Virtual Reality System for Rehabilitation of Children

with Cerebral Palsy: A Preliminary Study

Pierre Michel

, Paul Richard

, Takehiko Yamaguchi

, Adrien Verhulst

Eulalie Verhulst

and Micka

el Dinomais

Laboratoire Angevin de Recherche en Ing

enierie des Syst

emes (LARIS), Universit

e d’Angers, Angers, France

Faculty of Industrial Science & Technology, Tokyo University of Science, Tokyo, Japan

CRENAU UMR CNRS 1563, Universit

e de Nantes, Nantes, France

Keywords:

Virtual Reality, Adaptive Game, Rehabilitation, Children, Cerebral Palsy.

Abstract:

We present a non-immersive virtual reality (VR) system for the rehabilitation of children with Cerebral Palsy

(CP). Our objective is to encourage and motivate children to improve their limb motor control while playing

a game. Two tasks are available: (1) intercepting or (2) catching / releasing moving objects using a Kinect

T M

sensor. These tasks are achieved via the control of a virtual character placed in a virtual island. A control-

display ratio is used to virtually increase the child workspace allowing him/her to intercept moving objects

from any direction. In addition, a dynamic difﬁculty adjustment (DDA) is used to keep a good motivation

level. Furthermore, virtual coach is provided to support and congratulate the children. Twenty healthy children

participated in a preliminary experiment. The aim was (1) to collect control data concerning performance and

workload, and (2) to investigate the effect of the virtual coach. Results show a good usability of the game and

reveal a high ratio of acceptance and enjoyment from the children.

1 INTRODUCTION

Cerebral palsy (CP) is a term that refers to various

motor impairments caused by damage to the central

nervous system during foetal development. Tradi-

tional CP therapies are often of little interest to a

child, affecting his motivation to continue therapeu-

tic activities (Schmidt and Lee, 2005). For example,

Constraint-induced movement therapy (CIMT), often

used to improve upper limb function encourages the

use of the affected hand by restricting the unaffected

hand and asking for intensive movement with the im-

paired upper limb (Hoare et al., 2007). Having the

good arm blocked for long periods of time can gener-

ate frustration and might not be applicable in a long-

term rehabilitation program.

More child-friendly approaches are needed dur-

ing the neuro-development of children with CP. For

example, rehabilitation programs should offer thera-

peutic games that are speciﬁcally designed to encour-

age the child to use their affected limb and even both

arms in coordination. The proposed games should be

easy to use and non-intrusive in order to affect chil-

dren movements.

The paper is organized as follows: the next sec-

tion provides a survey of the related work concerning

interactive motor rehabilitation systems and adaptive

games. Section 3 presents an overview of our sys-

tem and its innovative characteristics. In Section 4,

we describe our approach for the auto-adaptation of

the system. Section 5 is dedicated to the experimen-

tal study. Section 6 concludes the paper and discuss

directions for future work.

2 RELATED WORKS

2.1 Rehabilitation Systems

Virtual reality (VR) or Virtual Environments (VEs)

may be described as multi-sensory, interactive and

immersive computer-based 3D environments that

could be used to simulate some aspects of the real

world. VR is now recognized as a powerful tool for

the assessment and rehabilitation of both motor and

cognitive impairments and provides a unique medium

for the achievement of several requirements of ef-

fective rehabilitation: controlled conditions, repeti-

tive practice and feedback about performance (Riva,

306

Michel, P., Richard, P., Yamaguchi, T., Verhulst, A., Verhulst, E. and Dinomais, M.

Virtual Reality System for Rehabilitation of Children with Cerebral Palsy: A Preliminary Study.

DOI: 10.5220/0005720503040311

In Proceedings of the 11th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2016) - Volume 1: GRAPP, pages 306-313

ISBN: 978-989-758-175-5

2003; Burdea, 2003; Gaggioli et al., 2009; Halton,

2008; Raspelli et al., 2012; Cipresso et al., 2012;

Pallavicini et al., 2013). In addition, VR offers

boundless variations of augmented feedback, objects,

and allow the users immersion in attractive environ-

ments (Cikajlo et al., 2010; Berger-Vachon, 2006) so

that VR remains motivating and entertaining (Rand

et al., 2009).

Few virtual rehabilitation systems have been de-

veloped for motor rehabilitation using full-body in-

teraction. For example, Kizony et al., (Kizony et al.,

2003) proposed IREX, based on a video-capture sys-

tem. A single camera is used for vision-based track-

ing to capture the users movements. The captured

video images can be displayed on a connected TV

screen, corresponding in real time to his movements.

