DEVELOPMENT OF AN ALTERNATIVE SYSTEM

FOR SUSPENDED GAIT ANALYSIS

Gustavo Freitas de Lima and Alberto Cliquet Jr.

Electrical Engineering Department, University of São Paulo, Av. Trabalhador São-carlense, 400, São Carlos-SP, Brazil

Department of Orthopedics, State University of Campinas, R. Tessalia Vieira de Camargo, 126, Campinas-SP, Brazil

Keywords: Functional Electrical Stimulation, suspended gait, ground reaction forces, knee joint angles, paraplegic,

rehabilitation, Spinal Cord Injury.

Abstract: Spinal Cord Injury (SCI) may impair an individual’s gait. For these cases, a rehabilitation technique that has

become more popular is Functional Electrical Stimulation (FES). Gait analysis is an important technique to

evaluate rehabilitation of patients undergoing FES-assisted therapy. This work proposes a system that

monitors gait variables – knee joint angles, and ground reaction forces (heel and metatarsal) – and uses them

as inputs for gait analysis of paraplegic patients. The methods for building the data acquisition hardware

(transducers and interface) and software are described, along with the transducer calibration methods. The

results show the final prototype for the gait analysis system, which allows comparison between different

individuals’ gaits, as well as different rehabilitation stages for the same individual. The software has a

recording feature, as well as digital control outputs, which may be used in the future for training an

Artificial Neural Network (ANN) and controlling the individual’s FES stimulator. In the near future, the

system may be of great applicability for suspended FES-assisted gait analysis and control.

1 INTRODUCTION

1.1 Human Gait

Gait may be defined as a form of biped progression

in which lower limb repetitive movements include

periods of double support – in which both feet are in

contact with the ground – followed by periods in

which only one foot supports the body (stance) and

the other is being moved above the ground (swing)

(Wall, 1999).

On a normal gait, the stance phase constitutes

60% of the gait cycle, and is defined as the interval

in which the reference foot is in contact with the

ground. The swing phase begins with heel contact

and ends when the foot leaves the ground (toe off

surface).

1.2 Gait Analysis

For individuals that suffer Spinal Cord Injury (SCI),

a technique that has contributed for rehabilitation is

Functional Electrical Stimulation (FES) (Castro and

Cliquet Jr., 2000). FES treatment may be associated

with dynamic suspension (Field-Fote, 2001). This

suspension allows a weight reduction, maintaining

the load on the lower limbs at a level they are able to

stand. It also stabilizes the trunk, resulting in better

balance for the patient, and lowers upper-limb

overload, frequently observed on walker-aided

training for paraplegic patients.

Gait analysis is an important tool for

biomechanical studies of the rehabilitation process.

Veltink, Liedtke, Droog, and van der Kooij (2005)

developed a gait analysis system using two

commercial six-degrees-of-freedom force and

moment sensors under a sandal. Giacomozzi and

Macellari (1997) constructed a compound

instrument by superimposing a dedicated pressure

platform on a commercial force platform.

The objective of this work is the development of

an alternative system for gait analysis, to be used on

the evaluation of patients’ rehabilitation for

suspended FES-assisted gait.

260

Freitas de Lima G. and Cliquet Jr. A. (2008).

DEVELOPMENT OF AN ALTERNATIVE SYSTEM FOR SUSPENDED GAIT ANALYSIS.

In Proceedings of the First International Conference on Biomedical Electronics and Devices, pages 260-263

DOI: 10.5220/0001049702600263

 SciTePress

2 METHODS

2.1 Determination of Variables

The first task was to determine which variables

would serve as inputs for the system. Various works

have been done proposing different models (Tong

and Granat, 1999; Pappas et al., 2004; Popovic et al.,

1998), each one showing advantages and

disadvantages. For this work, the idea was to be able

to measure the actual values for ground reaction

forces, and use them for comparison on gait

analysis, not as triggers for detection of foot contact.

Popovic et al. (1998) stated that ground reaction

forces alone could not be used for satisfactory gait

phase characterizing. In order to avoid this problem,

an extra variable was chosen to serve as system

input. The data posted on the CGA Normative Gait

Database (Kirtley, 2006) show kinematic and kinetic

analysis for healthy adults and children. Taking in

consideration the movement amplitude, easiness to

mount sensors, sensor stability during movement

and amount of signal noise, the variable selected was

knee flexion/extension.

