STATE OF THE ART

Biomechanica Hungarica II. évfolyam, 2. szám

VIRTUAL REHABILITATION – A FOCUS ON MOVEMENT FUNCTION

Gábor Barton
Research Institute for Sport and Exercise Sciences
Liverpool John Moores University
[email protected]

Abstract
For a long time virtual reality had been an exciting concept commonly associated with science
fiction rather than science but with the accelerated development of technology now we have an
accessible tool which is used in a multitude of applications. Rehabilitation is one of the fields
where physical or intellectual interaction of the patient with a reactive environment can lead to
health benefits and so the concept of “virtual rehabilitation” enjoys a lot of attention. The com-
plementary expertise of health professionals, biomechanists, engineers and computer scientists
working together in virtual rehabilitation allows us to formalise research questions related to
movement function and the answers can help individuals coping with their impairments in a
daily living context. This article describes our initial steps on a long and exciting journey.

Keywords: virtual reality, rehabilitation, movement analysis, cerebral palsy, targeted training,
core control

Introduction Institute for Health Research has consoli-

dated the theme of Virtual Rehabilitation.
Our Institute received major funding from A brief overview of the past five years will
the UK’s Science Research Investment Fund cover the technically focused initial work fol-
in 2004 to establish a cutting edge scientific lowed by the more applied research projects
laboratory aiming to extend research towards closing with an indication of future directions
new directions by combining biomechanical of research.
movement analysis technology with virtual
reality. This approach has taken us into the
exciting arena of “serious games” which are The CAREN system
games designed for purposes other than
entertainment (see www.seriousgames.org.uk). A lot of delicate details of human balance and
Packaging a research paradigm as a computer posture have been uncovered following the
game builds heavily on the power of motiva- conventional scientific approach of breaking
tion which together with repetition and feed- down the phenomenon of standing balance
back provides the three components of motor into its conceptual building blocks (inputs,
learning1,2. Our initial studies were of techni- processing and outputs) and examining the
cal nature exploring the potential uses of the components in great detail. The complexity of
only CAREN system in the UK but we soon the experiments however is still inferior when
moved on to bona fide research projects in a compared to the challenges of real life situa-
clinical context. Considerable research fund- tions where balance seamlessly or consciously
ing by The WellChild Trust, Alder Hey Chil- plays an integral part of coping with our envi-
dren’s NHS Foundation Trust, and LJMU’s ronment. How a figure skater can land after a

7
STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

jump on a narrow blade, how a child learns to SoftImage) can be used to generate faith-
ride a bicycle, how a mother can manoeuvre a ful scenarios which provide the task for
buggy on a bus braking in a bend, or even the individual experiencing the interactive
how a frail old lady can stand by the kitchen environment. The Computer Assisted reha-
sink with poor vision and painful hips, are all bilitation Environment (CAREN system,
challenges to the experimenter which so far www.motekmedical.com) links together all the
had to be avoided or largely simplified before above components making it possible to set
they could be approached. up real time cause-effect relationships and
feedback loops of events. For example the
Inputs, outputs and the internal processing Boat demo (Figure 1) uses reflective markers
of the neuro-musculo-skeletal system can be attached to the trunk to reconstruct trunk lean
manipulated and evaluated by the integration sideways which rolls a boat on the screen and
of a number of sub-systems. Vision, proprio- rolls the platform at the same time. Forward
ception and the vestibular system are the three tilt of the segment is linked to forward accel-
inputs to balance and posture and can be eration of the boat. In this example leaning of
influenced by computer generated graphics the trunk moves the supporting platform
presented to a subject standing on a comput- which in turn moves the trunk, thereby clos-
er driven movable Stewart platform. Output ing a positive feedback loop. Responses of
of the subject is segmental movement cap- the person immersed in the virtual environ-
tured in real-time by an optoelectronic ment carefully designed to interact with the
multi-camera setup (e.g. Vicon or Qualisys). individual can be recorded and analysed
Advanced computer graphics software (e.g. offline.