Its suitability has been investigated for use during mo-

tor or cognitive rehabilitation (Rand et al., 2004). The

Cybex Trazer

T M

employs a single infra-red beacon

which is mounted on a belt worn around the waist of

the user. The user’s motion is captured by monitor-

ing the sensor’s position by the tracker bar (Fitzgerald

et al., 2007). The EyeToy

T M

game uses a single cam-

era to capture the users movements. Interaction with

an on-screen user avatar can only track movements in

a single plane and is not able to record body move-

ments (Fitzgerald et al., 2007). For a review, a review

see Adamovich et al., (Adamovich et al., 2009).

Some research used others interaction techniques

to rise the motivation of children with CP in their mo-

tor rehabilitation exercises. For example using VR

devices with Wii

T M

or with Kinect

T M

, the children

showed improvement in the motor rehabilitation with

VR devices (Sharan et al., 2012; Deutsch et al., 2008;

Ortiz-Gutirrez et al., 2013; Chang et al., 2013). In-

deed Chang, Han and Tsai (2013) asked two teenagers

to move their arms in front of a screen. When they

made correct movement a song was broadcast and a

cartoon character was displayed. In our study, to sup-

port children’s motivation, we propose to use a virtual

coach and found the rehabilitation process on a play-

ful approach.

2.2 Adaptive Games

Dynamic Difﬁculty Adaptation (DDA), also known

as Dynamic Game Balancing (DGB), is the process

of automatically changing parameters, scenarios, and

behaviours in a video game in real-time, based on

the player’s ability, in order to avoid them becoming

bored (if the game is too easy) or frustrated (if it is too

hard) (Huang et al., 2010). In recent years, studies

have used different approaches to handle DDA. For

example, Parnandi et al., (Parnandi et al., 2013) pro-

posed an approach based on control theory’s princi-

ples. They used variation of actual and desired arousal

of the user as the variation error to minimize. They

conducted an experiment on 20 subjects through a

Car-racing application. Hocine and Goua

ıch (Hocine

and Goua

ıch, 2011), described an approach based on

prior assessments of the capability of the user. The

adaptation is done through an ability zone, which con-

tains information on the difﬁculty to do a given task

at given position. They conducted an experiment on 8

subjects through a reaching-task application.

Goua

ıch et al., (Goua

ıch et al., 2012) proposed

a digital pheromone approach based on the ant al-

gorithm introduced by (Dorigo and St

utzle, 2004).

The adaptation is done through an ability zone up-

dated regarding users performance. They conducted

an experiment on 10 subjects through a reaching-task

application. Arulraj et al., (Arulraj, 2010) proposed

a differential learning approach for NPC. The agent

learning-rate reduces with time, while being impacted

by users performance. The approach feasibility has

been tested using the Minigate game. Andrade (An-

drade and Ramalho, 2005) and Tan (Tan et al., 2011)

both proposed a Reinforcement Learning (RL) ap-

proach for NPC. Andrade implemented it by using

the Q-learning algorithm, the adaptation being done

by choosing the action-value which ﬁt the level of

the user. The approach feasibility has been tested

using a ﬁghting application. Tan implemented it by

using Adaptive Uni-Chromosome Controller (AUC)

and Adaptive Duo-Chromosome Controller (ADC)

algorithms, the adaptation is being done by activating

controller’s behaviour which ﬁt the level of the user.

Figure 1: Screen shot of the rehabilitation game.

3 SYSTEM DESCRIPTION

The goal of our research is to develop an innovative

interactive virtual rehabilitation system which enables

children with CP to play while improving their motor

Virtual Reality System for Rehabilitation of Children with Cerebral Palsy: A Preliminary Study

307

(a)

(b)

(c)

Figure 2: Interaction techniques : (a) using the game-pad

only, (b) using the game-pad and the TrackIR

T M

, (c) and

(d) using the kinect

T M

and the TrackIR

T M

control of the limb. This requires the use of a non-

intrusive motion capture device such as the Microsoft

Kinect

T M

and the possibility to increase the children

workspace allowing him/her to reach and intercept

any moving objects. The proposed approach enables

a modiﬁable control-display ratio together with dy-

namic difﬁculty adjustment (DDA) capabilities used

to keep children motivation.

The game takes place in a virtual island. Figure 1

shows a screenshot of the rehabilitation game. The

avatar is controlled (arms only) in real-time by the

child facing the visual display. The goal is to animate

the avatar in order to intercept, or catch and release

the virtual objects. A Kinect

T M

sensor is used to get

the 3D motion data from the child’s arms.

