A Low Cost Platform based on FES and Muscle Synergies for Postural

Control Research and Rehabilitation

D. Galeano

, F. Brunetti

1,2

, D. Torricelli

, S. Piazza

and J. L. Pons

Catholic University of Asunci

on, Asunci

on, Paraguay

Bioengineering Group, CSIC, Arganda del Rey, Spain

Keywords:

Posturography, FES, Muscle Synergies, Kinect, Wii Fit.

Abstract:

This paper presents a low cost system for the assessment, diagnosis and training of balance based on static

posturography and functional electrical stimulation (FES). This system includes low cost technology as the

Wii Fit Balance Board and the Kinect. The posturography is a complementary tool to clinical diagnosis,

and allows to ﬁnd sensory systems and inputs degraded by different pathologies. The presented system also

allows to explore new rehabilitation techniques based on functional electrical stimulation. Precisely, this paper

describes the implementation of a novel balance and posture control rehabilitation approach based on muscle

synergies.

1 INTRODUCTION

There are several diseases that can affect human bal-

ance and posture control. Such diversity requires the

participation of different specialists in the diagnosis

and treatment process like neurologists, otolaryngol-

ogists and ophthalmologists among others. Posturog-

raphy is deﬁned as an objective assessment technique

of postural control based. In this way, the monitor-

ing of the center of pressure of the person has proven

to be an effective tool complementary to clinical di-

agnosis in order to quantify this neuromotor disorder.

This technique also can be be used as a complemen-

tary tool to help clinicians with the diagnosis of ver-

tigo.

Posturography evaluates each of the sensory sys-

tems (visual, somatosensory and vestibular) involved

in the complex balance system. Its purpose is to iso-

late the contribution of each of these systems to eval-

uate the status of each one separately. It also assesses

movement strategies for maintaining balance, exam-

ines the stability limits of the person and the ability to

control voluntary movement.

Balance control is an important functional compo-

nent of human gait. After spinal cord injury (SCI) or

stroke, balance control if one of the ﬁrst rehabilitation

objectives towards the restoration of functional gait.

In this scenario, posturography also plays a key role

to evaluate the progress of the affected subject. Clas-

sic therapies of posture control rehabilitation include

exercises to improve stability limit or guided move-

ments to reinforce control efforts of patient.

Over the last years muscle synergies have been de-

scribed for several composed movements like those

exerted during normal postural control. Muscle syn-

ergies can be understood as functional muscle co-

activation patterns (D’Avella and Bizzi, 2005). This

theory proposes the existence of simpliﬁed mecha-

nisms and signals that can control several muscles at

the same time. The most interesting aspect of this

theory is the consistency of these synergies among

subjects, and its stability intra subject. The use of

this knowledge for rehabilitation is still a research

goal, as well as the assessment of muscle synergies in

functional tasks after stroke or SCI, (Torricelli et al.,

2012).

The use of Functional Electrical Stimulation

(FES) to interact with muscle synergies during the re-

habilitation of balance is a novel approach proposed

by Piazza et al., (Piazza et al., 2012). This paper

presents a low cost system that enables the imple-

mentation of this novel rehabilitation paradigm. It

is a posturography tool to help with the assessment

of postural control and its rehabilitation. The main

contribution of the work lies in its simplicity and its

potential use in rehabilitation. It is an exploratory de-

vice to study new rehabilitation approaches of balance

control while monitoring the status of human balance

and postural control system. The presented tool en-

able the evaluation the effectiveness of current treat-

Galeano D., Brunetti F., Torricelli D., Piazza S. and L. Pons J..

A Low Cost Platform based on FES and Muscle Synergies for Postural Control Research and Rehabilitation.

DOI: 10.5220/0004638300210031

In Proceedings of the International Congress on Neurotechnology, Electronics and Informatics (NEUROTECHNIX-2013), pages 21-31

ISBN: 978-989-8565-80-8

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

ments and the design of new ones. The paper presents

technical details of the system and preliminary re-

sults.

Further stages of this work include the validation

of the designed posturography system comparing to

similar ones like the NedSVE/IBV

(Baydal et al.,

2010) of the Institute of Biomechanics of Valencia,

or the SMART Balance Master of Neurocom

. Af-

ter this validation, the design of new herapies based

on FES and muscle synergies will be possible and its

evaluation in clinical environment.

2 ASSESSMENT

METHODOLOGY AND

POSTURAL CONTROL

REHABILITATION

In this section, balance assessment methods used in

posturography are reviewed, as well as the tests de-

signed for this purpose and existing proposals for re-

habilitation based on synergies.

2.1 The Computerized Dynamic

Posturography (CDP)

Computerized Dynamic Posturography (CDP) was

designed and developed by Nashner. It was clinically

studied in collaboration with Black and marketed in

1986 as Equitest by Neurocom Inc., (Faraldo, 2009).

