A PORTABLE REAL-TIME MONITORING SYSTEM

FOR KINESITHERAPIC HAND REHABILITATION EXERCISES

Danilo Pani

, Gianluca Barabino

, Alessia Dess`ı

, Alessandro Mathieu

and Luigi Raffo

DIEE - Dept. of Electrical and Electronic Engineering, University of Cagliari, Cagliari, Italy

Chair of Rheumatology and Rheumatology Unit, University and AOU of Cagliari, Cagliari, Italy

Keywords:

Kinesitherapy, Real-time, Rheumatic disease, Hand rehabilitation.

Abstract:

Rheumatic diseases, such as rheumatoid arthritis and systemic sclerosis, may seriously reduce the quality of

life of the patients. Nowadays, their progress can be controlled only through personalised pharmacological

treatments. Kinesitherapy can also help in faster movement recovery, also contrasting the disability worsening.

This paper presents a portable low-cost system for the real-time quantitative monitoring and evaluation of hand

rehabilitation exercises. The system, based on a MSP430 microcontroller central unit, provides a platform for

the analysis of ﬁne characteristics hitherto unavailable of 4 exercises required for the hand rehabilitation in

rheumatic patients. The systemcan be controlled, through a Bluetooth connection, by a graphical user interface

running on the physician’s PC. The ﬁrst prototypical systems have been developed for experimental outpatient

trials.

1 INTRODUCTION

Rheumatic diseases, such as rheumatoid arthritis and

systemic sclerosis, may severely reduce the quality

of life of the patients, which are required to undergo

an integrated therapy including kinesitherapy and a

personalised pharmacological protocol. If the lat-

ter presently represents the only way to control the

progress of the disease, the former both allows a faster

recovery after inactivity periods caused by active dis-

ease and contrasts the progressive disability. For in-

stance, patients with scleroderma suffer a skin thick-

ening, typically localized on the hands, whose con-

sequence is a limited mobility which in turn exac-

erbates the problem giving rise to a vicious circle.

When the lesions are localized on the hand, such in-

validating diseases hamper the execution of normal

daily life activities such as hair brushing, dressing or

cooking. Both hand strength and ﬁne movements are

often compromised.

Speciﬁcally designed physical exercises associ-

ated to an appropriate pharmacological therapy can

help in restoring the motor function of the hand. In

order to achieve the best results, such exercises must

be properly performed, with the right number of se-

ries and repetitions. During outpatient examinations,

an expert physician can evaluate the quality of the

movements in order to effectively guide the patient

through the training, avoiding the onset of inﬂamma-

tory ﬂares involving the hand. However, both a quan-

titative measurement and analysis of the patient’s ef-

fort, and the deﬁnition of the best suited protocol for

a speciﬁc patient, are hampered by the lack of instru-

ments expressly designed and packaged to this aim.

Beyond qualitative analyses including visual inspec-

tion and questionnaires administration, only for some

exercises (typically grip and pinch strength) some

digital devices are able to provide one-shot measure-

ments. Unfortunately they are quite expensive and

hardly integrable in a complete rehabilitation moni-

toring framework.

In this paper, a portable prototypical system for

the real-time quantitative evaluation of hand rehabili-

tation exercises is presented. Compared to the typical

procedures at the state of the art, the proposed system

has been designed in cooperation with expert rheuma-

tologists to monitor 4 agility/strength exercises, al-

lowing to analyse with a ﬁner resolution character-

istics of the execution otherwise hitherto unavailable

(e.g. speed, frequency, execution precision). On-line

monitoring is provided by a MSP430 microcontroller

(MCU) based subsystem able to perform real-time

event detection and measurements on the incoming

signals from the 4 sensorized devices. The system,

battery-powered for patient’s safety and conveniently

accommodated in a metal briefcase, is controlled by

Pani D., Barabino G., Dessì A., Mathieu A. and Raffo L..

A PORTABLE REAL-TIME MONITORING SYSTEM FOR KINESITHERAPIC HAND REHABILITATION EXERCISES.

DOI: 10.5220/0003793400820089

In Proceedings of the International Conference on Biomedical Electronics and Devices (BIODEVICES-2012), pages 82-89

ISBN: 978-989-8425-91-1

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

a desktop PC via a Bluetooth connection which re-

ceives both the raw signals, the measurements col-

lected until then and on-line reﬁned statistics on them.

