scientists and Engineers in Medicine 2013

Abstract This work presents the development of a portable, wireless activity monitoring system for the estimation of biomechanical gait parameters. The system uses a pair of instrumented insoles able to measure pressure from different points of the foot including four commercial piezoresistive pressure sensors and a three-axis accelerometer, all together integrated in the insole to determine foot forces during stance and swing phases. The system includes two kinds of analysis data, one on line with a RF communications to a computer, and another off line reading the data from SD memory card. Our system has been validated and tested in different trials, extracting several features during walking for ten participants by means of the combined information from the two kinds of sensors. With the combined data from the complete set of sensors, we can obtain highly valuable information on foot movement during the non-contact period, such as supination or pronation characteristics or anomalous movement during flight time. From our preliminary results, the variation of the lateral acceleration of the foot seems to be correlated with the amount of supination. Keywords Instrumented insoles  Plantar pressure  Accelerometer  Gait  Foot

A preliminary version of this work was presented at the conference ‘‘IWBBIO 2013, Proceedings Granada, 18–20 March, 2013 [1]‘‘ F. Martı´nez-Martı´ (&)  M. S. Martı´nez-Garcı´a  S. G. Garcı´a-Dı´az  J. Garcı´a-Jime´nez  A. J. Palma  M. A. Carvajal ECsens, Departamento de Electro´nica y Tecnologı´a de Computadores, ETSIIT, Universidad de Granada, 18071 Granada, Spain e-mail: [email protected]

Introduction The study of static and dynamic plantar-pressure and footmovement patterns is an increasing requirement of many specialists in the health and sports activity fields. An accurate and reliable measurement of several motion parameters (contact and non-contact times, plantar pressure and force distributions, gait symmetry, acceleration, and orientation of foot movements) allows walking patterns to be classified as normal or pathological. Similarly, by means of specific software, these data provide quantification of other parameters such as caloric use and power as a function of contact and non-contact during walking. A system designed to study the gait motion should be able to monitor major gait events: (1) heelstrike or initial-contact: the first contact of the foot with the ground happens in the heel; (2) foot-flat: the whole foot lands on the ground; (3) mid-stance or tibia-vertical: at this moment, the entire body weight rests on one foot and the leg is extended in the vertical position; (4) heel off: the heel takes off; (5) toe off: the big toe takes off; (6) mid-swing: the foot moves in the air in the direction of movement during the swing phase or flight, and the mid-swing point happens when the tibia of the other leg is in the vertical position and the inclination of the foot changes sign. Next step begins with a new heelstrike [2]. The stance time, considered as the period of time between the heelstrike and the toe off, is an important parameter to determinate asymmetries in gait analysis [3, 4]. Traditionally, this study has been done with plantar pressure [5]. Clinical gait analysis is one of the main applications. A remarkable development in this field is a shoe-integrated sensor system for wireless gait analysis and real-time feedback, in which spatial-pressure distribution and acceleration of the foot were used for pattern

123

26

recognition and numerical analysis [6]. Plantar-pressure distribution was employed to detect gait patterns: normal gait, toe in, toe out, oversupination, and heel-walking gait abnormalities [7]. Plantar pressure was also applied in the evaluation of foot pathologies such as flat foot diagnosis [8], assessment of the diabetic foot [9], strephenopodia, and strephexopodia [10]. Its application to individuals with neurodegenerative diseases and/or in rehabilitation, postoperative, or undergoing readaptation makes it possible to provide precise, objective information about their gaits. Moreover, alterations in running or walking can thereby be detected early. The second application is related to athletic training. Frederick and Hartner [11] endeavoured to achieve their visions of optimized sports performance with thin-film pressure sensors and relatively inexpensive dataacquisition hardware. Recently, more research has taken place on athletic plantar-pressure analysis in order to improve sports achievements, such as in soccer [12, 13], tennis [14], running [15, 16], and skiing [17]. In fact, these studies have enabled a biomechanical analysis of human walking [18, 19], running, jumping, and skiing from a cinematic and temporal point of view in sports activities. Plantar-pressure distribution has also been used to determine the effectiveness of therapeutic and athletic shoes with and without viscoelastic insoles using the mean peak plantar pressure as the parameter [20]. Guidelines were given by Mueller for the application of plantar-pressure assessment in the evaluation and design of footwear for people without impairments [21], and the differences in running patterns for diverse sport shoes [22]. Moreover, data from tri-axial accelerometers mounted on the tibia were processed for estimating contact time during sprinting [23]. Nowadays, diverse plantar-pressure measurement systems are available in the marketplace and in research laboratories. There are two main types of devices based on their technical specifications and intended applications: platform systems and in-shoe systems. Platform systems are usually embedded in a footpath. A typical commercial product is EMED-SF, a ground-mounted capacitance transducer matrix platform (Novel USA, Inc., Minneapolis, MN) frequently used in clinical research for diabetes mellitus [24, 25], gait analysis [18, 26], hallux valgus [27], and so on. However, this kind of system is restricted to use in a laboratory or hospital and is used for barefoot measurements. In-shoe systems can be used to record plantarpressure distributions with a sensorized insole within a shoe. Commercial products include the F-scan measurement system (Tekscan, Inc., South Boston, USA), the Novel Pedar system (Novel USA, Inc.), and the Biofoot/ IBV (Universidad Polite´cnica de Valencia, Spain) [28], all of which capture dynamic in-shoe temporal and spatial pressure distributions that are used for dynamic gait-