The CDP is a technique that analyzes subject’s

postural control in static standing and and his/her re-

sponse to destabilizing conditions. It is is based on

the idea that the center of gravity (COG) oscillations

reﬂect postural instability. Generally CDPs are based

on dynamometric platforms. These systems analyze

the postural oscillations by recording the vertical pro-

jection of gravity force, known as Center of Pressure

(COP). More frequent tests made with similar plat-

forms are:

• Sensory System or Romberg’s Test. It is aimed

at determining the ability of the patient to inte-

grate the three systems responsible for assessing

standing balance and body sway while different

sensory conditions are applied. The results of

this test are compared with results of normal sub-

jects. It is performed with eyes open and eyes

closed, with and without foam on which the sub-

ject stands. It can also be performed with the

patient’s head retroﬂexed, causing distortion in

neck proprioceptors. These tests can also be used

to evaluate proprioceptive information by making

patients to rely in vestibular information to main-

tain the balance, (Khasnis and Gokula, 2003).

• Stability Limit Test. It assess the capacity of the

subject to bring his COP to the border of his/her

stability limit. Basically, this test is used to as-

sess the maximum distance the patient can move

his/her COP without changing the base of sup-

port, i.e. without moving his/her feet. During the

test, the subject can see his/her COP representa-

tion on a computer screen in front of him, and

he/she should move it toward the stability limits

without moving its base of support. The test in-

cludes up to eight sequential different targets lo-

cated around theoretical stability limits (according

to previous measurements with healthy subjects).

• Rhythmic and Directional Tests. These tests

try to assess subject’s ability to perform rhythmic

movements around of its center of gravity (COG).

The subject isa asked to follow with his/her COP

moving targets whose speed and range are conﬁg-

urable. The target is moved to a percentage of the

stability limit previously calculated for the sub-

ject. This test is usually performed in the antero-

posterior and mediolateral directions.

2.2 Hybrid Approaches in Assisted

Neuromotor Rehabilitation

Hybrid exoskeletons have emerged as a way to im-

prove motor assistance using the beneﬁts of FES and

robotic exoskeletons. They overcome individual lim-

itations of the methods used separately. The FES uses

natural muscles as actuators to generate a movement,

which provides beneﬁts not only functional but physi-

ological. Robotic exoskeletons artiﬁcial actuators are

used to move the members that can not be fully or

partially controlled voluntarily.

Generally, people affected by stroke and SCI have

healthy muscles. The hybrid approach proposes the

use of their own muscles to complement the action

of the robotic exoskeleton. Muscles are activated co-

ordinately with the exoskeleton controller by means

of an electrical stimulation system, (Del-Ama et al.,

2012). This approach results in a eduction of energy

demand and allows the exoskeleton to use lighter and

less powerful actuators. Moreover, this solution is

considered more natural and help to preserve existing

biological structures. Main problem when using FES

is that it can produce muscle fatigue after long peri-

ods of stimulation. This problem limits the time of

use. This is not a problem when using exoskeletons,

which can be used for a longer time.

Balance control is an important not only when re-

habilitating after stroke or SCI but also when using

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

exoskeletons. Hyper is a spanish research project

aimed at developing new neurorobotics and neuro-

prosthetics therapies for people affected by stroke or

SCI. First clinical interventions include the rehabili-

tation of balance and postural control. The use use of

hybrid approaches is well considered by clinicians but

they way they are used in this rehabilitation process

and it effectiveness is not clear yet for the scientiﬁc

community.

The control of the assistive device is also not clear

in terms of compensation actions and movement rou-

tines. The most most common approach over the last

years, was the so called “assist-as-needed” (AAN)

paradigm, (Cai et al., 2006). Following this paradigm,

the interaction between the assistive device and the

natural involved mechanisms in the considered task is

given in terms of the ﬁnal results, and not consider-

ing the underlying status of biological control mech-

anisms. In this way, Hyper encourages the study and

development of new therapies to support classic ones.

These novel treatments are mainly driven by bioin-

spired mechanisms for better and deeper interaction

between the assistive device and remaining neuromo-

tor control structures, in order to reinforce and reha-

bilitate them in a more natural way.

2.3 Muscle Synergies

The study of human control system is a open research

ﬁeld where there are still many questions to answer.

One of them is how is coded the information to con-

trol the large number of degrees of freedom of human

movements. More speciﬁcally, this problem states

that to generate a speciﬁc motor task, there are multi-

ple combinations of muscle activations that can gen-

erate similar results. Muscle synergies theory is a

proposed answer to this question. The central ner-

vous system can solve the complex task by choosing

a speciﬁc set of muscle activations through a combi-

nation of a small set of neural patterns, called synergy,

(D’Avella and Bizzi, 2005).

Each muscle receives as input a modulating signal

from higher neural centers, and outputs a weighted

activation signal to activate a set of muscles. The ac-

tivation of each muscle can be seen is a weighted sum

of all synergies commands connected to it, (Torricelli

et al., 2011). Then, muscle synergies can translate

small sets of variables coming from the central ner-

vous system into higher dimensional signals. They

are strictly correlated to the functional performance

and their modulation are related with user workspace.

The most interesting characteristic of muscle syner-

gies is that they are consistent inter healthy subjects,

(D’Avella and Bizzi, 2005)(Piazza et al., 2012).

Mathematically, as indicated in (Torricelli et al.,

2011), can be expressed by the following equation

that describes the activation of a single muscle m:

m(t) =

∑

(t)w

(1)

in which

• m is the activation of the muscle function of time.

• c is the neutral command i-th synergy function of

time.

• w is the constant weight of the i-th synergy to the

muscle in question.

• K is the number of synergies.