2.2 Transducer Selection

Based on the studies conducted by Cunha (1999),

the chosen transducers for knee flexion/extension

angle determination were shape sensors. These

sensors use 0.25 mm diameter fiber optics, specially

treated to lose light by refraction proportionally to

the deflection suffered by the fibers. They present

some important characteristics, such as light weight

and a simple electronic signal processing package,

incorporated to the fiber optics.

For the ground reaction forces, the analysis was

based on the studies conducted by Leite (2003). The

author developed an instrumented crutch to measure

vertical reaction forces applied by patients during

gait, using strain gages.

Strain gages are resistors composed of a very

thin conductive layer over an isolating compound.

The sensor is glued on a structure, and when there is

some deformation caused by applied forces on the

structure, it is possible to determine the value of the

force, since it depends only on the type of material

and the geometry of the structure.

The fact that strain gages require a rigid structure

to operate represents a disadvantage for this kind of

sensors. The solution found for the problem was to

construct cylindrical aluminum rings to work as load

cells for the strain gages. For each load cell, a metal

base and a semi-spherical top were constructed, so

that the top’s radius helps bring the applied loads as

close as possible to vertical. Figure 1 shows one of

the load cells with the strain gages attached (left)

and with the protective covering, metal base and top

(right).

Figure 1: Ground reaction force sensors (load cells).

Three load cells were made for each foot, and

they were attached to a sandal, mounted on

cylindrical metal plates. One load cell was

positioned in the heel area, and the other two were

positioned in the metatarsal area, connected in

parallel – in order to compensate different styles of

stepping patterns.

2.3 Hardware Interface

The hardware interface circuit consists of anti-alias

filters (cutoff frequency of 18 Hz), voltage

regulators (± 5VDC), and amplifiers with adjustable

gains for the strain gages. The information was

acquired using a data acquisition board.

The circuit is enclosed in a box, connected to the

data acquisition board, and also to a smaller

interface box, which is positioned around the

patient’s waist, and has connection hubs for the

instrumented sandals and shape sensors. The

transducers connected to the patient’s interface box

are shown in Figure 2.

Figure 2: Transducers and patient’s interface.

2.4 Transducer Calibration

To calibrate the shape sensors, they were attached to

two articulated flat metal bars, and a protractor. The

voltage values corresponding to each 5º angle

interval, from -140º to 140º, were collected.

DEVELOPMENT OF AN ALTERNATIVE SYSTEM FOR SUSPENDED GAIT ANALYSIS

261

For the load cells, the calibration was performed

using a dynamometer, and the voltage values for

each 50 N interval, from 0 to 1000 N were collected.

For both cases, the results showed a linear behavior.

2.5 Software Interface

The software used to program the system was

LabVIEW 6.1. It has some advantages, including

user-friendly graphic interface, and compatibility

with the data acquisition board.

The interface software has a monitoring module,

which allows visually following the behavior of

knee flexion/extension angles, and ground reaction

forces – heel and metatarsal. It is also possible to

record them in a spreadsheet file for data analysis.

The visual interface shows three screens (selectable

using tabs): the “Main” screen, which has all the

controls and configuration options; the “Angles”

screen, which shows the graphics for the knee

flexion/extension angles, and the “Forces” screen,

which shows the graphics for the ground reaction

forces. Figure 3 illustrates the “Forces” screen.

Figure 3: Graphical interface for force measurements.

On the main screen, there is a calibration

function. This function should be used when the

patient already has the sandals on, but has not stood

up yet. The function performs 200 force acquisitions

for each load cell, and then calculates the average of

these forces. The result is then used as linear

coefficients for the calibration curves of the force

transducers. This way, the value shown with the

patient sitting is zero, despite the applied force for

tightening the sandals.

3 RESULTS

The final prototype was mounted on a healthy

individual and tested, to check its functionality. The

individual’s characteristics are: male, 25 years old,

height of 1.81m, weight of 69.0kg. Figure 4 shows

the subject wearing the instrumented sandals and the

shape sensors.

Figure 4: (A) System mounted on healthy individual; (B)

Detail of sensor positioning in the sandal; (C)

Instrumented sandal.

Figure 5: Ground reaction forces on right foot for healthy

individual.

Figure 6: Knee joint angles for healthy individual.

The individual was asked to walk normally in a

straight line for about 15m while the values of

ground reaction forces and knee joint angles were

recorded. Figures 5 and 6 show the resulting values

for the ground reaction forces on the left foot, and

the knee joint angles, respectively.

For the angle values, the convention adopted was

zero for straight vertical position, positive angles for

knee flexion and negative angles for knee extension.