123

Australas Phys Eng Sci Med (2014) 37:25–35

stability analysis [29], gait detection [30], and altered gait characteristics during running [31]. However, most of these systems use electrical wires to connect in-shoe sensors with data-acquisition systems around the waist, causing inconvenience and discomfort during strenuous exercise. Consequently, they are not suitable for long-term outdoor measurements. In these systems, hard devices have usually been used as the sensor units, but they are not comfortable and cannot last for long because of sensor fatigue and breakage. Moreover, each insole has a high number of pressure sensors ([60, typically piezoresistive sensors), entailing very high equipment cost and low-energy efficiency with a short battery life. In addition, the parts are usually expensive and fragile. The data-acquisition systems are often bulky and cannot be configured to connect with different remote receivers. Therefore, in most cases, these systems are conceived for indoor or laboratory use. Given the rapid development of different sensors for portable applications (small size and low energy consumption), apart from the above-mentioned single-type sensor gait analysis systems, in the last years more complex in-shoe systems have been reported, including additional sensors to the traditional pressure sensors. Indeed, the inclusion of accelerometers in this kind of system provides very useful information on foot movement (force and velocity) between foot-ground contacts [32, 33]. For example, Bamberg et al. [34] included an inertial measurement unit (IMU with an accelerometer, a gyroscope and a magnetometer) in order to obtain the foot orientation during swing phase as well as the heel-strike and toe-off for Parkinson patients. Sazonova et al. [35] developed a system with a three-axis accelerometer to get information about the energy consumption of patients in daily physical activities. Consequently, the availability of a low-cost, easy-to-use, portable system capable of collecting valid long-term data in gait analysis could provide important contributions to clinical practices supporting professional caregivers (for reporting and decision-making in hospitals or in outdoors environments), and to home-care practice to support patients for self-rating (for instance, in the home environment or during everyday activities). It therefore seems clear that the ideal device for ease of use by anyone, in hospital or outside it, should be low-cost, high-comfort, easy-mounting, and low-maintenance. This report focuses on the development of the prototype of a portable, wireless activity-monitoring system for the estimation of temporal gait parameters. Our aim has been to develop a compact and comfortable insole which includes all the sensors, and the analysis of the combined data from all these sensors to obtain more complete information, instead of the usual separate analysis of each kind of sensors. The system uses a pair of instrumented

Australas Phys Eng Sci Med (2014) 37:25–35

insoles able to record pressure from different points on the foot using four commercial pressure sensors. Unlike other works, the three-axis accelerometer has also been included in the insole (under the foot arc for subject comfort) to determine foot forces. The monitor includes an RF transceiver to enable real-time monitoring by computer using a software application. Similarly, the gait-activity data can be stored by the monitor and uploaded to the computer to analyze the data and generate selected reports. The system has been tested on ten participants, extracting several features during gait motion with the combined information from both types of sensors to get information about neutral, supinator and pronator participants. The opportunity to work with a real-time monitoring system will mean a breakthrough in working patterns, enabling new applications of this type of systems and new results that will surely end up being used in daily life. The remainder of the paper is organized as follows. The ’’Experimental‘‘ section provides a complete description of the measurement system, including sensor selection, insole conformation, processing electronics, firmware and software for configuration, specifications and system validation, control and display of results. The ’’Results‘‘ section discusses data acquisition and processing, showing experimental results obtained from walking. Finally, some conclusions are provided.

27

Fig. 1 Measurement system

Experimental In this section a detailed description of the instruments is provided. First, a block diagram of the system is presented and analysed, after which the firmware and communications system are briefly explained, and finally the system validation is shown. Sensors and electronics The system consists of two instrumented insoles connected to two data-processing and logging modules (Fig. 1). The sensors in the insole are powered and read-out by the respective data logger. One of the data-processing modules acts as master and the other as slave. Therefore, the master sends the measurement start and stop commands to the slave, and the data acquired by the slave are then sent to the master and stored on a lSD memory card. Additionally, the results can be sent to a computer via radio depending on the operation mode. The system can operate in two different modes: in manual mode, in which measurements are started or stopped by a button located on the master module; or in remote mode, where the computer controls the experiment and the results are received and displayed in the real-time application. In both operation modes, the