The use muscle synergies knowledge to rehabili-

tate postural control is not clear. However, their role

in functional movements and their importance have

being already reported (D’Avella et al., 2006) (Piazza,

2013). This encourage Hyper scientiﬁc team to take

them into account to favour the development of more

efﬁcient rehabilitation therapies by closely interacting

with involved muscle synergies in balance control. In

this way, FES can be used to develop and interact with

synergies and muscle activation patterns.

3 PROPOSED SYSTEM

In this section we describe the low-cost platform de-

veloped to perform static posturography tests and sup-

port treatments based of FES and muscle synergies.

Balance control assessment platforms are usually not

open and they are commercially available only as a

posturography tool. Thus, a novel low cost and open

posturography platform was developed. The main ob-

jective of this platform is to support the development

of novel balance control rehabilitation therapies in the

framework of Hyper project.

Figure 1 shows the outline of the developed plat-

form which is further described. The platform is

based on a distributed architecture, which include sev-

eral components: the posturography controller, the

real-time neuroprosthetic controller, a Wii Fit Balance

Board, a Kinect camera and the TEREFES electros-

timulation system, (Brunetti et al., 2011).

3.1 Wii Fit Balance Board

Wii Fit balance board is an input device included in

the Wii Fit from Nintendo

. It is a wireless device

that uses Bluetooth technology to communicate with

the Wii console. It is equipped with four resistive

pressure sensors located in each corner of the table.

ALowCostPlatformbasedonFESandMuscleSynergiesforPosturalControlResearchandRehabilitation

Wii Fit Balance Board

(pressure center)

Bluetooth

Posturography

Software

(visual feedback

the patient)

Kinect

(skeleton tracking and video)

TEREFES

(Functional Electrostimulation)

Computer Doctor

(Control and Configuration

Posturography software)

Bluetooth

FES and Sinergys Controls

(Matlab Real-Time)

TCP/IP Internet

(Feedback control)

HDMI

USB 2.0

Figure 1: Proposed platform.

In effect, it measures the displacement of the center

of pressure and the weight of the user. It also gives an

indication of the battery status.

Over the last years, the Wii Fit Balance Board

have been used by the scientiﬁc community, specially

as computer interface for disabled, (Martin, 2008).

This device has two attractive features: it is wireless

and low-cost. In our project, the Board will be used

to measure the COP.

Data from Wii Fit Balance Board is accessed

through a Microsoft Visual Studio C# application,

using a library called WiimoteLib available at wi-

imotelib.codeplex.com. Visual C# was chosen be-

cause it is also compatible with the Kinect and its Win-

dows Software Development Kit (SDK). Thus, the Wii

Fit is connected as a HID interface device. Provided

services by the Board are detected using the Service

Discovery Protocol (SDP) of Bluetooth.

An important aspect to consider is the sampling

frequency at which the Wii Fit sends the data to

the PC, or more speciﬁcally, how often the data

arrives, considering the nature of wireless trans-

mission and the operating system behavior. To

answer this question, we measured the time in-

terval between samples using methods and public

properties of the Microsoft Visual Studio C# class

System.Diagnostics.Stopwatch. The program

is executed in a almost dedicated HP Pavilion g6-

1b70us Notebook (Intel Core i3 CPU M 370 @

2.4GHz, 4GB of RAM) running Windows 8 64-bit.

Wii Fit Balance Board measured data sampling peri-

ods are depicted in ﬁgure 2. As it can be seen, the

average sampling frequency is 100 samples/second.

The probability density function obtained with

from the data shown in Figure 2 is depicted in

Figure 3. This function was calculated with the

Distribution Fitting Tool from Matlab. The

parameters deﬁned in the tool were: core (normal),

bandwidth (auto), domain (unlimited).

The aim of this analysis it to know how determin-

istic is the access to the data of the the Board in terms

0 500 1000 1500 2000 2500 3000 3500 4000

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

Measured

Measuring Average

Number of Samples

Sampling period (s)

Figure 2: Time oscillations between consecutive samples

received from the Wii Fit Balance Board. Sampling period

is expected to be 10 ms.

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

100

Sampling period(s)

Density

Measured

Fit (non−parametric)

Figure 3: Probability density function of the sampling pe-

riod of the Wii Fit.

of time. In other words, we want to know the prob-

ability that the sampling period of the Wii Fit is less

than or equal to any given time. Cumulative distribu-

tion function for the Wii Fit data is depicted in ﬁgure

4. For example, the probability that the sampling pe-

riod remain less than or equal to 0.02 s is 94.02%.

3.2 Microsoft Kinect

Since 2009, when it was announced, the Kinect for

Microsoft Xbox 360 has been widely used in many

applications. The Kinect device is a natural user inter-

face, which allows users to interact with games with-

out physical contact with any command. It was devel-

oped by PrimeSense Company. The user becomes the

controller itself, having to rely on movements, natu-

ral gestures and voice commands to control game el-

ements.

Kinect is equipped with an RGB-D camera that

acquires images of 640x480 at 30 fps. It has a vi-

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Cumulative(probability

Measured

Fit (non−parametric)

Sampling period (s)

Figure 4: Cumulative distribution function as a function of

the sampling period (s).