A graphical user interface (GUI) developed in MAT-

LAB allows a real-time qualitative and quantitative

analysis of the exercise execution. The presence of

different sensorized devices into a single system im-

proves usability and it opens to the integration of ad-

ditional devices. The system is going to be evaluated

in a clinical trial in Italy from October 2011.

The remainder of this paper is structured as fol-

lows. In Section 2 a brief description of the state of

the art is provided. Section 3 provides an overview

of the proposed system, whose hardware structure is

explained in Section 4, whereas the physician GUI is

presented in Section 5. Section 6 concludes this work

with ﬁnal remarks and perspectives for future devel-

opments.

2 STATE OF THE ART

Several diseases can affect the human hand, impair-

ing its functionality and then negatively inﬂuencing

the daily life. The largest part of studies in the ﬁeld of

the hand rehabilitation with biomedical devices deals

with the post-stroke recovery (Dovat et al., 2008). In

this case, cable-driven units connected to each ﬁn-

ger by means of soft rings are exploited, being able

to move the ﬁngers with predeﬁned patterns (passive

movements) and/or to provide a tunable resistance to

the hand movement. Other approaches make use of

complex mechanical infrastructures (Huang and Low,

2008) or exoskeletons in order to assist the movement

(Brokaw et al., 2011) or help in restoring the motor

function (Iqbal et al., 2010).

For functional assessment only, the most common

evaluation involves pinch and grip exercises. Both the

Jamar dynamometer (isometric) and the Vigorime-

ter (dynamic) represent well established instruments

for the clinical evaluation of the grip strength (Peters

et al., 2011). Commercial devices such as Pablo by

Tyromotion GmbH or the H500 Hand Kit by Biomet-

rics Ltd. allow monitoring also the single ﬁnger pinch

force. In principle, isometric wrist dynamometer can

be also used to estimate the torque applied with the

ﬁnger when the wrist is in a ﬁxed position, in order

to evaluate the hand performance with respect to this

task. Usually the digital versions of these devices are

able to provide maximum, average and standard devi-

ation of the force, but without any temporal analysis

within a series without additional electromyographic

signals (Seo et al., 2009). In (Helliwell et al., 1987),

a grip measurement device is presented, able to per-

form also some time measurements but only on a sin-

gle 4.4s grip exercise for the performance assessment

in rheumatic patients. A similar work has been pre-

sented in (Andria et al., 2006) for the parkinsonian

patients. In both cases the aim is a one-shot func-

tional assessment rather than the monitoring of a se-

ries of exercises, since multiple repetitions are some-

times used only for statistical purposes. An interest-

ing device for rehabilitation mixing torque and grip

force has been presented in (Lambercy et al., 2007),

but is not intended for monitoring purposes.

The hand agility (severely affected by rheuma-

toid arthritis and scleroderma) can be in principle

evaluated by means of ﬁnger tapping tests, originally

conceived to assess both motor speed and control in

neuropsychology. From the ﬁrst mechanical devices,

other approaches for the monitoring of this kind of ex-

ercise arose. Approaches including a passive marker-

based motion analyser (Jobb´agy et al., 2005) present

a very complex setup not suited for a fast evaluation.

Other approaches, based on sensorized gloves (Bus-

tamante et al., 2010), are uncomfortable for patients

with hand deformities caused by arthritis. In (Muir

et al., 1995), a touch system based on a 4-ﬁnger ac-

tive sensor (injecting on the hand a small sinusoidal

current at 1.5 kHz) has been presented along with its

support software. An App (Digital Finger Tapping

Test 1.0) with limited functionalities is also available

for IPhone users. An approach based on the detection

of the exertedforce in the tapping activity is presented

in (Macellari et al., 2006).

It is worth to note that, to the best of our knowl-

edge, the realization of a low cost device for the quan-

titative monitoring of both agility and strength ki-

nesitherapy exercises for rheumatic patients has not

been presented in literature until now.

Figure 1: A picture of the prototypical system.