Fig. 2 Scheme of the master/slave module

results can be stored on the lSD memory card or uploaded via radio or USB. The master and slave devices are two identical modules differing only in that the master holds the memory card to store the data and the button to start/stop the measurements in manual mode. In Fig. 2, the scheme of the master and slave is displayed. These modules are run by a microcontroller, in this case, the PIC24FJ256GB106 by Microchip (USA). For wireless communication, we used the MiWiTM protocol (development by Microchip, USA) [36, 37], which is a low-cost solution for radio communications at 2.4 GHz with a maximum range of 100 m. The master, the

123

28

Australas Phys Eng Sci Med (2014) 37:25–35

Fig. 4 Response of the sensor placed at the head of the fifth metatarsal of the left insole

Fig. 3 Top and side views of the instrumented insoles

slave, and the computer (only in remote mode) constitute a wireless network configured as peer-to-peer (P2P) topology, where the master module plays the role of the coordinator. The different network nodes are provided by the radio-transceiver MRF24J40MA (Microchip, USA), connected with the respective microcontroller by SPI bus for radio communications. A complete USB-transceiver module was also designed to link the computer with the master in remote mode. This module includes a PIC18LF2550 that is connected with the computer via USB, and with the RF transceiver via SPI, providing direct communication between the computer and the master module. The instrumented insoles are made of a flexible plastic substrate composed of polyethylene terephthalate, where the sensors are affixed. In order to prevent shifting, a plastic adhesive layer covers the entire insole, anchoring the pressure sensors and the accelerometer in place. Each insole has a total of four pressure sensors located at the heads of the first and fifth metatarsals, big toe, and heel, respectively (see Fig. 3). Four sensors are considered the minimum needed for a basic plantar-pressure study in accordance with previous studies [23]. The sensors at the big toe and heel indicate the start and end of contact during locomotion. The sensors at the first and fifth metatarsals have been added to determine the lateral pressure distribution in the front part of the foot. Obviously, a greater number of sensors would report more information, but the cost of such a system would increase considerably. Two different sizes of insoles were manufactured, UK size 7 and 8.5, to study participants with different foot size. We used the Flexiforce A201 (Tekscan, USA), which is a piezoresistive sensor connected with a planar flexible wire whose sensing area is a circle with a diameter of 9.53 mm registering pressures of upto 1,500 kPa. The manufacturer guarantees at least 106 impacts, so the

123

instrumented insoles should last for 700 km, considering a minimum step length of about 70 cm. In fact, these insoles have been used for around 100 km in our combined tests, without damage or appreciable aging in any of the eight sensors. Each sensor was individually calibrated in the range of interest (upto 400 kPa for the metatarsal pressure sensors and upto 1,000 kPa for the heel and toe sensors) with a press, obtaining a non-linearity and hysteresis total error of less than 5 %. An example of the response of the sensor placed at the head of the fifth metatarsal of the left foot is plotted in Fig. 4, showing its hysteresis cycle. The pressure and conductance of the sensors were fitted to a linear regression, and the coefficients were used to calculate the pressure in our application. In addition, a three-axis accelerometer has been included at the foot arch to monitor movement during the foot flight time and the forces generated. Below, we will use the combined information of both kinds of sensors (pressure and acceleration) for the gait analysis with a very low number of sensors. It must be noted that, while the foot is in motion, vertical gravitational acceleration cannot be separated from acceleration due to linear motion, but relevant information can be extracted from the lateral acceleration components, AX and AY. Moreover, the accelerometer reports useful data about the force undergone and applied by the foot, as shown below. We selected the ADXL330 (Analog Devices, USA) due to its small package, low power consumption, and sensitivity of 330 mV/g in its three-axis measurements, where g is the gravity acceleration (9.8 m/s2). The calibrations provided by the accelerometer datasheets were considered valid, assuming that an error of 5 % in acceleration would be acceptable for this application. Nonetheless, several independent tests of acceleration measurements were carried out. The instrumented insoles were orientated to the three spatial directions and acceleration was recorded by the three axes correctly. The low-pass filtered output of the ADXL330 comprises the internal output resistance of the accelerometer of 32 kX and two external parallel capacitors.

Australas Phys Eng Sci Med (2014) 37:25–35

29

Fig. 5 The measurement system affixed to the subject, showing the coordinate axes in the right foot of the right image

The filter and bypass capacitors are mounted over a thin printed circuit board in order to disturb normal activity as little as possible (insole 5.6 mm thick in this area). The rest of the instrumented insole is 1.0 mm thick, allowing to it be included in normal shoes without modifications (see Fig. 3). The low-pass filter has a bandwidth of 30 Hz. As concerns power-management electronics, each data-logger module is powered by two AA alkaline batteries, which provide three hours of autonomy for measuring and transmitting results to the computer. The master and slave modules were attached just under the right and left knees, respectively (Fig. 5) instead of at the waist, with the aim of being suitable in certain sports such as hurdling, where they are a significant safety concern. In Fig. 5, the coordinate axes are also shown, being the Y-axis the direction of advance. Moreover, with the proposed data-logger location, the connection wires are shorter than in waist systems, notably reducing electrical interference sensitivity. The size and weight of this initial prototype are around 14.5 9 7.5 9 2 cm and 210 g per data-logger (mainly due to overweighed enclosure boxes). This undesirable drawback can be improved with lighter enclosures, allowing a 50 % reduction in weight. Communications protocol and computer software A protocol that supports the standard IEEE 802.15.4 was considered as the most suitable for our system due to a transmission distance of upto 100 m, low power consumption, and a baud-rate of upto 250 kbp, enough for our datatransfer requirements. The ZigBee protocol was originally developed to satisfy applications with these requirements, and nowadays it is the most popular solution for wireless sensor networks in numerous fields such as domotics, body sensor networks, and industry [38–42]. To reduce protocol complexity and program memory requirements in the