Figure 5: A Kinect camera unwrapped (Miles, 2012).

sual ﬁeld range from 1.2 to 3.5 meters, but can be

reduced by optical coupling, as Niko Zoom Lenses

. Furthermore, its viewing angle is 57

◦

horizon-

tally, and 43

◦

vertically. The vertical visual ﬁeld

can be expanded 27

◦

with its servomotor. It is also

equipped with an array of four microphones, each

with a recording resolution of 16 bits sampled at 16

kHz. It also contains a stack of signal processing

hardware that is able to handle all the data that cam-

eras, infrared light, and microphones generate. By

combining the output of these sensors, a program can

track and recognize objects in front of it, determine

the direction of the sound signals, and isolate them

from the background noise. The disassembled hard-

ware is shown in Figure 5.

This unique device has not gone unnoticed by the

scientiﬁc community. Proof of this is the immedi-

ate development of free SDK such as OpenNI

and

OpenKinect

, used in many research projects as a

novel human-computer interface, (Brunner, 2012).

The role of the Kinect in the platform is to enrich

www.openni.org

openkinect.org/wiki/Main Page

Center of hip

Center of

shoulders

Left ankle

Left foot

Left hip

Left knee

Left

wrist

Left

elbow

Left

shoulder

Head

Right foot

Right hip

Right knee

Right

wrist

Right

elbow

Right

shoulder

Right hand

Left hand

Spine

Right ankle

Figure 6: Schematic of the 20 points of the human skeleton

that can track the Kinect (Catuhe, 2012).

Sampling Period (s)

0 200 400 600 800 1000 1200

0.01

0.02

0.03

0.04

0.05

0.06

Measured

Measuring Average

Number of samples

Figure 7: Sampling time the Kinect sends data to the PC.

the visual feedback provided to the patient. Com-

mon posturography platforms are limited to provide

users information about the center of pressure but the

user does not know precisely his/her position and how

good it is for the intended task. In this way, the Kinect

provides kinematic information of full body segments

(Figure 6), thus providing more complete information

to users as well to the neuroprosthetic controller en-

abling better actuation commands.

Both the Wii Fit as the Kinect, help to give vi-

sual feedback to the evaluated subject, and the infor-

mation generated can be used by the neuroprosthetic

controller to generate more precise and adequate stim-

ulation patterns. A similar analysis done with the Wii

Fit Board, was carried out with Kinect to evaluate

the jitter effect when acquiring the frames. Samples

taken during a period of approximately 20 seconds are

shown in Figure 7.

The probability density function of the data in Fig-

ure 7 corresponds to a normal distribution pattern,

now depicted in Figure 8. The distribution has a mean

ALowCostPlatformbasedonFESandMuscleSynergiesforPosturalControlResearchandRehabilitation

0 0.01 0.02 0.03 0.04 0.05 0.06

100

150

Sampling period (s)

Density

Measured

Fit (normal distribution)

Figure 8: Probability density function of the sampling pe-

riod kinect.

0 0.01 0.02 0.03 0.04 0.05 0.06

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Cumulative probability

Sampling period (s)

Measured

Fit (normal(distribution)

Figure 9: Cumulative distribution function of the sampling

period Kinect.

of 0.032 s (∼ 30 fps) and a variance of 3.62e-5.

The cumulative distribution function in Figure 9.

For example, the probability that the sampling fre-

quency is maintained below 35 fps is 75%, approx-

imately.

3.3 Posturography Controller

The Posturography Controller is implemented in a

personal computer running Windows 8 operating sys-

tem. The developed software includes traditional pos-

turography tools and tests like Romberg’s test, test of

the stability limits and rhythmic directional test.

The software was developed for easy use by med-

ical personnel. It includes a database in which data of

each patient is stored, allowing the physician to eval-

uate subject progress after several sessions. It also

helps to diagnose and program rehabilitation exercise

routines for each one subject. The application is also

able to generate Matlab scripts containing the center

of pressure points recorded during each rehabilitation

session. In this way, the therapists can analyze data

recorded in previous sessions.

The Posturography Controller receives all the data

from the Kinect camera and the Wii Fit Balance

Board. It fuses and displays the acquired data show-

ing information like the center of pressure, the rigid

body kinematic chain of the studied/analyzed subject,

and information about current routines and tests. This

controller is connected to the Neuroprosthetic Con-

troller thought a TCP/IP connection.

In next sections details of the various parameters

used to quantify the results of each test will be de-

scribed, (Garc

ıa, 2012).

• Romberg’s Test. The subject is positioned on the

Wii Fit Balance Board in an upright position with

arms straight and close to the body trying not to

move the head in neutral position facing forward,

bare feet at an angle Opening of 30

◦

. In this po-

sition is assessed for T seconds (conﬁgurable by

the doctor) their ability to maintain balance in the

following conditions:

– Eyes Open (REO) and Eyes Closed (REC).

– Foam on Wii Fit with Eyes Open (RGA) and

Eyes Closed (RGC).

The parameters evaluated in each test are:

– Shift angle (degrees). The angle of the vector

extending from the initial point to the subject

portion to the end point of the trajectory.

– Swept Area (mm

). It estimates the area swept

by the COP by mean of an ellipse whose axes

correspond to the maximum mediolateral and

anteroposterior displacement.

– Average speed (cm/s). It estimates the average

speed, which is the ratio between the displace-

ment and time T that lasts the test.

