MEASURING INTENTION TREMOR IN MULTIPLE SCLEROSIS

USING INERTIAL MEASUREMENT UNIT (IMU) DEVICES

Laurence P. Ketteringham, Simon A. Neild, Richard A. Hyde

University of Bristol, Dept. Mechanical Engineering, Queen’s Building, University Walk, Bristol, BS8 1TR, U.K.

Rosemary J. S. Jones, Angela M. Davies Smith

Bristol & Avon Multiple Sclerosis (BrAMS) Centre, Frenchay Hospital, Bristol, BS16 1LE, U.K.

Keywords: Measurement, Tremor, Multiple sclerosis, MS, Inertial measurement unit, IMU.

Abstract: This paper describes research to create a sensor based measurement system in order to provide detailed and

accurate data on the movement disorder known as intention tremor, a condition that affects a significant

proportion of individuals with multiple sclerosis. Intention tremor is a complex movement disorder that

worsens during goal directed movements and can therefore be extremely disabling. Multiple inertial

measurement unit devices were used to measure the upper limb of subjects with multiple sclerosis and

intention tremor during standard clinical finger-to-nose tests and reach-retrieve tasks, which were designed

to mimic activities of daily living. Analyses allowed information on tremor characteristics to be ascertained

during these movements. The equipment and software provide a useful tool for clinical assessment of

tremor, displaying a variety of relevant information at differing levels of detail, obtainable at several points

over the torso, shoulder, upper arm, lower arm and hand. Examples of this data are discussed. The system

allows tremor assessment in more detail than is possible with clinical tests that rely on visual assessments,

and provides a tool that can accurately assess the benefits of future tremor reduction devices, or other

interventions.

1 INTRODUCTION

MS is the most common disabling neurological

condition affecting young adults (Calabresi, 2004).

The onset of symptoms is typically between 20 and

45 years of age. More than 2.5 million individuals

worldwide have MS. MS leads to substantial

disability in more that 50% of patients (Prat, et al.,

2002).

Though currently incurable, several treatments

are available which may slow the appearance of new

symptoms and reduce the severity of existing ones.

Movement disorders, including tremor, often

occur as a result of MS. Tremor is estimated to

affect 75% of people diagnosed with MS; it can be

severely disabling and extremely difficult to treat

(Alusi, et al., 1999). A form of tremor called

intention tremor (sometimes described as, or as a

sub-category of, kinetic tremor) is especially

common in MS. This tremor exhibits itself during

purposeful movement, such as reaching out or

picking up an object, rather than at rest. It is often

accompanied by slower, uncoordinated movements

and a tendency to overshoot or undershoot targets,

referred to as ataxia and dysmetria, respectively

(Alusi, et al., 1999). Muscle weakness and sensory

impairments are also found (Alusi, et al., 2000);

these are likely to further complicate the movement

disorder. Wrist, elbow and shoulder tremors were

found to be particularly disabling when they

occurred during activities of daily living (ADL)

(Alusi, et al., 2001).

Initial work in this area discussed the use of

miniature accelerometer and gyroscope micro-

electromechanical systems (MEMS) sensors for

measuring positions and movement (Hyde, et al.,

2008). Such sensors have been combined into

devices that provide three orthogonal axes each of

accelerometer, gyroscope and magnetometer

measurements, together with microprocessors that

convert these readings into 3D orientations in space.

The early development of these sensors is detailed in

Luinge, et al., 2004 Luinge, et al., 2005. These have

since been developed into commercial devices and

204

P. Ketteringham L., Neild S., Hyde R., S. Jones R. and Davies Smith A..

MEASURING INTENTION TREMOR IN MULTIPLE SCLEROSIS USING INERTIAL MEASUREMENT UNIT (IMU) DEVICES.

DOI: 10.5220/0003156002040211

In Proceedings of the International Conference on Biomedical Electronics and Devices (BIODEVICES-2011), pages 204-211

ISBN: 978-989-8425-37-9

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

several publications discuss their use in human

measurement, monitoring and rehabilitation

applications (e.g., Heinz, et al., 2006, Beauregard,

2007, Moore, et al., 2007 and Zhou, et al., 2006).

2 MOVEMENT MEASUREMENT

Several commercial systems that can provide 3D

orientations in this manner are discussed in

Ketteringham, 2010. The sensors used in this work

were chosen for their ability to provide 3D

orientation measurements in real-time, in an easily

applied, unobtrusive system. These data can be used

to give the position and movement of body segments

in space.