Fig. 6 Synchronization and data flow among master, slave, and computer

microcontroller, Microchip (USA) developed a simpler protocol, Microchip Wireless (MiWiTM), which supports upto 127 nodes and is free of charge if Microchip devices are used [36, 37]. However, since our system consists of only three nodes, P2P topology is enough to provide communications among the master, the slave, and the computer. As mentioned above, the microcontroller of the master module works as the network coordinator, and therefore it has to build the network, add new nodes to the net, and manage the channel access of the other nodes (the slave module and computer). The data flow timing is described in Fig. 6. The start measurement command can be triggered from the start/stop button (on the master) in manual mode, or it can be received from the computer in remote mode. Then, the master calculates the pressure and the acceleration from the master and slave nodes, applies the calibration curves, and stores them on the memory card.

123

30

Australas Phys Eng Sci Med (2014) 37:25–35

Fig. 7 a Graphic panel with real-time operation. b Historical format data in animation

Finally, if the system is operating in remote mode, the results are transmitted to the computer. The master starts a new measurement when the sample time, defined by the user, is completed. The system measures continually until the start/stop button is pressed again (in manual mode) or the end measurements command is received from the computer (remote mode). As Fig. 6 shows, the master waits in idle mode (the operation frequency of the microcontroller is reduced, but the peripherals continue to be active) when it is out of measurement cycles, whereas the slave unit waits in sleep mode, reducing energy consumption to a minimum. In this synchronization between master, slave, and computer, a maximum delay of 12 ms was measured. Packets are handled with acknowledgements and timeouts between the three nodes of the network (PC, master and slave). Timeouts set a maximum time for going through with the different task between the net members.

123

Fig. 8 Pressure patterns obtained from our system (solid line) and a F-scan system (dashed line) on the left foot. Heel sensor (a), 5th Metatarsal sensor (b), 1st Metatarsal sensor (c) and Big Toe sensor (d)

A software application has also been developed for controlling measurements (Fig. 7, downloading acquired data, and displaying the results. In real-time mode, the data received from the master module are displayed on the screen in real time. In animation, a previously downloaded file from the master module can be reproduced, as in realtime operation. System validation and specifications We measured the plantar pressure during gait, and compared data with those recorded using the F-scan system (Tekscan Inc., Boston, USA). To measure plantar pressure in equivalent regions of our sensors, we fixed the F-scan

Australas Phys Eng Sci Med (2014) 37:25–35

31

Table 1 Technical specifications of the measurement system Parameter

Value

Acceleration linear range (g)

3.6

Acceleration resolution (g)

0.017

Acceleration errors (%)

5

Pressure range (kPa)

1,000

Pressure resolution (kPa)

2.5

Pressure error (%)

5

Master/slave weight (with batteries) (g)

219/213

Master/slave size (cm)

14.5 9 7.5 9 2

Insole weight (g)

32

Thickness of arch insole (mm)

5.6

Thickness of rest of insole (mm) Battery time (h)

1.0 3

Wireless transmission distance, in open air (m)

60

Sampling frequency (Hz)

62

Memory capacity in a 2-GB uSD (h)

50

insoles (used adhesive tape) to our insoles. Each FlexiForce sensor covered the area between four or six sensing points on the F-scan insoles. The average of the pressure values for all the F-scan sensing points was calculated and compared to the corresponding FlexiForce sensor value. The variations in the plantar pressure were measured simultaneously by our system and the F-scan system for three participants. Typical plantar pressure curves obtained from both devices are shown on the same graph in Fig. 8. The solid line is the result of our device and the dashed line is the F-scan result. The differences in peak pressure ranged from -1.8 to 27.9 kPa, where the maximum difference was in big toe sensor. We can observe a very good agreement between both systems with an error below 8.1 % pressure. Moreover, from the dynamic point of view, the average delay between the signal from our system and F-scan system was below 30 ms. Therefore, the developed system was confirmed to provide a quantitative estimation of plantar pressure. The main technical specifications of this initial prototype are summarized in Table 1. In the present stage of system development, the sampling frequency (62 Hz) obtained allowed the system to be used only for walking analysis [43, 44].