– Maximum mediolateral and anteroposterior

displacements (mm). These parameters repre-

sent the longest displacement in the mediolat-

eral and anteroposterior axis during the exer-

cise.

Figure 10 shows a screenshot during the execution

of the application designed in this project. Specif-

ically, this screenshot corresponds to a REO Test.

The ﬁgure shows two visual feedbacks. The ﬁrst

one is the position of the center of pressure on the

Wii Fit Balance Board, and the second, provided

by Kinect, is the RGB image and a trace of its

skeleton. The screen provides information about

subject’s skeleton (skeleton blue) and a given ref-

erence (red skeleton). The reference indicates cor-

rect estimated position during the tests.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

Figure 10: Screenshot during the execution of REO Test.

On the right,it is shown the center of pressure on the Wii

Fit while on the left the subject with his/her skeleton (blue

lines) and the given reference (red lines).

Some parameters are provided in real time by the

application. For example, regarding the Kinect, it

monitors the status of the tracking task, which can

be Tracking (OK) or not skeleton (Subject not de-

tected). Another parameter is the quality of the

skeleton, this parameter indicates if the Kinect

is showing the complete skeleton of the subject

(Good Quality). If the quality is poor, the clini-

cian or therapist can ask the patient to move up,

down, left or right, depending on the case, to get a

better result. This will help the doctor to point the

camera in the correct position. Regarding the Wii

Fit Balance Board, the parameters observed are

subject’s weight and the coordinates of the COP.

• Stability Limit Test. This test evaluates the fol-

lowing parameters:

– LE max (mm). It is the maximum value reached

by the COP in the corresponding direction (8

targets separated of 45

◦

and whose radial dis-

tance from the origin is conﬁgurable).

– Stability zone (cm): It is approximately the

mean distance at which the patient is 90 % of

the time. It is calculated for each direction.

Figure 11 shows a screenshot during the execu-

tion of limit test. The variation of the COP on

the Wii Fit for the Limit Test is depicted on the

right side. The red circle represents the current

target to which you should direct your COP, while

the green ones represent those already targeted.

Traces of COPs in these directions have been

deleted to not disturb the patient with current tar-

get.

• Rhythmic and Directional Test. In this test, the

patient is asked to follow the movements of a

moving target (conﬁgurable frequency ) in medi-

olateral and anteroposterior directions. The max-

imum excursion limit is calculated based on the

parameters of normal stability limits (previously

recorded with healthy subjects).

Figure 11: Screenshot during the execution of the limit test.

On the right side of the screen, the COP on the Wii Fit is

shown in real time. On the left side is shown the patient

with his/her stickman representation (blue lines).

Figure 12: Screenshot for Rhythmic Test ML execution. On

the right, we see the COP on the Wii Fit. To the left is shown

the patient with his skeleton (blue lines).

The following parameters are evaluated for each

direction.

– Reaction Time (s). It is deﬁned as the time that

the subject takes to bring his/her center of grav-

ity closer than two centimeters from the refer-

ence target.

– Tracking Capability (%). It quantiﬁes subject’s

ability to follow the movement of the target in

ML or AT directions. This parameter is cal-

culated as the mean of error (DesiredCOP −

MeasuredCOP), after the reaction time. If the

error is lower than 2cm (conﬁgurable), in other

words the COP is inside the target circle, it is

considered as zero for this sample.

– Directional Control (%): quantiﬁes the sub-

ject’s ability to remain in the expected direction

of the test. For example, if the target moves in

the axis ML, is evaluated AT axis error using

the same process for calculating the tracking

capability.

Figure 12 shows a screenshot during the execution

of the application for Mediolateral (ML) Rhyth-

mic Test. On the right, the screen shows the mov-

ing target.

Figure 13 shows a screenshot during the execution

ALowCostPlatformbasedonFESandMuscleSynergiesforPosturalControlResearchandRehabilitation

Figure 13: Screenshot for AT Rhythmic Test execution. On

the right, the COP on the Wii Fit is depicted. On the left, it

is shown the subject with his stickman representation (blue

lines).

of the application, more speciﬁcally, a anteropos-

terior (AP) Rhythmic Test.

3.4 Neuroprosthetic Controller

The Nueroprosthetic controller is responsible for the

generation of muscle activation patters and for control

of the actuation system: the TEREFES electrostimu-

lator. It receives from the Posturography controller

all the kinematic data of the subject (acquired with

the kinect) and the coordinates of the center of pres-

sure (COP). A driver will decode or convert the in-

formation to muscle activation patterns and speciﬁc

TEREFES commands accoding to previously pro-

grammed synergies sets, theory and rehabilitation pa-

rameters. Full detail of the proposed synergistic con-

troller can be found in (Denis et al., 2012)

The functional stimulator TEREFES must act syn-

chronously according with the exerted movements.

This imposes the real-time nature that must fulﬁll the

Neuroprosthetic Controller. To achieve this and ease

the development of novel controllers a real time Mat-

lab kernel is used. Thus, the delay caused by the Pos-

turography Controller can be determines before gen-

erating actuation command. Further analysis of the

performance of system timing should be realized in

next stages.