A PORTABLE REAL-TIME MONITORING SYSTEM FOR KINESITHERAPIC HAND REHABILITATION

EXERCISES

3 THE SYSTEM AT A GLANCE

The system is conveniently packaged in a lightweight

metal briefcase, as shown in Figure 1. With the pro-

posed system, the patient can perform 4 exercises

with a single hand a time, with as many sensorized

devices.

There are 2 knobs on the vertical panel. The outer

one allows the evaluation of the patient manipulation

dexterity (exercise of dynamic rotation). The patient

must rotate as fast as possible the knob using his ﬁn-

gers, shaped in a pinch grasp, without any wrist ro-

tation and maintaining the forearm on the horizontal

plane. The inner knob allows to evaluate the clock-

wise and anticlockwise rotation torque (isometric ro-

tation exercise) with the same grasp type and restric-

tions of the previous exercise.

On the horizontal panel it is possible to perform

the other two exercises. One is a revised version of the

ﬁnger tapping exercise, which must be performed on

the exposed printed circuit board (PCB). The patient

must touch key-shaped pads on the PCB following a

speciﬁc sequence (little ﬁnger, ring ﬁnger, middle ﬁn-

ger, ﬁrst ﬁnger and thumb) as playing the piano. It is

allowed to have multiple ﬁnger on the keys provided

that the sequence is correctly performed and closed

with a thumb tapping. The last exercise allows eval-

uating the hand extension ability. The patient must

rest the hand between the two L-shaped aluminium

proﬁles, touching them with the thumb and the little

ﬁnger. Then he must open and close the hand (always

on the horizontal plane) in rhythm, allowing the sys-

tem to appreciate opening and closing agility. A small

counter-resistance is applied.

The system considers the exercise completed af-

ter a number of repetitions, previously established by

the physicians and hard coded in the system ﬁrmware,

have been executed. By using a GUI installed on his

PC, the physician can choose which exercise to exe-

cute, evaluating in real-time how the patient executes

it not only in terms of correct position but also look-

ing at barely perceptible execution parameters that the

digital system is able to reveal. For instance, a real-

time updated plot discloses sensors wave shape while

numerical data such as peak and running-average val-

ues are displayed on the GUI, allowing a ﬁner moni-

toring compared to a traditional visual inspection.

4 SYSTEM ARCHITECTURE

Figure 2 shows the most important parts of the system

and their interconnection. Beyond the MCU subsys-

tem controlling the whole system, we can see:

• the analogue sensorized devices;

• the digital sensorized device (for ﬁnger tapping);

• the analogue interface circuitry;

• a Bluetooth module, which provides a wireless

link to the host PC;

• additional components to provide a visi-

ble/audible feedback to the user.

The system can be easily supplied by a single-

cell Li-Ion battery. For improved safety, the inter-

nal battery can be recharged only when the system

is switched off. Along with the DC power supply

adaptor to recharge the internal battery, the Bluetooth

module (Bluegiga WT-11) is the only device that is

not embedded in the metal briefcase. The module

implements the Bluetooth stack and communicates

with the MCU via a standard UART port. A 25-pole

female D connector has been included providing a

clean way to access the Bluetooth module pins and

the JTAG ports to program the 2 MCUs embedded in

the system.

MSP430FG4618

Bluetooth

module

Feedback

Touch board

Signal

conditioning

stage

Sensors

UART I2C

GPIO ADC_IN

Figure 2: System block diagram.

4.1 The Sensorized Devices

The simplest device is that for the dynamic rotation

exercise. In this case, a precision multi-turn poten-

tiometer, equipped with a 30mm aluminium knob, has

been used. The potentiometer (Vishay 534, 20kΩ,

2W) is able to perform 10 turns opposing a torque

of 0.006 Nm. The resistance varies linearly with the

rotation of the knob so that it sufﬁces to measure the

voltage on the wiper to detect the angular position at

any instant. Due to the low opposing torque, the exer-

cise can be considered without any load.

On the contrary, the isometric rotation device is

composed of a 5-lobe 50mm plastic knob able to

slightly turn on its own axis pulling along with it a T

bar nut able to press one of two thin-ﬁlm force sensors

(the low-cost Tekscan FlexiForce A201, max 110N),

for clockwise and anticlockwise rotations. These sen-

sors linearly vary their conductance in response to the

applied force. Being an isometric exercise, thanks to

the aforementioned design, the knob cannot spin.

BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices

The hand extension exercise is dynamic but it

introduces a counter-resistance. It is evaluated by

means of an analogue draw wire position sensor (LX-

PA-15 by TME) mounted on a roller (CES30-88-ZZ

by Rollon) free to move on a 40cm linear zinc plated

guide (TES30-1040 by Rollon): the wire coming out

from the sensor is attached to a second roller mounted

on the same guide. The two rollers are attached to

as many L-shaped aluminium proﬁles actuated by the

patient opening and closing his hand. The sensor is

characterized by a nominal wire rope tension of 3.9N,

which must be overcome by the patient in order to

extend his hand.

Lastly, for the ﬁnger tapping exercise it has been

necessary to develop a capacitive touch board. Com-

pared to the one presented in (Muir et al., 1995), the

capacitive approach is still able to provide a detec-

tion of the touch without any counter-resistance from

the measuring device but also avoids any direct cur-

rent injection in the patient’s hand. The touch board

is based on the MSP430F2013 MCU, managing the

reading of the capacitance associated to 8 key-shaped

sensible areas on a PCB. The keys, which form a ca-

pacitor with the ground plane surrounding them, are

sequentially charged by the MCU, which is able to

measure the discharge time. Since the effect of touch-

ing a pad is the increase of the capacitance value, it

is easy to detect whether a sensor is touched or not,

comparing the measured discharge time with the base

value obtained when the pad is untouched. The de-

sign of this device followed the guidelines given in

(Albus, 2007) with some further consideration: the

layout of the board must accommodate both left and

right-handed exercises and the sensor shape should

lead to an ergonomic device (it should accommodate

different hand sizes and postures). Therefore the keys

were made slightly larger than the suggested value,

mesh-ﬁlled to keep the capacitance base value under

an acceptable level. The device provides over an I2C

bus, whenever required, the current status of the keys

in a single byte: the interpretation of the data in the

light of the exercise to execute is up to the main pro-

cessor ﬁrmware.

4.2 The Main Board

The main board, highlighted with a grey-shaded area

in Figure 2, hosts the MSP430FG4618 MCU, which

takes care of the actual processing and manages the

operation of the rest of the system. For the sake of

simplicity, a single power supply at 3.3V is available

on board.

The chosen processor embeds a 16-bit RISC CPU,

an 8kB SRAM, a 116kB ﬂash memory for program

storage, different I/O ports, a 12-bit multi-channel

ADC, three timers and other unused peripherals. It

is clocked at 1MHz by means of an external quartz

oscillator.

4.2.1 The Signal Conditioning Stage

Given the nature of the involved signals, which are

slowly time-varying, it is possible to operate at rather

low sampling frequencies, with consequent beneﬁts

in terms of real-time bounds for the signal processing

algorithms. On the other hand, the event detection

algorithms which underlie the system operations re-

quire an adequate time resolution: a fair trade-off be-

tween these two aspects led us to choose a sampling

frequency of 150Hz. The signal conditioning stage

must then implement a properly tuned anti-aliasing

ﬁlter.

The analogue interface block is essentially com-

posed of four non-inverting, active low-pass ﬁl-

ters, implemented with an operational ampliﬁer

(TLV2375) and a single pole RC net. The value of its

cut-off frequency has been set to about 48Hz to ex-

ploit the ﬁlter as anti-alias with guard band of about

25Hz under the Nyquist frequency, also limiting the

50Hz mains noise. The outputs of the four ﬁlters are

connected to as many different channels of the MCU

ADC.

V = 0.5V

out

Figure 3: FlexiForce sensor conditioning stage.

Two different conﬁgurations have been employed:

the one used for the FlexiForce sensors is depicted

in Figure 3. The sensor has been connected between

ground and the operational ampliﬁer inverting input,

making the stage a variable gain ampliﬁer. Using a

ﬁxed input, provided by a voltage reference at 0.5V,

the output varies linearly with the force applied to the

sensor (between 0.5 and 3.3V), according to:

out





= V

= K

(1)

where K

and K

are constants. Equation (1) shows

the linear dependency between output voltage and the

sensor conductance G

. The optimal value of R

has

A PORTABLE REAL-TIME MONITORING SYSTEM FOR KINESITHERAPIC HAND REHABILITATION

EXERCISES

been chosen in order to provide an adequate response

when the isometric rotation exercise is performed by

a rheumatic patient, even if this limits the operating

range of the sensor.

out

Figure 4: Potentiometer-based sensors conditioning stage.