This research used a set of 5 MTx sensors (in an

Xbus kit, Xsens Technologies, The Netherlands).

The computed 3D orientation from each of the 5

sensors was recorded over time during each run. The

MTx sensors communicated with a battery powered

Xbus Master unit that was worn on the body and

communicated wirelessly with a PC.

The five sensors were positioned over the body

surfaces at the torso, shoulder, upper arm, lower arm

and hand, allowing the orientation of each body

segment to be found at each point in time. The base

of the torso was assumed to be static. Applying the

measured body segment orientations to the known

body segment dimensions allowed calculation of the

body segment displacements, along with the

displacements of points in space that move in

relation to the body segments (e.g. a held object).

The relative orientations between segments (joint

rotations) can also be found, and these can each be

converted into a set of three rotations (Euler angles)

that represent the rotation of each the joint around

three orthogonal axes. These angles could prove

useful in clinical practice if there was a requirement

to study a specific joint angle, or set of joint angles.

The displacement and angle data can also be

differentiated with respect to time to produce

velocities and accelerations, and angular rates and

angular accelerations. Further analyses can reveal

the frequency content of the movements. The joint

angles, angular rates and angular accelerations have

also been used in inverse dynamics models, created

in the Matlab

SimMechanics™ toolbox, for

estimating joint torques during the movements, as

described in Ketteringham, 2010 and Ketteringham,

et al., 2008.

The movement measurement system therefore

provides a useful tool for measuring tremor levels

and movement capabilities in individuals during

clinical assessments. It allows quantitative

assessment of the efficacy of other treatments in a

less subjective manner and in far more detail than is

possible with visual assessment alone.

3 EQUIPMENT AND

EXPERIMENTAL DESIGN

The Xbus Master was wirelessly interfaced via

Bluetooth

to a portable PC running code written in

Matlab, to obtain orientation and sensor data from

the MTx sensors. Data were obtained at 50 Hz for

each of the 5 MTx sensors.

3.1 Initial and Clinical Measurements

Ethical approval was obtained before carrying out

these studies, and all subjects completed consent

forms at or prior to the time of testing.

The subjects were initially tested to determine

their symptom characteristics, before attachment of

sensors. The initial clinical tests consisted of:

 A check for pain-free range of joint movement;

 An evaluation of eyesight;

 Muscle strength evaluation assessed on a

Motricity scale;

 A Fahn test for upper limb tremor and ataxia.

These joint range of movement, eyesight and

muscle strength tests were carried out to ensure that

the subjects could comfortably carry out the tasks

required during the recorded movements. Further

details are given in Ketteringham, 2010.

The Fahn test consisted of a tremor assessment

by observation while the subject held a pose, to

assess postural tremor, and performed a finger-to-

nose test, to assess kinetic and intention tremor. The

subject was asked to hold a pose and perform these

actions while a clinician observed the resulting

movements.

Resting tremor was assessed with the subject

sitting upright and fully supported by a high-backed

chair, with arms fully supported against gravity on

the chair’s armrests. Postural tremor was assessed

while the subject sat upright, with their arms held

out in front of them (elevated to 90° flexion) so that

the upper arm, lower arm and hand were horizontal,

and pointed directly forwards, with the hands

shoulder width apart and palms faced downwards.

Kinetic and intention tremor were assessed during a

finger-to-nose test, described below. The movement

towards the nose in the finger-to-nose test was used

to assess kinetic tremor, while the part of the finger-

MEASURING INTENTION TREMOR IN MULTIPLE SCLEROSIS USING INERTIAL MEASUREMENT UNIT (IMU)

DEVICES

205

to-nose test where the finger approached and briefly

remained at the nose was used to assess intention

(goal oriented) tremor. Feys et al., 2003 discuss the

reliability of the observed finger-to-nose test for

rating tremor.

Body segment dimensions were measured before

attachment of the sensors, as described in

Ketteringham, 2010.