Figure 1. The CAREN system integrates biomechanical movement capture, three dimensional
computer graphics and a movable platform to provide immersive virtual environments

8
 STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

Initial studies indicated by the small bounding rectangles

containing the instantaneous axes of rotation
In one of our earliest studies3 we aimed to while rotating around the ankle, knee and
determine the technical performance of the hip of a subject standing on the platform
CAREN platform by defining its kinematic (ankle: 1.87 × 0.81 mm, knee: 3.04 × 1.23 mm,
response characteristics to commonly used hip: 3.14 × 1.63 mm).
input functions for each of its six translation-
al and rotational degrees of freedom. Limits The ability to aim rotational perturbations
of the platform’s linear and angular displace- of the standing support at specific joints was
ment (0.13 m and 16°), velocity (0.4 ms-1 and used to test9 if rotation of a person around the
40°s-1) and acceleration (4.5 ms-2 and 450°s-2) hip joint leads to a more focused movement
compare favourably with other single degree perturbation when compared to the conven-
of freedom platforms. Full range of the tional rocking movement routinely used in
CAREN platform’s amplitude can be utilised targeted training10. Movement responses of
at 0.3 Hz sine movement and 25% of its one healthy female participant were recorded
range at 1.2 Hz. The low maximum average under the two conditions replicated by pro-
displacement crosstalk (1.5%) and velocity grammed movements of the CAREN plat-
crosstalk (3.3%) suggest that single degree of form (Figure 2). In comparison to the rocker
freedom movements can be generated faith- movement, rotation of the body around the
fully although the real strength of the plat- hip joints resulted in significantly reduced
form is in the generation of realistic multiple range of motion (ROM) in the hip and spine
degree of freedom movements. In our assess- (mean±SD: 1.4°±0.3 and 6.8°±1.2 respec-
ment the CAREN system is an appropriate
device for postural and balance research with
some unique features.

There is mounting evidence in the literature4

that triggered responses to balance perturba-
tions involve not only the ankle but also neck,
thigh and trunk muscles. Current research5
demonstrated that a single inverted pendu-
lum model of human posture is not sufficient
and muscles crossing the ankle, knee and hip
account for the majority of forces driving a
musculoskeletal model. Such evidence raises
the need to generate joint specific rotational
perturbations by rotating the CAREN plat-
form around an axis crossing the targeted
joint of the subject standing on the moving Figure 2. The upper part of a Targeted Training
platform. Our first published results with the equipment is mounted on the CAREN platform
CAREN platform6,7,8 described locations of which was programmed to replicate the default
rocking movement normally provided by a rocker
the platform’s default axes of rotation and the base. Responses of hip and lumbar flexion/
mathematical algorithm needed to translate extension were recorded and compared
the three axes of rotations into arbitrary loca- to a condition in which the body was rotated
tions. Accuracy of the platform's movement is around an axis joining the two hip joint centres

9
STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

Figure 3. Tilt of the body using the same ramp function (black curve) resulted
in different responses of hip and lumbar movement when rotating around the hip centre (left chart)
and simulating the default rocking motion (right chart)

tively when rocking and 1.1°±0.4 and of the hip. The ability to focus the movement
2.5°±0.4 respectively when rotating, p<0.05) perturbation on a selected joint while reducing
and a significantly increased ratio of hip-to- the stimulus acting on adjacent joints allows
spine ROM (mean±SD: 0.2±0.1 when rock- more finely graded levels of difficulty when
ing and 0.5±0.2 when rotating, p<0.05) training patients with cerebral palsy (CP).
(Figure 3). Rotation around the targeted joint
simplifies the balance task by minimising the
antero-posterior accelerations and related Virtulal rehabilitation – proof of concept
proprioceptive, vestibular and visual stimuli,
as indicated by the smaller ROMs. Rotation A collaboration with the Gait Laboratory
around the hip joint re-directs the focus of of Alder Hey Children’s NHS Foundation
movement perturbation from the spine, con- Trust (Liverpool) and The Movement Centre
firmed by the increased relative contribution (Oswestry) resulted in an award winning pres-

a) b) c)
Figure 4. The Magic Carpet game (a) played by a girl with cerebral palsy diplegia.
Typical trajectories of the magic carpet approaching the balloon in the origin, showing the differences
between strategies used by the patient (b) and the healthy control (c)