The second conﬁguration, used for the potentio-

metric sensors and depicted in Figure 4, has a ﬁxed

gain and a variable input voltage. The sensors are in-

serted in a voltage divider, with the wiper connected

to the stage input, so that the output value is propor-

tional to the voltage present at the wiper. A series

resistor limits the current sunk but the side effect is

that the voltage at the wiper cannot reach V

, so a

gain greater than one must be used, which can be cal-

culated as

outmax

inmax

4.2.2 Patient Interface

The system includes low-level user interface elements

and some patient feedbacks, motivating him and aid-

ing a correct execution of the exercises. Two leds in-

dicate which hand must be used to execute the exer-

cise and another led gives a time reference blinking at

1Hz, which is useful for sustained position tests. They

are placed on the front panel for improved visibility.

Moreover, a buzzer chimes whenever the system de-

tects a successful event, letting the user know that the

system has effectively captured his action. The sys-

tem has been also provided with a double digit 7 seg-

ments display, which has different functions depend-

ing on the exercise. It displays:

• the percentage of the effort with respect to the

maximum bound (extension and torque),

• the number of correct sequences performed (ﬁn-

ger tapping),

• the percentage of rotation over 10 turns (dynamic

rotation).

Two buttons, white and red coloured, placed on

the horizontal plane and connected to two different

external interrupt pins of the MCU, provide a way for

the patient to interact with the device. The ﬁrst one

starts the exercise when the patient is ready, allowing

to correctly position the hand, whereas the second one

can be used to skip a single repetition of an exercise

(the whole exercise can be aborted from the GUI).

init_MSP430

check_BTooth

LPM0

init_resources

LPM0

exercise code

recieved from host

processing

LPM0

filter

detection

algorithm

send samples

to host

stop?

start signal

recieved

Figure 5: Firmware ﬂow diagram.

4.3 Firmware

The operation of the MCU is controlled by the

ﬁrmware loaded onto its ﬂash memory. This piece of

software is written in C and has been developed under

the CCS v4.0 IDE by Texas Instruments.

The ﬁrmware ﬂow, depicted in Figure 5 is quite

simple: as soon as all the initializations have been

carried out, the MCU enters the low power mode

(LPM). The rest of the processing is then managed

asynchronously by interrupt service routines (ISR). In

the ﬁrst phase:

• the USCI A port, devoted to communicate with

the Bluetooth module, is conﬁgured in serial

mode;

• the USCI B port, assigned to the communication

with the touch board, is conﬁgured in I2C master

mode;

• the timer A and timer B, respectively used to beat

the sampling time and generate the reference time

for the timing led, are conﬁgured;

• the ADC is set to perform a single conversion on

a single channel and initially left disabled;

• the general purpose pins are set.

All the resources present on the board, as opera-

tional ampliﬁers, ﬁnger tapping MCU and Bluetooth

module, are initially held in reset. Then the Bluetooth

module is set up and conﬁgured by setting the operat-

ing mode, device name and password. The ﬁrmware

enters an endless loop, where each iteration corre-

sponds to the execution of an entire exercise. Inside

the loop the MCU goes immediately in LPM (both

CPU and MCLK disabled), waiting for the host to

receive the execution code of the exercise to launch.

Receiving the exercise code triggers the USCI A port

BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices

ISR, which wakes up the MCU. Depending on the se-

lected exercise, some initializations are carried out,

the ADC input channel is set to the corresponding in-

put pin and it is started (except for the ﬁnger tapping

exercise). After that the MCU goes in LPM again,

waiting the start signal (by pushing the white button),

which unlocks the execution. From now on the pro-

cessing is timed by the timer A. In the corresponding

ISR, either the value at the ADC input is sampled and

stored or the new digital word from the tapping board

is read. A global counter is incremented, to keep track

of the number of samples gathered and hence to ex-

tract time measurements from it. Then the actual sig-

nal processing takes place on a sample-by-sample ba-

sis, in different ways depending on the speciﬁc exer-

cise, as explained in the following session. The cur-

rent sample is sent to the host machine through the

Bluetooth link, and only every second (150 samples)

a vector containing statistics which characterize the

execution is sent too. If the stop conditionwhich iden-

tiﬁes the end of an exercise is not met, the core enters

the LPM again from which it will be released by the

acquisition of a new sample, otherwise the processing

steps back to the main loop, entering in LPM until the

system gets triggered again from the GUI.