BIODEVICES 2008 - International Conference on Biomedical Electronics and Devices

262

4 DISCUSSION

The ground reaction forces observed in Figure 5

show peaks of about 780N on the heel, which

correspond to 113% of the individual’s weight. On

the metatarsal area, the peak force values are around

380N (55% of the individual’s weight). This may be

due to the softness of the sandal sole, which may

still absorb part of the applied forces. It’s possible to

distinguish the gait phases of initial contact (heel

force peak), mid stance (heel and metatarsal force

intersection), and terminal stance (metatarsal force

peak).

The knee joint angles observed in Figure 5

follow the pattern presented on the CGA Normative

Gait Database (Kirtley, 2006). The waveforms

present a repetitive pattern, confirming that the

shape sensors did not move during the acquisitions.

5 CONCLUSIONS

The values of ground reaction forces and knee joint

angles may be observed during the gait, and the

recorded values may be used for further analysis,

comparing different styles of gait, or different

rehabilitation stages for the same individual.

This system may be used as an alternative to the

force platforms. The disadvantages are that it

requires some time for donning, and it can only

measure vertical forces. But it presents some

advantages, such as: the subject may walk freely

(within the limitation of the cables), and does not

have to step exactly on the load cell, resulting in a

more natural gait; also, the system allows

monitoring two critical force points for each foot,

and not just the resulting force.

Considering the aforementioned advantages, an

important possible application for this alternative

system is suspended FES-assisted gait. In this case,

therapists may follow the recovery of patients

undergoing this kind of treatment by analyzing the

gait on different stages of the rehabilitation process.

In the future, the recordings of gait sessions may be

used as inputs for a closed-loop FES control. The

system already has two digital inputs and two

outputs, which may be used to trigger an electrical

stimulator. Since the software is open, an Artificial

Neural Network (ANN) may be programmed to

control the FES during gait, using the patient’s own

recorded data for training. With this implementation,

the patient will not need to trigger the stimulation

manually, and may direct all the attention to the

walking activity.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the support

of the State of São Paulo Foundation for Research –

FAPESP and the Coordination for the Improvement

of High Education Personnel – CAPES – Ministry of

Education, Brazil.

REFERENCES

Castro, M, C. F.; Cliquet Jr., A. An artificial grasping

system for the paralysed hand. Artificial Organs.

2000;24:185-188.

CGA Normative Gait Database. Produced by Chris

Kirtley. [cited 2007 Jul 10]. Available from:

<http://www.univie.ac.at/cga/data/index.html>.

Cunha, F. L. Obtenção e uso dos acoplamentos

cinemáticos interfalangianos e interdigitais no projeto

de próteses antropomórficas para membros

superiores. Vitória: UFES; 1999.

Field-Fote, E. C. Combined use of body weight support,

functional eletric stimulation, and treadmill training to

improve walking ability in individuals with chronic

incomplete spinal cord injury. Arch Phys Med Rehabil.

2001;82: 818-824.

Giacomozzi, C.; Macellari, V. Piezo-Dynamometric

Platform for a More Complete Analysis of Foot-to-

Floor Interaction. IEEE Transactions on

Rehabilitation Engineering. 1997;5: 322-330.

Leite, F. I. L. Desenvolvimento de uma muleta

instrumentalizada para fins de acompanhamento

clínico. São Carlos: EESC-USP; 2003.

Pappas, I.; Keller, T.; Mangold, S.; Popovic, M. R.; Dietz,

V.; Morani, M. A reliable gyroscope-based gait-phase

detection sensor embedded in a shoe insole. IEEE

Sensors. 2004;4:268-274.

Popovic, M. R.; Keller, T.; Ibrahim, S.; Von Büren, G.;

Morani, M. Gait identification and recognition sensor.

International Workshop on Functional Electrical

Stimulation; 1998; Vienna, Austria; 1998. p. 153-156.

Tong, H.; Granat, M. H. A practical gait analysis system

using gyroscopes. Medical Engineering & Physics.

1999;21:87-94.

Veltink, P. H.; Liedtke, C.; Droog, E.; van der Kooij, H.

Ambulatory Measurement of Ground Reaction Forces.

IEEE Transactions on Neural Systems and

Rehabilitation Engineering. 2005;13:423-427.

Wall, J. C. The Gait. In: Durward, B. R.; Baer, G. D.;

Rowe, P.J., editors. Functional human movement:

measurement and analysis. Oxford: Butterworth

Heinemann; 1999. p. 93-105.

DEVELOPMENT OF AN ALTERNATIVE SYSTEM FOR SUSPENDED GAIT ANALYSIS

263