3.2 Sensor Positioning

The MTx sensors were placed as follows, locating as

flat an attachment position under each sensor as

possible:

1. On the midline of on the torso, over the upper

part of the body of the sternum, pointing

superiorly (upwards);

2. On the superior aspect (top) of the shoulder, on

a flat region (where one could be found),

medial to the acromion process, pointing

laterally (to the side, along the shoulder);

3. On the distal end of the upper arm (the end of

that segment that is closest to the hand),

proximal to the lateral epicondyle, pointing

distally (towards the hand end of the arm);

4. On the distal end of the lower arm, proximal to

the wrist, on the dorsal surface (at the position

that a watch would normally be worn),

pointing distally;

5. On the dorsal surface (back) of the hand,

pointing medially (inwards, towards a plane

that divides the left and right sides of the

body).

These sensor locations were chosen, in

consultation with medical professionals, as being

positions on the body where minimal skin movement

artefacts would occur, due to movement of the

flexible and compressible soft tissues that cover the

more rigid skeletal structures beneath.

Sensor positions can be seen in Fig. 1 (with the

exception of the sensor on top of the shoulder).

Sensors on the top of the shoulder, arm and hand

were all positioned so that they were as horizontal as

possible when the arm was in the starting position

(described below).

The MTx sensors were adhered to the skin

surface using PALstickies™ hydrogel adhesive pads

(PAL Technologies Ltd, UK) in order to prevent the

sensors moving relative to the skin surfaces.

MaxWrap™ silicone elastic straps (La Pointique

Int’l Ltd., USA) were wrapped around the upper

arm, lower arm and hand over the sensors to ensure

minimal movement between the sensors and the

underlying rigid body structures. The straps are not

shown in Fig. 1, for clarity of the sensor positioning.

Figure 1: MTx sensor positions on the torso, shoulder (at

the top of the image, under clothing), upper arm, lower

arm and hand.

The Xbus Master data transmitter, which the

sensors were attached to, is shown as the white box

at the bottom of Fig. 1, attached to the subject with

the supplied waistband.

3.3 Sensor Calibration and Starting

Position

At the beginning of each movement test, which

consisted of three replicates of each type of

movement, the torso, shoulder, arm and hand were

stabilised in the starting position (described below)

for six seconds to allow the sensor readings to

stabilise and to obtain the initial sensor orientations.

The physiotherapist stabilised the arm in the starting

position by grasping the hand and the elbow, and

maintained as steady and stable a position as was

possible. The pose was held as steadily as possible

during this time, despite some subjects finding it

difficult to maintain a completely relaxed, steady

pose, even when fully supported and stabilised by

the physiotherapist.

The starting pose was defined as:

 The torso held as upright as possible facing

forwards;

 The shoulder and upper arm longitudinal axes

pointed laterally (to the side);

 The lower arm and hand longitudinal axes

pointed to the front;

 The fingers pointed forwards, with the hand

held flat, palm faced downwards.

All body segments were held as horizontal as

possible (except the torso, which was held vertical),

with orientations as close to the world X-, Y- and Z-

axes as possible, as shown in Fig. 2. It was the

central axis of each segment (considered to be the

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

206

line that directly linked the joints between the body

segments) that was held aligned to the world X-, Y-

and Z-axes when in this starting pose.

This starting pose was easy for the

physiotherapist to maintain and had the additional

advantage of starting each axis of rotation, of each

joint, at a position that was close to the centre of its

range of movement. This meant that the joint angles

would be much less likely to vary by as much as

±90° from this position during these movement tests,

which is a great advantage when working with Euler

angles, as discontinuities can be achieved if the

middle angle of a set of Euler angles reaches ±90°.

Figure 2: Representation of the body segments for the left

arm (thick, purple lines), in the starting position, viewed

from behind and to the left. Sensors are shown as orange

wireframe boxes, in their average orientations during this

period, and in the approximate positions relative to each

body segment. World axes are shown as red, green and

blue arrows (X-, Y- and Z-axes, respectively) at the base

of the torso. Local axes of each sensor are shown in the

same colours, emanating from the centre of each sensor’s

base. The thin green line connects the point of interest

(green dot, described below) with the hand.

Average sensor orientations were obtained

during the sixth second of measurement. Since the

orientation of each body segment was known during

this time, the orientation offset between each sensor

and the body segment that it was attached to could

be found. Each of these orientation offsets was

removed from the orientations measured at each

sensor over the entire data recording period,

providing the orientations of the body segments

themselves.

Once the orientations of the body segments were

calculated, the displacements of the body segments

in space, relative to the origin at the base of the

torso, were found by multiplying the rotation

matrices representing the body segment orientations

with vectors representing the dimensions of the body

segments, then attaching the rotated body segment

vectors end-to-end, to provide a model of the body

segments like that in Fig. 2.