4.3.1 The Processing Algorithms

For all the exercises but the tapping one, the samples

are ﬁrst low-pass ﬁltered by an 8-tap moving average

ﬁlter in order to further smooth the signal.

For the extension and isometric rotation exercises

the algorithm simply detects the signal peaks corre-

sponding respectively to a hand extension or a torque

application. This is done by comparing each sample

with a threshold, which is computed by averaging the

ﬁrst ten samples acquired. This value is stored and

used as lower limit for the threshold, which is updated

after the detection of a new peak to 0.3× peak

value.

All the values are referred to a zero represented by

the initial condition of the sensorized device when

the user is ready to start. The peak event is validated

only if at least 75 consecutive samples are above the

threshold and only as soon as the samples go under

the threshold again. The peak maximum value, its du-

ration and position are determined and used to com-

pute their incremental mean values as:

¯m

( ¯m

N−1

(N − 1) + s)

(2)

where ¯m

is the mean value computed over i samples,

and s is the value of the new sample. The system also

stores the absolute maximum and minimum values for

the peak amplitude. It is worth to underline that the

variables which hold the average valuesare ﬂoat num-

bers, though the CPU is a 16 bit platform and ﬂoating

point is not supported in hardware. Nevertheless all

these operations are translated by the compiler in the

proper microcode without additional coding effort.

The algorithm is different in the case of the dy-

namic rotation, since different features are needed.

The typical signal has a terraced waveform, where the

edges correspond to the spinning of the potentiome-

ter whereas the plateaus indicate that the transducer

is still. The duration of both edges and plateaus, and

the amplitude of each edge, are computed. To de-

tect both onset and end of an edge, a simple detection

mechanism based on thresholds has been designed,

exploiting the smoothness of the ﬁltered signal. A

FIFO buffer of 14 samples is linearly updated at ev-

ery new sample. The mean value of the oldest 4 sam-

ples is computed and compared with the most recent

sample. If the difference is greater than an empiri-

cally determined threshold, the algorithm detects an

edge and marks the onset n samples before the most

recent one. When the difference falls back under the

threshold, the edge end is marked and the processing

is repeated, until the potentiometer reaches the limit.

By using absolute values, the processing is the same

for both clockwise and counter-clockwise exercises.

The ﬁnger tapping exercise differs from the oth-

ers because there are no analogue signals involved.

The MCU on the main board acts as the master of the

I2C channel, requesting the 8-bit word provided by

the sensorized device whenever the sampling timer

expires. For this exercise, the timer A has been dif-

ferently set, in order to have a sampling frequency

of 50Hz, which is in line with the state of the art

(Jobb´agy et al., 2005) and allows the complete scan-

ning of the 8 keys in a sampling period. As a new

word is received, it is mirrored, if necessary, in order

to have the least signiﬁcant bit always referred to the

thumb key. When the ﬁrst not null data is received,

the algorithm detects the less signiﬁcant bit set to 1

and creates a mask used, at the next touch, to check if

the next key tapped corresponds to a less signiﬁcant

bit or not. If this is true, the mask is updated and the

processing goes on, otherwise an error ﬂag is set. The

sequence terminates when the thumb touch is detected

(lsb = 1). If the number of touches is equal to ﬁve

the valid sequence counter is incremented or, if ei-

ther the error ﬂag is set or the sequence length differs

from ﬁve, the bad sequence counter is. This process-

ing is performed in real-time and when the exercise is

complete, an additional routine computes the relevant

statistics, including average touch duration for each

ﬁnger, average distance between them, total consecu-

tive touches and total duration of the exercise.