Results and discussions First, tests were carried out in order to assess system comfort and usability. In trials, the instrumented insole was placed on top of and under the sports shoes’ insole. It was concluded that the instrumented insole must be under the insole of the sports shoes for maximum comfort. It must be

taken into account that the pressure was recorded under the shoe insole. Indeed, the measured pressure depends on the absorption of the sole of the shoes, however the accuracy and the time response of the system remained unchanged. The absorption coefficient of the used shoe insoles was 1.19, calculated as the ratio of the average pressure registered without and with shoe insole. This coefficient was also validated with the commercial system F-scan. Therefore, the pressure values have to be considered as dependent on the shoes used in the study. In addition, it was found that the foot slid less over the shoe insole than directly over the plastic cover of the instrumented insole. The instrumented insole does not significantly disturb locomotion patterns because the thickest area, the board with the accelerometer, is located under the foot arch; as a consequence, no modification of the shoes’ insoles was required. For the case of participant with low or collapsed arches, the thickness of the shoe insole could be reduced in the arch to increase comfort. Nevertheless, one participant with collapsed arches did not feel any discomfort while using the instrumented insole under the shoe insoles. Pressure and acceleration were recorded during 5 min walking for a total of ten subjects. The participants walked freely on a flat surface without any impositions concerning velocity or other parameters in order to study the participants’ natural gait. The characteristics of the participants are summarized in Table 2. By previous biomechanical and medical tests, four of the subjects were diagnosed as walkers who supinate slightly, one pronates, and the rest were considered neutral. In addition, participant 2 has a claw foot and number 9 has collapsed arches. These characteristics are evident in the results, as compared to a normal walker as shown below. The results from the experiment of participant 1 are plotted in Fig. 9a, where AX, AY and AZ represent the acceleration components. The major events in gait can be observed. The step starts with the heelstrike, during which acceleration in the direction of movement, AY, reduces to its minimum value, and pressure in the center of the heel starts to increase [3, 43, 44]. The maximum pressure value in the center of the heel is recorded during the flat-foot stage, when the foot reaches horizontal and the fifth metatarsal touches the ground. At the beginning of the experiment, when the foot is totally resting on the ground, the acceleration components are constant, and the inclination of the accelerometer can be zeroed. Then, the midstage starts, and pressure decreases in the rear of the foot and increases in the front due to the mass center movement of the subject. The mid stage phase ends when the foot acceleration changes, as the Fig. 9a shows. During the mid stage the pressure in the heel drops down to zero; and ends with the heel-off. The acceleration in Z-axis changes due to heel elevation movement, as Fig. 9a shows. During the

123

32

Australas Phys Eng Sci Med (2014) 37:25–35

Table 2 Participant characteristics Participant

Sex

Mass (kg)

Height (m)

Age (years)

Body mass index

Foot type

1

Male

76

1.82

30

22.9

Neutral

2

Male

80

1.81

33

24.4

Supinated, claw feet

3

Male

75

1.78

27

23.7

Supinated

4

Male

94

1.83

35

28.1

Neutral

5

Male

73

1.77

41

23.8

Supinated

6

Male

72

1.71

23

24.6

Neutral

7

Male

73

1.76

18

23.6

Over-supinated

8

Male

79

1.85

38

23.1

Neutral

9

Female

61

1.60

17

23.8

Supinated, collapsed arches

10

Female

54

1.65

17

19.8

Pronated

Fig. 9 Right plantar pressure and accelerations during walking: participant 1 (a), participant 2 (b)

heel-off the sensor at the fifth metatarsal reaches a maximum, and after falls to zero when the pressure in the big toe starts to increase. The contact of the foot with the ground, the stance phase, finishes with the big toe take off (toe-off). Finally, the foot advances through the air in the swing phase, or foot flight, until the next heel contact. During the swing phase, the foot inclination changes, and the point where AY is equal to zero can be considered the mid-swing. The set of four pressure sensors could be enough to detect certain anomalous gait parameters. As an example,

123

Fig. 9b shows the results of participant 2, who presented a slight claw foot. During the heel-off phase, the pressure increases in the first and fifth metatarsal, but when the first metatarsal makes contact with the ground, pressure in the big toe is reduced for a moment, and then increases once again at the end of contact with the ground. This fact is accentuated in subjects with a slight claw foot, such as participant 2. In addition, the maximum pressure in the heel is slightly reduced for a moment in participant 2, just when pressure increases at the fifth metatarsal, due to the claw foot. In both cases (Fig. 9a, b), the acceleration in the Y direction (direction of participant movement) is negative during landing due to the braking of the foot, and AY reaches values of upto -5 g, exceeding the linear range of the sensor (3.6 g). This fact indicates that the accelerometer should be replaced by another with a higher range in this direction. Once the heel is on the ground, the time the Z acceleration component takes to reach a resting value will be longer in shoes with higher absorption. However, the minimum of AZ component marks the exact moment when the heels impacts on the floor, as Fig. 9 shows. The maximum of pressure happens some milliseconds later than the minimum of AY due to the pressure sensor is placed in the center of the heel and not on the edge. Observing the X acceleration component at the moment of the impact (heel strike), the inclination of the foot can be estimated and then, we can know the orientation of the foot with respect to the rest position (when the foot is resting on the floor). If the AX of participant 1 and 2 (Fig. 9a, b, respectively) is compared at the heel strike moment, we can see that the maximum for the participant 1 is higher than the participant 2. Finding the changes of AX (DAX ) with respect to the rest positions, the Fig. 10a, b, c are obtained patient 1, 2 and 7. These patients were diagnosed as neutral, supinated and over-supinated foot, respectively, by previous biomechanical tests (see Table 2). As it can be observed in Fig. 10, during the swing phase, the foot of over-supinator suffers more changes of DAX than supinator or neutral foot.