3.4 Movements Measured using

Sensors

Both clinical finger-to-nose and reach-retrieve tests

were measured using the sensors. The reach-retrieve

tests were designed to measure movement and

tremor during task-oriented (“functional”) everyday-

type movements. The movements were

unconstrained to make them as representative as

possible of movements in ADL.

The clinically assessed, and sensor-measured,

finger-to-nose tests used in these tests were similar

to those tested in Feys, et al., 2003 (which compared

a variety of methods). The movement sequence is

described below:

1. Starting at a holding position, with the arm

outstretched, and all limb segments pointing

laterally (to the side), with the elbow fully

extended, all arm segments held as horizontally

as possible;

2. Flexing the elbow to move the finger to the

nose;

3. Arriving at the nose and holding the finger at

the nose (or as close to it as possible) for a

short time;

4. Extending the elbow, and returning the arm to

the outstretched position.

The physiotherapist stood in front of the subject

and demonstrated the required movement before the

tests began. The clinical finger-to-nose test was

observed from in front of the subject.

The reach-retrieve movement required the

subject to move a ball between a near and a far cup,

in the sequence described below:

1. Starting at a resting position moving the hand

forwards to pick up the ball, initially situated

at the far cup;

2. Carrying the ball from the far cup to the near

cup and depositing it;

3. Moving back to the rest position after

depositing the ball, and resting briefly at the

rest position;

4. Moving forwards, from the rest position to pick

up the ball, now situated at the near cup;

5. Carrying the ball from the near cup to the far

cup and depositing it;

6. Moving back to the rest position after

depositing the ball.

The subject rested their arm before, between and

after each reach-retrieve part in a position where it

MEASURING INTENTION TREMOR IN MULTIPLE SCLEROSIS USING INERTIAL MEASUREMENT UNIT (IMU)

DEVICES

207

was completely supported by the armrest of the chair

in which they were seated. This allowed their tremor

to subside between movements, which gave visual

pointers to the duration of the movements in the

resulting data and provided a relatively static period

in which the sensors could reset the drift that can

occur in the readings during higher frequency

movements.

The balls in the reach-retrieve tests were moved

forwards and backwards between shallow cups,

which stabilised their position when being picked up

and deposited and provided a clear goal of where the

ball should be deposited, fixing the positions that the

ball was moved between. The far cup was positioned

at a distance in front of the subject that provided a

comfortable full reach, while the near cup was

positioned 5 cm from the edge of the table. Both

cups were positioned directly in front of each

subject’s shoulder joint. The distance between cups

depended on the subject’s stature, and varied from

14 to 20 cm.

3.5 Selecting a Point of Interest

A “point of interest” (POI) was chosen in order to

simplify study of the measured movements. This

was considered to be the part of the body, or a

position relative to a body part, that the subject was

concentrating on during the tests. For this reason, the

POI was considered to be in a different position

depending on the test being carried out. The POI

was considered to be the tip of the middle finger if a

finger-to-nose test was being carried out, or below

the palm of the hand (as shown in Fig. 2.), at the

position that the centre of the ball would be when

the ball was being carried in the hand. This POI

rotated with the rotation of the hand, and so it was

always in the same position relative to the hand.

Some example data, showing the displacement

of the POI in space, are discussed below.

3.6 Data Analysis and Filtering

In order to study the tremor in the movements, the

displacements of the POI in space were filtered to

leave only the high frequency components. A high-

pass fifth order Butterworth filter with a cutoff

frequency of 2 Hz was applied to the data in

forwards and reverse directions, to provide acausal

filtering. This filter gave a sharp separation between

the high and low frequencies, with minimal ripple in

the passband, and 50% attenuation of the data at

frequencies of 2 Hz (as it was applied to the data

twice).

The X-, Y- and Z-axis displacement data were

filtered separately, to leave only the high frequency

components that were superimposed on top of the

lower frequency voluntary (intentional) and ataxic

movements.