10
 STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

entation11 exploring the potential of Virtual movement control of the pelvis and trunk
Rehabilitation in cerebral palsy. The study (core). We proposed that exposure to comput-
documented our proof of concept findings er games driven by active body movements of
with a virtual reality game which employed the core can lead to measurable improve-
a multiple-task driven visual and somatosen- ments of core control and gait, and that com-
sory bio-feedback environment that was puter based core control training is more effec-
expected to serve both as an assessment and tive than exposure to conventional physio-
training tool for core (pelvic) stability in CP therapy. In phase 1, games were designed and
children. Movement of the pelvis recorded developed and participants were recruited
in real time by Vicon was used to navigate the (Figure 5). In phase 2 after establishing their
“Magic Carpet” in order to burst a series of bal- baseline measures of core control and gait, the
loons appearing at random positions (Figure CORE group received core control training
4a). A comparison of the movement strategies by playing the games, and the CONTROL
of a diplegic patient and a typically develop- group played the games for the same dura-
ing control (Figure 4b and 4c) indicated that tion and frequency but without a focus on
the method had a great potential to evaluate core control (by using a joystick). This phase
core stability. A video of the game can be seen ended with recording of post-training meas-
at www.ljmu.ac.uk/sportandexercisesciences/ ures of core control and gait scores. We are
RISES/WellChildProject. currently in phase 3 when the results are
analysed leading to output generation.

Goblin post office – the WellChild project Related satellite studies have explored the
effect of game speed12 and the development
Motivated by the initial findings and finan- profiles of performance over time13. Our ini-
cially supported by The WellChild Trust, the tial results were presented at ESMAC 2009
next stage was to further develop and evaluate reporting some findings of the pre-training
custom made computer games to improve tests of core control, focusing on rotation and
movement co-ordination of children with tilt of the trunk and pelvis14. The game task
cerebral palsy who have impaired dynamic was similar to that of the “Magic Carpet”

Figure 5. One of our participants playing the Goblin Post Office game.
Note the two clusters of markers on the trunk and pelvis which drive the game

11
STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

game with the added feature that an adaptive Further results of the study will follow soon
algorithm adjusted the speed of the game in and our next project funded by LJMU’s
response to changes in target hit rates dur- Institute for Health Research is under way
ing game play. The maximum settled speed aiming to extend movement training to
reached under each segmental control scheme peripheral joints in order to improve reduced
was used to quantify performance. Significant selectivity of muscle control in children with
differences were found between the control cerebral palsy. Our newly developed games
schemes in that game speed with trunk rota- training methodology based on inertia sen-
tion was highest (54.5±0.6 ms-1) followed by sors (XSens, Enschede, The Netherlands)
trunk tilt (48.3±2.0 ms-1) and then pelvic will be applied in Liverpool schools accessing
rotation (44.1±1.6 ms-1). Children were more children (Figure 6). We aim to describe
unable to play the game using pelvic tilt. the interaction between core and peripheral
Sagittal plane tilt was more difficult to control movement control measuring the specific
than transverse plane rotation, the trunk was effects of the interventions on selective move-
better controlled than the pelvis, and there ment control. The specific implementation of
was least control over movements of the virtual reality based movement training in
pelvis. clinical and community settings was funded
by Alder Hey Charitable Funds.

Future directions

Our current research stream uses visual and

somatosensory feedback based on segmental
movement. Thanks to the recent develop-
ments by Motek Medical, their Human Body
Model software can visualise muscle actions
in real time derived from inverse dynamics
(joint moments and powers) combined with
optimisation algorithms (Figure 7). Muscle
specific visual bio-feedback realised in flexible
virtual environments opens new directions to
improving the movement control of children
with neuromuscular conditions.