The user may intercept or catch and release the

virtual objects

 using virtual hands

. While the

virtual character’ arms are moving is 3D space, both

the virtual hands and the approaching objects are con-

strained in a vertical plane. Thus, the subject does

not have to manage the depth. In addition, as previ-

ously mentioned, to allow the children with motor dis-

ability to perform the task, a control-display ratio has

been implemented (ampliﬁcation of movements). An

important aspect is that a control/display ratio (less

than 1/1) could be used to encourage the children with

CP not to use his/her ”good” hand, but rather his/her

paretic hand.

In order to keep the children motivation to play the

game and keep exercising, we constantly display the

score

 on the upper right corner of the screen. At

the bottom of the screen, a scroll bar is displayed to

inform the children about the current difﬁculty level

of the game.

The system proposes different menus to set-up the

game parameters. The ﬁrst menu, illustrated in Fig-

ure 2 (a), allows to set the value of the control/display

ratio for both the right and the left hands

. We could

also set-up the afﬁne transform parameters

. Then,

we could select different avatars (boy, girl, animated

or not). We can display a circle around the avatar to

illustrate the hands movement constraint and also set

and display the collision area (on both hands) used for

intercepting the objects

. Finally, we can save the

parameters and load any set-up

.

The second menu, illustrated in Figure 2 (b), is

mainly used to select the task

 : intercept or catch

and release, and select with which hand

 the child

have to use to perform the task (right hand, left hand,

any of them, or with both hands joined). We can also

set the task’s parameters

 such the size, velocity, or

frequency of occurrence (spawn interval) of the ob-

jects (Figure 5).

The third menu, illustrated in Figure 2 (c), is used

GRAPP 2016 - International Conference on Computer Graphics Theory and Applications

308

to select the set of objects that will have to be caught

. Is this menu, we can also choose a level of dis-

traction

, ranging from a static white background

to a fully dynamic virtual island including moving ob-

jects and sounds. This menu also allows to set the way

(manually or automatically) the difﬁculty level

 will

be changed and the way the task will be displaced in

the island after each session

. In addition the system

allows to set the area from which the moving objects

will approach the avatar. For example, in the Figure 3,

the main direction (red line) can be tuned to a given

angle. In addition, a range (width) around this main

direction can be tuned manually on the circle around

the avatar.

Figure 3: Setting the area for the approaching objects.

3.1 Control/Display Ratio

The Control/Display (C/D) ratio is the ratio between

the amplitude of movement of the user’s real arm

and the amplitude of movements of the virtual cur-

sor (Dominjon et al., 2005). The C/D ratio may be

used to increase the user’s physical workspace allow-

ing him/her to move in a larger workspace. Since pa-

tients with impaired motor functions have a limited

(hopefully increasing) ranges of movement, a C/D ra-

tio may help them to perform task as the people with-

out any impairment.

3.2 Dynamic Difﬁculty Adaptation

In order to maintain the child’s attention and moti-

vation to an acceptable level, we implemented a Dy-

namic Difﬁculty Adaptation (DDA) protocol. We pro-

posed that the game difﬁculty is dependant of (1) the

task parameters, (2) the game ambiance, and (3) the

interaction technique used to perform the task. In the

context, we propose to use a virtual coach (Figure 6)

to support, congratulate or advise the child after each

game session.

Figure 4: Task-Difﬁculty-Automaton.

Figure 4 shows the automaton related to the evo-

lution of the task’s difﬁculty. In this automaton, TP

is the initial state vector of the parameters. This pa-

rameters are tuned according to the user’s personal

proﬁle. If the user performance (score), is less that

30% or more that 80% of catches, then the TP

state

switches to the TP

−1

(difﬁculty is decreased) or TP

(difﬁculty increase) states respectivelly. The task dif-

ﬁculty could then be further decreased (TP

−2

) or in-

creased TP

or returns to the initial state (TP

) ac-

cording to the same given thresholds. This approach

will keep the child motivated, because if the game is

too easy or too difﬁcult he/she will not keep playing.

3.2.1 Interaction Technique

The interaction technique parameters could also im-

pact the task difﬁculty. These parameters are: (1) the

control/display ratio (ratio between the child move-

ments and avatar movements), (2) the feedbacks

(sounds from collision with the moving objects or

haptic feedback), (3) movement constraints (in our

case 2D projection of 3D movements), and (4) time

delay between the child movements and the avatar

movements.