While the alignment of the tremor with the X-,

Y- and Z-axes can indicate, to some degree, which

joints are involved in the tremor movements, it may

be better, from a clinical point of view, to simplify

the data by combining the tremor displacements in

the three X-, Y- and Z-axes into a single measure of

tremor displacement, in an arbitrary direction. These

data represent the magnitude of the tremor at any

point in time, and can be calculated as a “3D

hypotenuse” of individual displacements in the X-,

Y- and Z-axes:

,, ZtYtXtmt

dddd ++=

(1)

where d

t,m

is the tremor displacement magnitude

in an arbitrary direction, and d

t,X

, d

t,Y

and d

t,Z

are the

tremor displacement magnitudes in the individual

X-, Y- and Z-axes. This reduces the data set

threefold, while maintaining information on the

magnitude of tremor, in an arbitrary direction.

Obtaining the tremor magnitude in an arbitrary

direction requires squaring the data, however,

leading to loss of sign information. This can give

inaccurate results when analysing the resulting

rectified data for frequency information (as an

example, a rectified and non-rectified sine wave

contain different frequency components). A method

was created to negate portions of the resulting data,

in order to return it to being a waveform that

represented the tremor movements alternating to

“either side” of the lower frequency data, as

discussed in Ketteringham, 2010.

Windowed power spectra were calculated at

regular intervals through the resulting tremor

magnitude data. The power spectra were calculated

at time increments of 0.3 s, with a time window size

of 1.3 s. This window size was found to be optimal

in terms of producing good frequency and time

resolution; a shorter window was more responsive to

rapid changes in movement while a longer time

window gave better differentiation of the

frequencies contained in the data. The power

spectrum obtained from each analysed window of

data was normalised to the maximum power in that

time window, as this provided a clearer indication of

the dominant frequencies throughout the data. The

power spectrum analysis was limited to a maximum

of 8 Hz.

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

208

4 RESULTS: EXAMPLE DATA

FROM FINGER-TO-NOSE AND

REACH-RETRIEVE TESTS

The following results are example data from a

subject completing the first of three replicates of a

finger-to-nose test and a reach-retrieve test, in Fig. 3

and Fig. 4, respectively. The data represent the

movements of the POI in space, analysed as

described above, shown as a contour plot of the

normalised, windowed power spectra, together with

a concurrent plot of the displacement data.

Figure 3: Tremor frequencies and displacements of the

POI during a finger-to-nose test.

Figure 4: Tremor frequencies and displacements of the

POI during parts 1 to 3 of a reach-retrieve test.

The dotted line that passes through the power

spectra from left to right indicates the frequency

with the maximum power in each of the time

windows. The vertical dash-dot lines in Fig. 3

(nominally) separate the four regions of the finger-

to-nose test, as described above. Fig. 4 shows only

parts 1 to 3 (of parts 1 to 6, described above) of a

reach-retrieve test, as the data in parts 4 to 6 were

fairly similar to the first three parts, and the duration

of parts 1 to 3 was similar to that of one replicate of

the finger-to-nose test, giving a better comparison.

Again, the parts are separated by vertical dash-dot

lines.

The lightness of the regions on the contour plot

shows the relative power of the particular

frequencies at a particular time, where a darker

region indicates the dominant frequencies (those

with a higher power). No power magnitude (colour

bar) is shown with the contour plot, as the maximum

power values in each time window were normalised,

so the maximum “elevations” are of value 1.

High frequency Euler angle data (three

orthogonal rotations for each joint in the body

model) are also shown, in Fig. 5 and Fig. 6. The high

frequency angle data were obtained from the Euler

angle data by applying the same high-pass, acausal,

fifth order Butterworth filter, with a cutoff frequency

of 2 Hz, as described above.

Figure 5: High frequency components of the joint angles

during a finger-to-nose test.

Figure 6: High frequency components of the joint angles

during a reach-retrieve test.

MEASURING INTENTION TREMOR IN MULTIPLE SCLEROSIS USING INERTIAL MEASUREMENT UNIT (IMU)

DEVICES

209

The subscripts Ts, Sh, UA, LA and Ha in the

legends indicate the particular joint that the torso,

shoulder, upper arm, lower arm and hand segments

are rotating about. X, Y and Z indicate the axis that

the rotation is about. These rotations were aligned

with the world X-, Y- or Z-axis, respectively, when

the body was held in the starting position. This is

discussed in more detail in Ketteringham, 2010.

5 DISCUSSION OF RESULTS

The tremor frequencies found in the data were

mainly between 3 and 5 Hz. This agrees with

descriptions given in several previous studies (e.g.