3.5 TEREFES

The TEREFES was proposed within the framework

of the TERERE and Hyper projects (Brunetti et al.,

2011). The TEREFES electrostimulator provides up

to 32 stimulation channels driven by controllable and

stable and close loop current sources. In addition, the

system is portable and ﬂexible. This functional stim-

ulator is powered by 4 AA batteries and includes a

USB communication interface that allows its conﬁg-

uration via external software. Monophasic and bipha-

sic stimulation signal can be obtained in its 32 avail-

able channels. This channels are divided in two inde-

pendent groups of 16 channels each, that can be stim-

ulates simoultaneously.

4 PRELIMINARY RESULTS

In this section preliminary results of posturography

software are presented. Described results were ob-

tained with 6 healthy people, 4 men and 2 women.

The purpose of this functional validation is to tech-

nically verify the platform and to compare result be-

tween different subjects. Unfortunately, at this stage

of the work, the system could not be tested with previ-

ously diagnosed pathological subjects, and the results

could not be compared with those obtained with other

commercial platforms like Neurocom.

The procedures for the tests were explained in pre-

vious sections. REO and REC tests were conducted,

as well as Stability Limit and Rhythmic tests. All of

them were realized a couple of times in order to make

sure that the subjects understand the test but with-

out producing fatigue or previous learning/training

(Garc

ıa, 2012). The sampling frequency was 30

frames/second, enough to detect any COP variation

(Enbom et al., 1988).

4.1 Romberg’s Test

Each Romberg’s test lasted 30 seconds. The results of

the 6 subjects are shown in Table 1.

Table 1: REO and REC Test results.

Subject Sex Years Disp. angle (

◦

) S. Area (cm

) A.Speed (cm/s) Disp. ML (cm) Displ. AT(cm)

REO REC REO REC REO REC REO REC REO REC

1 M 23 108,22 114,45 18,28 17,21 1,94 2,28 2,42 3,16 1,82 3,00

2 M 26 90,77 91,45 2,71 6,06 1,25 1,65 0,817 1,552 1,01 1,22

3 M 34 95,89 113,77 12,91 12,52 1,69 2,15 2,46 2,71 2,01 2,31

4 M 47 76,55 75,71 9,05 4,88 1,23 1,34 1,81 1,55 1,85 1,21

Average 92,86 98,84 10,74 10,17 1,53 1,86 1,88 2,23 1,67 1,94

5 F 19 122,49 112,1 4,94 5,85 1,53 1,95 2,06 2,05 1,06 2,45

6 F 18 106,79 103,85 8,99 7,45 1,34 1,65 2,15 2,03 1,13 2,32

Average 114,64 107,96 6,97 6,65 1,44 1,80 2,11 2,04 1,095 2,385

All proposed parameters were calculated and they

are presented in Table 1. Results suggest a decrease

of ﬁne postural control in most subjects when they

close their eyes. For both men and women, the dis-

placement angle is usually in the second quadrant, and

no signiﬁcant differences are found among REO and

REC tests. In fact, according to (Garc

ıa, 2012) this

parameter does not change signiﬁcantly under these

test conditions.

Figure 14 shows the results of subject 4. Using

similar data, proposed parameters were calculated for

each subject.

Regarding the swept area, calculated by many

professionals in the ﬁeld according to the literature

(Black et al., 1989), it does not reﬂect noticeable

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

0 5 10 15 20 25 30

Time (s)

COPx (cm)

0 5 10 15 20 25 30

Time (s)

COPy (cm)

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

COPx.(cm)

COPy.(cm)

Figure 14: Subject 4 (S4) REC Romberg’s test plotted using

Matlab. Parameters for S4 are calculated with these data.

changes with the changing sensory conditions. Bal-

aguer, in his work, (Garc

ıa, 2012), has suggested that

the calculation by ﬁtting a geometric ﬁgure may not

be adequate to quantify this parameter.

Finally, the average speed of displacement, is

found to increase without visual feedback. This same

behavior is observed in the mediolateral and antero-

posterior displacement. Therefore, these parameters

are used to differentiate visual system impact and po-

tential dysfunction in balance control. Both men and

women present larger variations in the mediolateral

direction, being even larger in men in these particular

tests.

4.2 Stability Limit Test

The stability limit test lasted 10 seconds for each di-

rection, and each target was located at a distance of

10 cm from the origin. The subject was asked to make

his/her best effort to reach the targets.

Table 2: Stability Limit rest results. The average values

are shown. Similar data can be used to obtain normality

patterns.

Average Max. LE (cm) Average Stability Zone (cm)

Direction

Sex

Male Female Male Female

Front 7,51 9,04 5,77 8,32

Front-right 9,68 9,695 8,05 8,73

Right 10,59 9,165 9,28 7,70

Rear-right 9,44 8,58 8,92 7,33

Rear 10,06 8,00 8,68 6,915

Rear-left 9,88 9,33 8,32 8,64

Left 11,01 10,05 9,00 8,35

Front-Left 9,335 9,97 8,2675 8,55

Figure 15 shows the results of subject 1. Using

similar data, proposed parameters were calculated for

each subject.