3.2.2 Virtual Coach

The behaviour of the virtual coach is the following.

At the beginning of the game, the coach tell the story,

explain the goal of the game and the task. Then he dis-

appears from the screen. At the end of each session,

the virtual coach appears to (1) congratulate the child

in case of a good score (more than 80% of catch), (2)

advise the child in case of a bad score (less than 30%

Virtual Reality System for Rehabilitation of Children with Cerebral Palsy: A Preliminary Study

309

Figure 5: Diagram for task parameters.

Figure 6: Illustration if the virtual coach.

of catch), or (3) to support the child if the score is

between these two thresholds.

4 USER STUDY

4.1 Aim of the Study

The aim of this study is twofold. Firstly, we want

to collect control data with healthy children in order

to have a baseline for comparison with CP children.

These data are related to performance (score) but also

to the usability of the system, the enjoyment, the task

and the game-play. Secondly, we would like to inves-

tigate the effect of the virtual coach on both perfor-

mance data and subjective data.

4.2 Method

4.2.1 Design and Procedure

Twenty children from a primary school in Angers

(France) participated in the study. The task consisted

Figure 7: Experimental set-up.

in catching a total of 200 objects (fruits: pineapple,

banana, kiwi and apple) during four sessions (50 ob-

jects in each session). The children were split in two

groups of 10 children each. The ﬁrst group performed

the task in condition C

(without the virtual coach),

while the second group performed the task in condi-

tion C

(with the virtual coach). Children were al-

lowed to get a rest of 5 minutes between sessions.

The fruits were launched randomly from starting

point above the avatar’s shoulders (direction was set

to 0 degree and width was set to 90 degrees). The

control/ratio display was set to 1/1. The task difﬁculty

was kept constant throughout the experiment. Thus,

the objects always moved at a constant velocity of 0.2

m.s

−1

. The children have to use their dominant hand.

The task took place in a different position for each

session. Concerning the sounds, a tropical forest am-

biance was displayed including birds, monkey, etc.

The assertions submitted to the children in the

main questionnaire were the following:

1. I had a lot of fun playing the game,

2. I found the game interesting,

3. I found the game boring,

4. The game caught my attention,

5. I found this game somehow complicated,

6. I liked the graphics of the game,

7. I liked the sounds of the game,

8. I liked the game scenario,

9. I liked the character that caught the fruits.

4.2.2 Apparatus

The experimental set-up is illustrated in Figure 7.

The system is composed of a 60 LCD TV monitor,

a laptop, and a Microsoft XBox 360 Kinect

T M

sen-

sor. Each child was placed in front of the screen, at

GRAPP 2016 - International Conference on Computer Graphics Theory and Applications

310

the center of the Kinect’s workspace. The Kinect

T M

SDK v1.8 for Windows was used for motion tracking.

In order to measure children performance, we

recorded the total number of caught objects in each

sessions. To get subjective data about the system us-

ability, preference, enjoyment, task, and game-play, a

non standardized questionnaire was used (seven Lik-

ert scale). The children were also asked about the time

they spend in playing video games. We also used the

NASA Task Load Index (TLX) to assess the task’s

mental, physical and temporal demand, user’s per-

ceived performance, effort and frustration. Finally,

we observed the children while performing the task

and noted his/her comments, strategies and speciﬁc

behaviours.

4.3 Results and Discussion

Results about performance are illustrated in Figure 8.

We observed that the scores (number of fruits caught)

obtained in both conditions are quite good. Indeed,

we obtained a mean score of 39.18 (std : 1.1) for the

condition and a mean score of 42.45 (std : 0.50)

for the C

condition.

A statistical analysis (MannWhitney U test) re-

vealed that the difference between the C

condition

(no avatar) and the C

condition (with avatar) is not

signiﬁcant (U= 25.5, p-value= 0.068). However, we

could consider this results as a trend and state that the

avatar somehow affect the children performance.

Results revealed that the children enjoyed playing

the game (6 points). We observe that the results of

the main questionnaire are not signiﬁcantly different

between the 2 groups and are very similar for both the

condition and the C

condition.

The children found the game interesting (5.7

points for both conditions) and stated the the game

(6.0 points for the C

condition and 5.75 for the C

condition). They stated that they liked the graphics

(5.7 points for the C

condition and 6.0 for the C

condition ) and the sounds (5.9 points for the C

con-

dition and 6.05 for the C

condition ) of the game.

The children liked the scenario (5.9 points for the C

condition and 5.85 for the C

condition ), the child

character (6.2 points for the C

condition and 6.1 for

the C

condition).