Australas Phys Eng Sci Med (2014) 37:25–35

33 Table 3 hjDAX ji of right and left feet of participants Participant

hjDAX jiðgÞ Right

Left

Average

Symmetry (%)

1

0.291

0.287

0.289

101

2

0.375

0.302

0.339

124

3

0.434

0.371

0.403

117

4

0.261

0.277

0.269

94

5

0.389

0.302

0.346

129

6

0.236

0.190

0.213

124

7 8

0.512 0.256

0.399 0.263

0.455 0.260

128 97

9

0.354

0.368

0.361

96

10

0.272

0.283

0.278

96

Fig. 11 AY of the left foot of participant 4 over ten steps

Fig. 10 DAX of the right foot: participant 1 (a), participant 2 (b), and participant 7 (c)

The experimental data presented in Fig. 10, can be justified as follows: During the swing phase, the foot can move in the X direction (normal to advance direction) and this movement can be quantified by analysing the data provided by the accelerometer. The AX increment can be found, zeroing the acceleration components in the rest time just before starting the test, when the participant is motionless. The temporal evolutions of the AX increments are plotted in Fig. 10 for participants 1, 2 and 7, the latter diagnosed as a walker who supinates considerably (oversupinated foot, 2). The temporal average of the absolute value of AX, h|AX(t)|i , indicates how much the foot moves

in the X direction, and this parameter was calculated over 60 s for each participant. In our study, supinator walkers show a higher value of h|AX|i, as Table 3 shows. Once the participants are classified as pronator, supinator or neutral, the hjDAX ji could provide the amount of pronation or supination. And in fact, the results of participant 7, who supinates his right foot considerably, exceed the rest of the participants. However, participant 6 moves his feet in the X-direction less than the other participants; therefore, this subject will lose less energy due to lateral movements during walking. The symmetry of the two feet has been calculated as the rate between the hjDAX ji for the right and left feet. Significant asymmetries can be found in several participants due to muscular or bone imbalances. The acceleration component in the direction of movement can also provide useful information for finding certain temporal gait parameters. As mentioned above (see Fig. 11), the gait cycle starts with the heelstrike, which involves a minimum in AY; and the cycle finishes with the toe-off with a maximum in AY. Therefore, the stance and swing times can be found from AY by finding the maxima and minima, as Fig. 11 shows. The stance time and time

123

34

Australas Phys Eng Sci Med (2014) 37:25–35

Table 4 Temporal parameters in gait test for the ten participants Right foot

Left foot

Time per step (s)

Stance time (s)

Swing time (s)

Mean

SD

Mean

SD

Mean

SD

1

1.030

0.009

0.685

0.017

0.346

0.011

2

1.058

0.028

0.712

0.017

0.346

0.022

3 4

1.048 1.097

0.027 0.015

0.717 0.745

0.018 0.014

0.331 0.352

5

1.086

0.010

0.712

0.012

6

1.117

0.017

0.758

0.015

7

1.023

0.013

0.686

8

1.201

0.020

9

1.052

0.016

10

1.113

Global

1.082

Duty(%)

Time per step (s)

Stance time (s)

Swing time (s)

Mean

SD

Mean

SD

Mean

SD

66.4

1.030

0.011

0.675

0.016

0.356

0.015

65.5

67.3

1.056

0.023

0.706

0.016

0.350

0.014

66.8

0.020 0.006

68.4 67.9

1.050 1.114

0.013 0.040

0.716 0.716

0.016 0.029

0.334 0.397

0.008 0.025

68.2 64.3

0.373

0.014

65.6

1.085

0.022

0.701

0.019

0.385

0.015

64.5

0.359

0.016

67.8

1.112

0.018

0.742

0.008

0.369

0.019

66.8

0.014

0.337

0.011

67.1

1.025

0.012

0.679

0.019

0.347

0.014

66.2

0.791

0.019

0.410

0.017

65.9

1.201

0.030

0.835

0.020

0.366

0.015

69.5

0.713

0.013

0.339

0.013

67.8

1.053

0.016

0.723

0.011

0.330

0.008

68.7

0.021

0.754

0.012

0.359

0.019

67.7

1.097

0.022

0.710

0.014

0.387

0.013

64.7

0.053

0.727

0.034

0.355

0.023

67.2

1.082

0.052

0.720

0.045

0.362

0.023

66.5

per step have been calculated over 60 seconds, and the average and the standard deviations are summarized in Table 4. The period of a step is practically equal for the different participants when they walked freely, and the averaged cadence is (0.92 ± 0.05) Hz, very similar for all the subjects. In addition, the ratio between the stance time and the time per step (duty) has been calculated, resulting in around 2/3 for most of the participants walking freely without restrictions which is in agreement with the literature [45]. The mean stance time for the set of participants (global) is (0.727 ± 0.034) s.