A PORTABLE REAL-TIME MONITORING SYSTEM FOR KINESITHERAPIC HAND REHABILITATION

EXERCISES

5 THE PHYSICIAN GUI

By using a user friendly GUI developed in Matlab, the

physician can monitor in real-time on a host PC the

execution quality of the kinesitherapic exercises, also

extracting useful information to evaluate the rehabil-

itation progress over time. As already said, the link

between the system and the host PC exploits a Blue-

tooth technology. The Bluetooth device driver exports

a serial interface towards the user applications, which

is easy to manage using the built-in Matlab functions.

At launch, the very ﬁrst window contains a list

of radio buttons enabling the selection of the exer-

cise and the hand to use. By pushing the start but-

ton on the GUI, a numerical code which identiﬁes the

chosen exercise and the hand to use is sent to the de-

vice. The callback function associated to this button

also creates a new window which is speciﬁc for the

selected exercise. All the exercise windows, except

that of ﬁnger tapping one, contain an area where the

raw signal acquired by the sensor can be plotted over

time. The signal is sent to the host PC on a sample-by-

sample basis. Since every sample is a 16 bit integer,

before reading the GUI waits inside a while loop the

availability of at least 2 bytes in the serial port input

buffer. The received samples are then converted to

the corresponding real physical quantity by means of

the calibration values. For the sake of efﬁciency, the

time plot is refreshed only when a block of 75 input

samples has been acquired, shifting towards left the

previous blocks in the plot linear buffer: the oldest

block is overwritten and the new one is inserted on

the right. All the received samples are logged thus, at

the end of the execution, the user can visualize a static

plot of the whole signal.

Figure 6: GUI screenshot.

The GUI also receives the peaks position detected

in real-time by the system and used to analyse the

signal, together with other relevant parameters (e.g.

speed of execution, position and the amplitude of the

last peak, the maximum, the minimum and the mean

value of the executions). These values are sent by

the system to the host PC only every 150 samples of

the signal and the most important ones are presented

on the GUI (Figure 6). Counting the received sam-

ple, the GUI is able to properly receive such param-

eters as a data chunk. A ﬂag at the end of the chunk

is used by the system to signal the end of the exer-

cise. The interface uses this ﬂag to allow the visual-

ization of the whole signal plot, including the markers

to the peaks found by the system during the execution.

The interface enables the visualization of the “speed-

value plot” (Figure 7), which overprints to a bar graph

showing the peak values, a line graph representing the

frequency of the repetitions. This information can be

useful to evaluate how much the performance is de-

pendent by the execution speed, being important to

know if smaller values achieved by the patient are

caused by a higher execution speed or by fatigue.

Figure 7: GUI speed-value plot.

In every window there is the possibility to stop

the execution by using a push button, whose callback

function sends a numerical code to the system in order

to signal the premature end of the exercise. Another

push button enables going back to the main window,

where the user can select a new exercise.

6 CONCLUSIONS

The portable kinesitherapic monitoring system pre-

sented in this paper is proposed as a support tool to ex-

ploit along with the latest treatment techniques in the

rheumatic patients hand rehabilitation practice. Com-

BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices

pared to other devices at the state of the art, the pro-

posed system presents several advantages. In fact it

embeds the sensorized devices necessary to execute

different kinds of exercisesin a single low-cost frame-

work, exporting the main real-time monitoring fea-

tures to a host PC via a wireless connection. Here,

an accurate analysis of the patient’s performances can

be easily performed, thanks to a user-friendly GUI.

In particular the real-time performance simpliﬁes the

physician’s task of evaluating and correcting the pa-

tient’s training, being immediately available quantita-

tive measurements also involving the time-related as-

pects of the exercise. In the next future the system is

going to be employed in clinical trials on rheumatic

patients, with the aim of verifying the effectiveness of

the approach and the usability. The system could be

further expanded including additional sensorized de-

vices, in order to offer a wider selection of exercises.

Furthermore, being a compact and portable device, it

could represent a good solution to delivery rehabilita-

tion services in the patient’s home.

ACKNOWLEDGEMENTS

The research leading to these results has received

funding from the Region of Sardinia, Fundamen-

tal Research Programme, L.R. 7/2007 “Promotion

of the scientiﬁc research and technological innova-

tion in Sardinia” under grant agreement CRP2 584

Re.Mo.To. Project. The authors wish to thank V.