Figure 6. The Goblin Post Office

game running on a laptop computer can be
driven by XSens orientation trackers.
Strategic placement of the sensors to selected body
segments can be used to address the role of core
and peripheral movement control
in motor learning

12
 STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

Figure 7. Motek Medical’s Human Body Model

in action. The colour of each muscle indicates
its activation level calculated in real time
from inverse dynamics using an optoelectronic
movement analysis system and two force
platforms. Addition of an instrumented treadmill
with dual force platforms enables visualisation
of muscle activity during gait

REFERENCES
1. Rizzo AA, Buckwalter G, van der Zaag C. Virtual 7. Vanrenterghem J, Barton GJ, Lake M, Lees A.
environment applications in clinical neurophysi- Changing the axes of rotation in a six degrees of
ology. In: Stanney K, editor. Handbook of vir- freedom moving platform used for postural
tual environment technology. London: Lawrence research [abstract]. Gait and Posture 2005;21
Erlbaum; 2002. p. 1027–64. (Suppl 1):152. ISPGR 17th Conference; May 29–
June 2; Marseille, France.
2. Holden MK. Virtual environments for motor
rehabilitation: review. Cyberpsychology and 8. Barton GJ, Vanrenterghem J, Lees A, Lake M.
Behavior 2005;8:187–211. A method for manipulating a movable plat-
form’s axes of rotation: a novel use of the
3. Lees A, Vanrenterghem J, Barton GJ, Lake M. CAREN system. Gait and Posture 2006;
Kinematic response characteristics of the CAREN 24(4):510–4.
moving platform system for use in posture and
balance research. Medical Engineering and 9. Barton GJ, Butler PB, Hawken MB, Jarvis SE.
Physics 2007;29(5):629–35. Re-direction of movement perturbation by
rotating around the targeted joint [abstract].
4. Allum JH, Bloem BR, Carpenter MG, Hulliger M, Journal of Sports Sciences 2009;27(Suppl 1):
Hadders-Algra M. Proprioceptive control of pos- 3–4. The Ergonomics Society: 6th International
ture: a review of new concepts. Gait and Posture Conference on Sport, Leisure and Ergonomics.
1998;8:214–42. 2007 Nov 14.

5. Jansen K, Levin O, De Groote F, Duysens J, 10. Butler PB. A preliminary report on the effec-
Jonkers I. Modelling postural control: Validity of tiveness of trunk targeting in achieving inde-
the inverted pendulum assumption. Gait and pendent sitting balance in children with cere-
Posture 2009 (In Press). 18th ESMAC Meeting bral palsy. Clinical Rehabilitation 1998;12(4):
and Conference; Sept 17–19; London, United 281–93.
Kingdom.
11. Barton GJ, Holmes G, Hawken M, Lees A,
6. Barton GJ, Vanrenterghem J, Lees A. Joint specif- Vanrenterghem J. A virtual reality tool for train-
ic proprioceptive perturbation of standing bal- ing and testing core stability: a pilot study
ance on a movable platform [abstract]. Gait and [abstract]. Gait and Posture 2006;24(Suppl 2):
Posture 2005;22(Suppl 1):50. 14th ESMAC Meet- 101–2. 1st Joint ESMAC–GCMAS Meeting;
ing and Conference, Sept 22–24; Barcelona, 2006 Sept 28–30; Amsterdam, Netherlands:
Spain. Winner of the Best Poster Prize.

13
STATE OF THE ART
Biomechanica Hungarica II. évfolyam, 2. szám

12. Hawkins PJR, Hawken MB, Barton GJ. Effect of 28:10–1. 17th ESMAC Meeting and Confer-
game speed and surface perturbations on pos- ence; Sept 11–13; Antalya, Turkey.
tural control in a virtual environment. Pro-
ceedings of the 7th ICDVRAT with ArtAbilita- 14. Barton GJ, Hawken MB, Butler P, Holmes G,
tion; 2008; Maia, Portugal. Foster RJ. Movement control of the trunk and
pelvis in cerebral palsy diplegia [abstract]. Gait
13. Foster RJ, Hawken MB, Barton GJ. Movement and Posture 2009 (In Press). 18th ESMAC
Co-ordination of the pelvis in a virtual game Meeting and Conference; Sept 17–19; London,
environment [abstract]. Gait and Posture 2008; United Kingdom.

Our research is funded by The WellChild Trust, Alder Hey Children’s NHS Foundation Trust,
and the Institute for Health Research of Liverpool John Moores University.

Dr. Gábor Barton

Research Institute for Sport and Exercise Sciences
Liverpool John Moores University
15–21 Webster Street, Liverpool, L3 2ET, United Kingdom
Tel: +44 (0)151 2314333

14