Alusi, et al., 2001, Gillard, et al., 1999 and Deuschl,

et al., 1998). The tremors described in such

publications are often reduced to simple values,

without reference to changes over time. These data

show that the tremor frequencies are by no means

constant throughout an unrestricted, full arm

movement, but vary due to the configuration of the

arm, and may be affected by the stiffness of the

joints due to muscle co-contraction. There are also

points in the data where the tremor movements are

not at a single, dominant frequency, but are

distributed over a range of frequencies.

Many previous studies have restricted movement

to one joint axis, one joint or a small range of joints.

In these cases a more limited range of frequencies

may be witnessed. These kinds of studies can not

represent the tremor as seen in full, unrestricted

movement, though. Measurements of unrestricted

movements are arguably more useful when studying

movement in ADL, in order to ascertain

improvements in tremor due to interventions or other

methods for controlling tremor.

Some frequencies below 3 Hz can be seen during

the more restful periods at the beginning and end of

the reach-retrieve movements. These parts also had

low amplitudes in the tremor, as tremor subsided on

cessation of movement. At other times during the

reaching task, the tremor was relatively stable in

terms of frequency and amplitude, though the

amplitude varied somewhat over a period of 2-3

cycles, probably due to changes in the configuration

of the arm during the task and the interactions

between the tremors that arose in the joints. Similar

characteristics can be seen in the joint angles in Fig.

6. The tremor took 2-3 cycles to reach full amplitude

on beginning the task.

The finger-to-nose test showed quite different

characteristics. While the tremor was of a similar, or

slightly higher, frequency during the parts where the

finger was held at or near the nose, larger, lower

frequency tremor occurred at the beginning and end

of the movements.

At the beginning of each finger-to-nose

movement, the subject held their arm out to the side.

A typical stabilisation tactic for subjects with tremor

is to “lock” the lower arm and wrist joints at the end

of their range of movement while the arm is in this

position, with the elbow (Z

in Fig. 5 and Fig. 6)

extended, the lower arm (Y

) supinated (to turn the

palm upwards) and the wrist (X

) extended. This

was the case during this test.

Little tremor can originate at the elbow, lower

arm or wrist when in this position, but it can once

the movement to the nose has commenced and the

joints can move more freely. The arm is also longer

and more rigid in this configuration, making it act

like a longer pendulum of larger mass, compared to

when the elbow, especially, is more flexed. These

factors can lead to lower frequency, larger

magnitude tremor displacements at the hand.

Similar features can also be seen in the joint

angles in Fig. 5, and the tremor can be seen to

originate more from the torso and shoulder at the

beginning and end of the test movements, where the

distal joints are less free to move.

6 CONCLUSIONS

This kind of coupled measurement and analysis

system can be seen to provide detailed and useful

data for measuring and investigating movement and

tremor.

The system can be used to generate a wide

variety of data, including displacements of the body

segments in space and joint angles. As discussed

earlier, these data can be further analysed to produce

kinematic data sets which can be used to drive

inverse dynamic models to obtain joint torque

estimates.

On their own, however, these displacements can

still be a useful measure of the tremor present in an

individual's movements, and should prove to be a

useful tool in clinical practice, whether used for

assessment of tremor progression or for assessment

of the effects of an intervention to improve tremor.

The different movements in the two tests

measured here result in observable differences in the

resulting tremor. The characteristics of the tremor

change more during the finger-to-nose test than

during the reach-retrieve test. The configuration of

the arm when fully extended can be seen to lead to

lower frequency, larger amplitude tremors. These

BIODEVICES 2011 - International Conference on Biomedical Electronics and Devices

210

are similar to those seen in a typical clinical

“postural” tremor test, where tremor is observed

while the arms are held out, fully extended in front

of the body. The tremor can not occur to such a

degree in (especially) the elbow when in this

position. The extended arm represents a longer

pendulum of larger mass than when the elbow is

flexed, so a lower frequency movement results,

emanating from the torso and shoulder joints.

It could be argued that the positions held, and

movements made, in the finger-to-nose tests are not

particularly “functional” (i.e. representative of a

typical everyday action, or ADL), and that the

tremor seen during a task with the arm held

outstretched is somewhat an artefact of the position

that the arm is held in.

Movements such as the reach-retrieve task

described here could be said to be more

representative of ADL. The characteristics of the

tremor during those tasks were relatively consistent

throughout, and there was no opportunity for joints

to be “locked” at the limit of their range of motion.