Conclusions The system presented in this work can simultaneously provide the pressure values for the big toe, first and fifth metatarsals, heel and the acceleration components in the three spatial axes including all the sensors in the instrumented insole. The data analysis can be carried out with a software program or with a standard spreadsheet. The software analysis shows the results in real time on the computer screen or in replay mode for experiment analysis. These display modes make immediate feedback possible, and therefore the subject gets immediate feedback on his or her pathology and can try to correct it at once. Moreover, the accelerometer readouts report additional information during swing and stance time in walking tests. For instance, detection of anomalous pressure distribution, excess lateral movement, determination of stance time, and the detection of asymmetry could prevent future injuries or help a person choose the most suitable shoes. In this work, we have combined the analysis of the pressure and acceleration data to clearly distinguish among neutral, pronator or supinator participants, even estimating different amount of supination. This

123

Duty (%)

system could be very useful in studying walking patterns in people affected by illnesses such as Parkinsons, astronauts after a weightless period, rehabilitation from different injuries, or to choose the most appropriate shoes. Acknowledgements The authors would like to thank Jose´ L. Gonza´lez Montesinos (Associate professor in Physical and Sports Activity) for his tips on fixing the pressure sensor in the instrumented insole and Carlos J. Carvajal Rodrı´guez (M.Sc. in Sports Science) for collaborating in data analysis. This work was funded by Junta de Andalucı´a (Spain), under Project P10-TIC5997. This project was partially supported by European Regional Development Funds (ERDF).

References 1. Martı´nez-Martı´ F, Garcı´a-Dı´az SG, Garcı´a-Jime´nez J, Martı´nezGarcı´a MS, Martı´nez-Olmos A, Carvajal MA (2013) IWBBIO 2013 Proceedings of the international work-conference on bioinformatics and biomedical engineering, pp 657–665 2. Whittle MW (2003) Gait analysis: an introduction. ButterworthHeinemann, Oxford 3. Groot A, Decker R, Reed K (2009) Third joint eurohaptics conference and symposium on haptic interfaces, pp 190–195 4. Johanson M, Cooksey A, Hillier C, Kobbeman H, Stambaugh A (2006) Heel lifts and the stance phase of gait in subjects with limited ankle dorsiflexion. J Athl Train 2:159 5. Urry S (1999) Plantar pressure-measurement sensors. Measurement Sci Technol 10(1):R16 6. Morris SJ, Paradiso JA (2002) Proceedings of the second joint engineering in medicine and biology 24th annual conference and the annual fall meeting of the biomedical engineering society EMBS/BMES Conference vol 3, pp 2468–2469 7. Chen M, Huang B, Xu Y (2008) Proceedings IEEE international conference on robotics and automation, pp 2019–2024 8. Tareco J, Miller N, MacWilliams B, Michelson J (1999) Foot & ankle international / American orthopaedic foot and ankle society [and] Swiss foot and ankle society 20(1071–1007):456 9. Mueller M, Hastings M, Commean P, Smith K, Pilgram T, Robertson D, Johnson J (2003) Forefoot structural predictors of

Australas Phys Eng Sci Med (2014) 37:25–35

10.

11. 12.

13.

14.

15.

16.

17. 18.

19. 20.

21.

22. 23. 24.

25.

26.