According to these tests, areas of stability in both

men and women vary with direction. In general terms,

0 2 4 6 8 10 12

−20

−10

Limit Test Results − Left (Session 1)

Time (s)

COPx (cm)

0 2 4 6 8 10 12

−2

Time (s)

COPy (cm)

−14 −12 −10 −8 −6 −4 −2 0 2 4

−2

COPx (cm)

COPy (cm)

Measured

Reference

Measured

Reference

Measured

Figure 15: Subject 1 (S1) Limit Test Results plotted using

Matlab. Parameters for S1 are calculated with these data.

there are no signiﬁcant differences. These results

agree with (Cort

es, 2007). However, a larger popula-

tion is needed to obtain robust conclusions. Balaguer

found that the subject own subjective perception (Pre-

vious Q&A about disability condition of the subject)

of his/her skill or disabilities does not inﬂuence the

stability limits, (Garc

ıa, 2012).

4.3 Rhythmic and Directional Control

Test

For the rhythmic tests, windows of 10 cm (con-

ﬁgurable) long were deﬁned directionally. First in

the mediolateral direction and then in the anterior-

posterior one. The subject was asked to follow a mov-

ing target traveling at a frequency of 0.25 Hz. Each

test lasted 20 seconds. The results are shown in Table

3 for each patient.

Table 3: Test results rhythmic control.

Subject Sex Years Reaction time(s) Tracking capability(%) Directional control (%)

ML AT ML AT ML AT

1 M 23 0,037 0,50 81,86 85,7 81,7 99,9

2 M 26 0,119 0,039 80,55 70,44 90,17 99,05

3 M 34 0,12 1,059 87,7 63,8 78,22 99,83

4 M 47 0,198 1,046 79,6 57,8 77,51 99,92

Average 0,1185 0,66025 82,43 69,43 81,9 99,675

5 F 19 0,40 0,035 66,5 70,23 79,7 98,5

6 F 18 0,42 0,21 71,5 55,26 88,53 96,94

Average 0,4125 0,1225 69 62,74 84,11 97,72

Figure 16 shows the results of the subject 5. Sim-

ilar results were used to calculate all parameters for

each subject.

According to these results, the reaction time in-

creases with age. In addition, there is a shorter reac-

tion time in women. In men, the reaction time is better

in the ML direction with respect to the AT. Although

this work has not made a study of subjects with spe-

ciﬁc pathologies, (Garc

ıa, 2012) found that vestibular

disorders can affect rhythmic and directional control

in disagreement with the ﬁndings of Cort

es, (Cort

es,

2007).

ALowCostPlatformbasedonFESandMuscleSynergiesforPosturalControlResearchandRehabilitation

0 5 10 15 20 25

−5

AT Rhythmic Test Results (Session 1)

Time (s)

COPx (cm)

0 5 10 15 20 25

−10

Time (s)

COPy (cm)

−1.5 −1 −0.5 0 0.5 1 1.5 2 2.5

−10

COPx (cm)

COPy (cm)

Measured

Reference

Measured

Reference

Measured

Figure 16: Subject 5 (S5) Rhythmic test Results in the AT

direction plotted using Matlab. Parameters for S5 are calcu-

lated with these data.

5 DISCUSSIONS

Posturography helps to assess the inﬂuence of any

vestibular dysfunction in postural and balance con-

trol. However, a pathology that affect the balance in

one patient, in other word the vestibular-spinal reﬂex,

not necessarily will do it in another one. In this case,

tools like the one described in this work are not effec-

tive for the diagnosis of the impairment.

Regarding the tool presented in this paper, it i

not very clear in the literature the way how differ-

ent assessment parameters are calculated. This lack

of information make more difﬁcult to compare re-

sults. However, overall conclusions and trends ob-

tained with this tool are similar to those reported in

the literature and obtained with other platforms.

Nowadays, there is still a discrepancy between

scientist regarding the results of each parameter and

associated information. According to (Garc

ıa, 2012),

this discrepancy exits because it is difﬁcult to ﬁnd

clear relationships between functional assessment of

balance and patient-perceived disability. Tests may

be inﬂuenced by many factors like social, profes-

sional, technical, psychological, affective, and cogni-

tive ones.

The current drawback of classical static postur-

ography is limited only to study the subject during

standing position, so it does not provide information

on the dynamic aspects of postural control. To solve

this shortcoming, we have followed the line proposed

by (Garc

ıa, 2012) and set dynamic tests, such as the

rhythmic test.

6 CONCLUSIONS AND FUTURE

WORK

Postural rehabilitation boosts patient conﬁdence and

contribute to their self-improvement. In addition,

knowledge of the particular deﬁcit in postural control

helps clinician and patient to develop prevention plans

to avoid falls, and as mentioned before, it is the ﬁrst

step towards the rehabilitation of more complex pro-

cesses like gait.

Current research projects in neuromotor rehabili-

tation like Hyper, are devoted to develop novel bioin-

spired rehabilitation treatments. The use of hybrid

solutions including neurorobotics and neuroprosthet-

ics devices has been shown as an efﬁcient approach.

However, the use and development of modern rehabil-

itation therapies based on novel knowledge need the

support of non existing research tools.

We have seen how to make a low cost posturogra-

phy system. It is based on a Wii Fit Balance Board,

the Microsoft Kinect and the TEREFES electrosti-

mumlator. This tool can serve as a low cost balance

control assessment tool and will allow the implemen-

tation of novel therapies that could improve current

ones for the rehabilitation of balance control.