Lussu, L. Piras, I. Secci, N. Zaccheddu and F. Boi

for their collaboration. A special acknowledgementto

Michele Crabolu for the development of the ﬁrst pro-

totypes of the ﬁnger tapping unit and the mechanical

realization of both the extension one and the briefcase

structure.

REFERENCES

Albus, Z. (2007). PCB-Based Capacitive Touch Sensing

With MSP430. Texas Instruments Inc. SLAA363A

Application report.

Andria, G., Attivissimo, F., Giaquinto, N., Lanzolla, A.,

Quagliarella, L., and Sasanelli, N. (2006). Func-

tional evaluation of handgrip signals for parkinsonian

patients. IEEE Transactions on Instrumentation and

Measurement, 55(5):1467–1473.

Brokaw, E. B., Black, I., Holley, R. J., and Lum, P. S.

(2011). Hand spring operated movement enhancer

HandSOME: A portable, passive hand exoskeleton for

stroke rehabilitation. IEEE Transactions on Neural

Systems and Rehabilitation Engineerig, 19(4):391–

399.

Bustamante, P., Grandez, K., Solas, G., and Arrizabalaga,

S. (2010). A low-cost platform for testing activities

in parkinson and ALS patients. In 12th IEEE Inter-

national Conference on e-Health Networking Appli-

cations and Services (Healthcom), pages 302–307.

Dovat, L., Lambercy, O., Gassert, R., Maeder, T., Milner,

T., Leong, T. C., and Burdet, E. (2008). HandCARE:

A cable- actuated rehabilitation system to train hand

function after stroke. IEEE Trans. on Neural Systems

and Rehabilitation Engineering, 16(6):582–591.

Helliwell, P., Howe, A., and Wright, V. (1987). Functional

assessment of the hand: reproducibility, acceptability,

and utility of a new system for measuring strength.

Ann Rheum Dis, 46:203–208.

Huang, Y. and Low, K. (2008). Initial analysis and design of

an assistive rehabilitation hand device with free load-

ing and ﬁngers motion visible to subjects. In IEEE In-

ternational Conference on Systems, Man and Cyber-

netics, SMC 2008, pages 2584–2590.

Iqbal, J., Tsagarakis, N., Fiorilla, A., and Caldwell, D.

(2010). A portable rehabilitation device for the

hand. In 2010 Annual International Conference of the

IEEE Engineering in Medicine and Biology Society

(EMBC), pages 3694–3697.

Jobb´agy, A., Harcos, P., Karoly, R., and Fazekas, G. (2005).

Analysis of ﬁnger-tapping movement. Journal of Neu-

roscience Methods, 141:29–39.

Lambercy, O., Dovat, L., Gassert, R., Burdet, E., Teo, C. L.,

and Milner, T. (2007). A haptic knob for rehabilitation

of hand function. IEEE Transactions on Neural Sys-

tems and Rehabilitation Engineering, 15(3):356–366.

Macellari, V., Morelli, S., Giacomozzi, C., Angelis, G. D.,

Maccioni, G., Paolizzi, M., and Giansanti, D. (2006).

Instrumental kit for a comprehensive assessment of

functional recovery.

Muir, S. R., Jones, R. D., Andreae, J. H., and Donaldson,

I. M. (1995). Measurement and analysis of single and

multiple ﬁnger tapping in normal and parkinsonian

subjects. Parkinsonism related disorders, 1(2):89–96.

Peters, M. J. H., van Nes, S. I., Vanhoutte1, E. K., Bakkers,

M., van Doorn, P. A., Merkies, I. S. J., and Faber,

C. G. (2011). Revised normative values for grip

strength with the jamar dynamometer. Journal of the

Peripheral Nervous System, 16:47–50.

Seo, N. J., Rymer, W. Z., and Kamper, D. G. (2009).

Delays in grip initiation and termination in persons

with stroke: Effects of arm support and active muscle

stretch exercise. J Neurophysiol, 101(6):3108–3115.

A PORTABLE REAL-TIME MONITORING SYSTEM FOR KINESITHERAPIC HAND REHABILITATION

EXERCISES