plantar pressures during walking in people with diabetes and peripheral neuropathy. J Biomech 36(7):1009 Hodgson B, Tis L, Cobb S, McCarthy S, Higbie E (2006) The effect of two different custom-molded corrective orthotics on plantar pressure. Sport Rehabil 1:33 Frederick E, Hartner K (1993) The evolution of foot pressure measurements. Sens Mag 6:30 Gioftsidou A, Malliou P, Pafis G, Beneka A, Godolias G, Maganaris C (2006) The effects of soccer training and timing of balance training on balance ability. Eur Appl Physiol 6:659 Wong Pl, Chamari K, Chaouachi A, Mao DW, Wislff U, Hong Y (2007) Difference in plantar pressure between the preferred and non-preferred feet in four soccer-related movements. Br J Sports Med 41(2):84 Girard O, Eicher F, Fourchet F, Micallef J, Millet G (2007) Effects of the playing surface on plantar pressures and potential injuries in tennis. Br Sports Med 41:733 Queen RM, Haynes BB, Hardaker WM, Garrett WE (2007) Forefoot loading during three athletic tasks. Am J Sports Med 35(4):630 Kuntze G, Pias M, Bezodis I, Kerwin D, Coulouris G, Irwin G (2009) in 27th International Conference On Biomechanics in Sports Michahelles F, Schiele B (2005) Sensing and monitoring professional skiers. IEEE Pervasive Comput 4(3):40 Saito M, Nakajima K, Takano C, Ohta Y, Sugimoto C, Ezoe R, Sasaki K, Hosaka H, Ifukube T, Ino S, Yamashita K (2011) An in-shoe device to measure plantar pressure during daily human activity. Medl Eng Phys 33(5):638 Dugan S, Bhat K (2005) Biomechanics and analysis of running gait. Med Rehabil Clin N Am 16:603 Lavery LA, Vela SA, Fleischli JG, Armstrong DG, Lavery DC (1997) Reducing plantar pressure in the neuropathic foot: acomparison of footwear. Diabetes Care 20(11):1706 Mueller M (1999) Application of plantar pressure assessment in footwear and insert design. J orthop sports phys ther 29(01906011):747 Cheung R, Ng G (2008) Influence of dIfferent footwear on force of landing during running. Phys Ther 88:620 Purcell B, Channells J, James D, Barret R (2006) Proceedings of SPIE: microelectronics, MEMS, and nanotechnology Stess RM, Jensen SR, Mirmiran R (1997) Reducing plantar pressure in the neuropathic foot: a comparison of footwear. Diabetes Care 20(5):855 McPoil TG, Yamada W, Smith W, Cornwall M (2001) The distribution of plantar pressures in American Indians with diabetes mellitus. J Am Podiatr Med Assoc 91(6):280 Kanatli U, Yetkin H, Bolukbasi S (2003) Evaluation of the transverse metatarsal arch of the foot with gait analysis. Arch Orthop Trauma Surg 123:148

35 27. Kernozek T, Sterriker S (2002) Chevron (Austin) distal metatarsal osteotomy for hallux valgus: comparison of pre- and postsurgical characteristics. Foot Ankle Int 23:503 28. Martı´nez-Nova A, Cuevas-Garcı´a JC, Pascual-Huerta J, Sa´nchezRodrı´guez R (2007) BioFoot in-shoe system: normal values and assessment of the reliability and repeatability. The Foot 17(4):190 29. Lemaire ED, Biswas A, Kofinan J (2006) Proceedings 28th annual international conference of the IEEE engineering in medicine and biology society EMBS ’06, pp 4465–4468 30. Catalfamo P, Moser D, Ghoussayni S, Edwins D (2008) Detections of gait events using and F-scan in-shoe pressure measurement system. Gait Posture 3:420 31. Kong P, Heer HD (2009) Gait & Posture 29:143 32. Kirtley C (2001) Proceedings of the 5th symposium on footwear biomechanics, Zuerich/Switzerland 33. Paulick P, Djalilian H, Bachman M (2011) . In: Duffy V (ed.) Digital human modeling, lecture notes in computer science, vol. 6777. Springer, Berlin Heidelberg, pp. 171–177 34. Bamberg S, Benbasat AY, Scarborough DM, Krebs DE, Paradiso JA (2008) Gait analysis using a shoe-integrated wireless sensor system. IEEE Trans Inf Technol Biomed 12(4):413 35. Sazonova NA, Browning R, Sazonov ES (2011) Prediction of bodyweight and energy expenditure using point pressure and foot acceleration measurements. Open Biomed Eng J 5:110 36. Microchip. Application note an1284, microchip wireless (miwi) application programming interface - miapp 37. Microchip. Application note an1283, microchip wireless (miwi) media access controller-mimac 38. Baronti P, Pillai P, Chook V, Chessa S, Gotta A, Hu Y (2007) Wireless sensor networks: a survey on the state of the art and the 802.15.4 and ZigBee standards. Comput Commun 30:1655 39. Jovanov E, Milenkovic A, Otto C, de Groen P (2005) A wireless body area network of intelligent motion sensors for computer assisted physical rehabilitation. J NeuroEng Rehabil 2(1):6 40. Milenkovic A, Otto C, Jovanov E (2006) Wireless sensor networks for personal health monitoring: Issues and an implementation. Comput Commun 29(13–14):2521 41. Wheeler A (2007) Commercial applications of wireless sensor networks using ZigBee. Commun Mag IEEE 45(4):70 42. Wang N, Zhang N, Wang M (2006) Wireless sensors in agriculture and food industryaˆ€’’Recent development and future perspective. Comput Electron Agric 50(1):1 43. Hausdorff JM, Ladin Z, Wei J (1995) Footswitch system for measurement of the temporal parameters of gait. Biomechanics 3:347 44. Karnik T, Kralj A (1997) Simple gait assessment system. Gait and Posture 6:193–199 45. Ounpuu S (1994) The biomechanics of walking and running. Clin Sports Med 13(0278–5919):843

123

Embedded sensor insole for wireless measurement of gait parameters.

This work presents the development of a portable, wireless activity monitoring system for the estimation of biomechanical gait parameters. The system ...
2MB Sizes 2 Downloads 0 Views