Future work includes the evaluation of the tool and

developed system in clinical environments. After this

validation, the ﬁnal integration of the neuroprosthetic

controller and the implementation of therapies based

on muscle synergies will be done.

REFERENCES

Baydal, J., Castelli, A., Garrido, J., Bermejo, I., Broseta, J.,

Amparo, M., J, P., and Moya, M. (2010). Nedsve/ibv

v.5 a new system for postural control assessment in

patients with visual conﬂict.

Black, F., Shupert, C., Peterka, R., and Nashner, L. (1989).

Effects of unilateral loss of vestibular function on the

vest

ıbulo-ocular reﬂex and postural control. Annals of

Otology, Rhinology, and Laryngology, 98:884–889.

Brunetti, F., Garay, A., Moreno, J., and Pons, J. (2011).

Enhancing functional electrical stimulation for emerg-

ing rehabilitation robotics in the framework of hyper

and project. In 2011 IEEE International Conference

on Rehabilitation Robotics Rehab Week Zurich, ETH

Zurich Science. IEEE.

Brunner, S. (2012). Using Microsoft Kinect Sensor to per-

form commands on virtual objects. Master’s thesis,

Polyechnic University of Turin, Italy.

Cai, L., Fong, A., Yongqiang, L., Burdick, J., and Edger-

ton, V. (2006). Assist-as-needed training paradigms

for robotic rehabilitation of spinal cord injuries. In

Proceedings of the 2006 IEEE International Confer-

ence on Robotics and Automation (ICRA’06).

Catuhe, D. (2012). Programming with the Kinect for Win-

dows Software Development Kit. Microsoft Press.

Cort

es, O. (2007). An

alisis cl

ınico y posturogr

aﬁco en an-

cianos con patolog

ıa vestibular y su relaci

on con las

ıdas. PhD thesis, University de Valencia, Spain.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

D’Avella, A. and Bizzi, E. (2005). Shared and speciﬁc mus-

cle synergies in natural motor behaviors. Proceedings

of the National Academy of Sciences of the United

States of America, 102(8):3076–3081.

D’Avella, A., Portone, A., Fernandez, L., and Lacquaniti, F.

(2006). Control of fast-reaching movements by mus-

cle synergy combinations. Journal of Neuroscience,

26(30):7791–7810. cited By (since 1996) 78.

Del-Ama, A., Koutsou, A., Moreno, J. C., and Pons, J.

(2012). Review of hybrid exoskeletons to restore gait

following spinal cord injury. Journal of Rehabilitation

Research and Development, 49:497–514.

Denis, W., Brunetti, F., Piazza, S., Torricelli, D., and Pons,

J. (2012). Functional electrical stimulation controller

based on muscle synergies. In Proceedings of the

First International Conference on Neurorehabiltia-

tion, Converging Clinical and Engineering Research

on Neurorehabilitation.

Enbom, H., Magnusson, M., Pyykko, I., and Schalen, L.

(1988). Presentation of a posturographic test with

loading of the proprioceptive system. Acta Otolaryn-

gol Suppl., 455:58–61.

Faraldo, A. (2009). Register postural in healthy: evalua-

tion of balance by comparative study of computerized

dynamic posturography and sway star system. PhD

thesis, University of Santiago de Compostela, Spain.

Garc

ıa, R. B. (2012). Valoraci

on de un m

etodo de postur-

ograf

ıa est

atica con pruebas din

amicas para evaluar

funcionalmente pacientes vestibulares en edad labo-

ral y su relaci

on con el

ındice de discapacidad. PhD

thesis, Polytechnic University of Valencia, Spain.

Khasnis, A. and Gokula, R. (2003). Romberg’s test. Journal

of Postgraduate Medicine, 2:169–172.

Martin, E. A. (2008). Study and development of man-

machine interface based on wirelles sensor. University

Pontiﬁcia Comillas.

Miles, R. (2012). Start Here!. Learn Microsoft Kinect API.

O’Reilly Media, Inc.

Piazza, S. (2013). Muscle synergies in postural sway move-

ments: neurophysiologial evidences and reabilitation

potentials. Master’s thesis, University Carlos III of

Madrid, Spain.

Piazza, S., Torricelli, D., Brunetti, F., del Ama, A. J., Gil-

Agudo, A., and Pons, J. (2012). A novel fes con-

trol paradigm based on muscle synergies for postu-

ral rehabilitation therapy with hybrid exoskeletons. In

Proceedings of 34th Annual International Conference

of the Engineering in Medicine and Biology Society.

(EMBC’12). IEEE.

Torricelli, D., Aleixandre, M., Alguacil, I., Cano, R.,

Molina, F., Carratal

a, M., Piazza, S., and Pons, J.

(2012). Modular control of mediolateral postural

sway. In Proceedings of 34th Annual International

Conference of the Engineering in Medicine and Biol-

ogy Society (EMBC’12).

Torricelli, D., Moreno, J. C., and Pons, J. L. (2011). A

new paradigm for neurorehabilitation based on mus-

cle synergies. In Proceedings of 34th Annual Interna-

tional Conference of the Engineering in Medicine and

Biology Society. (EMBC’12).

ALowCostPlatformbasedonFESandMuscleSynergiesforPosturalControlResearchandRehabilitation