ACKNOWLEDGEMENTS

This research was supported by the UK Engineering

and Physical Sciences Research Council (EPSRC)

under a Doctoral Training Account, EP/P501326/1,

and by the UK charity MS Research.

REFERENCES

Alusi, S. H., Glickman, S., Aziz, T. Z., Bain, P. G., 1999.

Tremor in multiple sclerosis. Journal of Neurology,

Neurosurgery, and Psychiatry, 66, pp.131-134.

Alusi, S. H., Worthington, J., Glickman, S., Findley, L. J.,

Bain, P. G., 2000. Evaluation of three different ways

of assessing tremor in multiple sclerosis. Journal of

Neurology Neurosurgery and Psychiatry, 68, pp.756-

760.

Alusi, S. H., Worthington, J., Glickman, S., Bain, P. G.,

Apr 2001. A study of tremor in multiple sclerosis.

Brain, 124(4), pp.720-30.

Beauregard, S., 2007. Omnidirectional pedestrian

navigation for first responders, in: 4th Workshop on

Positioning, Navigation and Communication 2007,

(WPNC07), Hannover, Germany.

Calabresi, P. A., 2004. Diagnosis and management of

multiple sclerosis. American Family Physician,

70(10), pp.1935-1944.

Deuschl, G., Bain, P., and Brin, M, 1998. Consensus

statement of the movement disorder society on tremor.

Movement Disorders, 13(S3), pp.2–23.

Feys, P. G., Davies-Smith, A., Jones, R., Romberg, A.,

Ruutiainen, J., Helsen, W. F., Ketelaer, P., 2003.

Intention tremor rated according to different finger-to-

nose test protocols: A survey. Archives of Physical

Medicine and Rehabilitation, 84, pp.79-82.

Gillard, D. M., Cameron, T., Prochazka, A., Gauthier, M.

J. A., 1999. Tremor suppression using functional

electrical stimulation: a comparison between digital

and analog controllers. IEEE Transactions on

Rehabilitation Engineering, 7(3), pp.385–388.

Heinz, E., Kunze, K., Gruber, M., Bannach, D., Lukowicz,

P., 2006. Using wearable sensors for real-time

recognition tasks in games of martial arts - an initial

experiment. In: 2006 IEEE Symposium on

Computational Intelligence and Games, Reno, NV.

pp. 98-102.

Hyde, R. A., Ketteringham, L. P., Neild, S. A., Jones,

R.J.S., 2008. Estimation of Upper-Limb Orientation

Based on Accelerometer and Gyroscope

Measurements, IEEE Transactions on Biomedical

Engineering, 55(2), pp.746-754.

Ketteringham, L. P., 2010. Measuring and Controlling

Upper Limb Tremor in Individuals With Multiple

Sclerosis. PhD thesis. Department of Mechanical

Engineering. University of Bristol, UK.

Ketteringham, L. P., Neild, S. A., Hyde, R. A., Jones, R. J.

S., Davies Smith, A., 2008. Intention tremor in

multiple sclerosis: measuring and modelling arm

dynamics and elbow torque. In: Proceedings of the

ASME International Mechanical Engineering

Congress & Exposition, 35, pp.19-26. Boston,

Massachusetts, USA. Paper number: IMECE2008-

66140.

Luinge, H. J., Veltink, P. H., 2004. Inclination

measurement of human movement using a 3-d

accelerometer with autocalibration. IEEE Transactions

on Neural Systems and Rehabilitation Engineering,

12, pp.112-121.

Luinge, H. J., Veltink, P. H., 2005. Measuring orientation

of human body segments using miniature gyroscopes

and accelerometers. Medical & Biological

Engineering & Computing, 43, pp.273-282.

Moore, S., H. G., M., J. M., G., H. S., C., W. G., O., 2007.

Long-term monitoring of gait in Parkinson’s disease.

Gait & Posture, 26, pp.200-207.

Prat, E., Martin, R., 2002. The immunopathogenesis of

multiple sclerosis. Journal of Rehabilitation Research

and Development, 39(2), pp.187-199.

Zhou, H., Hu, H., Tao, Y., 2006. Inertial measurements of

upper limb motion. Medical and Biological

Engineering and Computing, 44, pp.479-487.

MEASURING INTENTION TREMOR IN MULTIPLE SCLEROSIS USING INERTIAL MEASUREMENT UNIT (IMU)

DEVICES

211