The children who had the virtual coach had two

more assertions:

• I liked the virtual coach,

• I found the coach useful.

The ﬁrst question obtained in average 5.45 (std :

0.69) points overs the 7 points of the likert scale. The

second question obtained in average 6 (std : 0.82).

Figure 8: Number of fruits caught vs. condition C

and

condition C

(dash line).

Thus, the presence of the virtual coach was very ap-

preciated by the children. More interesting, they

found it useful for the task. Observation during the

experiment conﬁrm that the children liked the avatar.

Results from the Nasa TLX questionnaire are il-

lustrated in Figure 9 (a) and (b). We observe in Fig-

ure 9 that the results are about the same for both con-

dition C

and C

excepted for the question 3 con-

cerning the temporal demand where statical differ-

ence was found (U= 80, p-value= 0.04).

This result reveals that the children who per-

formed in C

condition felt less intensity and pace in

the game. Thus, the intervention of the virtual coach

at the end of each session allowed the children to get

some rest. Results from the NASA TLX question-

naire also reveal that the game does not involve a high

mental demand. Indeed, the rating for mental demand

was 44.5 (std : 15.71) for C

condition and 54.0 (std

: 14.87) for C

. However, and this is not very surpris-

ing, the physical demand was rated higher (about 60

for both conditions). Concerning the children perfor-

mance self estimation, we observed that it was rather

high for both conditions: 66.5 (std : 10.29) for C

condition and 73.5 (std : 10.01) for C

. Similarly,

the children felt developing rather a same level of ef-

fort: 55.0 (std : 14.72) for C

condition and 50.0 (std

: 15.63) for C

. Finaly, the children felt a rather low

frustration at the end of the experiment: 14.50 (std :

6.43) for C

condition and 12.0 (std : 3.50) for C

condition.

5 CONCLUSION AND FUTURE

WORK

In this paper, we presented VR system for the reha-

bilitation of children with Cerebral Palsy (CP). The

Virtual Reality System for Rehabilitation of Children with Cerebral Palsy: A Preliminary Study

311

(a)

(b)

Figure 9: Results for the Nasa TLX questionnaire (dash

lines correspond to the C

condition ) : (a) rating of the

6 criteria of the Nasa TLX questionnaire, and (b) between

subjects comparison.

system is non-intrusive and includes interesting char-

acteristics such as Dynamic Difﬁculty Adjustment

(DDA) and a control/display ratio. A Kinect

T M

sen-

sor allows the user to control a 3D avatar and inter-

cept or catch and release moving objects. Further-

more, virtual coach is provided to support the child.

Twenty healthy children participated in a preliminary

experiment. The aim was to collect control data con-

cerning performance and workload, and to investigate

the effect of the virtual coach on both performance

and subjective data. Results show a good usability of

the game and reveal a high ratio of acceptance and

enjoyment. In the near future, experiments with both

healthy children and children with cerebral palsy (CP)

will be carried out to investigate the effect of the con-

trol/display ratio on performance. The DDA proto-

col will be extended to integrate both behavioural and

physiological data.

ACKNOWLEDGEMENTS

This work was supported by the Institut des Sciences

et Techniques de l’Ingnieur d’Angers (ISTIA) and by

the ENJEU[X] project, funded by the Region Pays de

la Loire, France. We wish to thank all children who

participated in the study.

REFERENCES

Adamovich, S. V., Fluet, G., Tunik, E., and Merians, A.

(2009). Sensorimotor training in virtual reality: A re-

view. NeuroRehabilitation, 25(1):29:1196–1207.

Andrade, G. and Ramalho, G. (2005). Challenge-sensitive

action selection: an application to game balancing. In-

ternational Conference on Intelligent Agent Technol-

ogy, pages 194–200.

Arulraj, J. (2010). Adaptive agent generation using machine

learning for dynamic difﬁculty adjustment. Computer

and Communication Technology (ICCCT), 2010 Inter-

national Conference on, pages 746–751.

Berger-Vachon, C. (2006). Virtual reality and disability.

Technology and Disability, 18:163–165.

Burdea, G. (2003). Virtual rehabilitation : beneﬁts and chal-

lenges. Methods Inf Med., 42(5):519–523.

Chang, Y.-J., Han, W.-Y., and Tsai, Y.-C. (2013). A kinect-

based upper limb rehabilitation system to assist people

with cerebral palsy. Research in Developmental Dis-

abilities, 34(11):3654–3659.

Cikajlo, I., Rudolf, M., Goljar, N., and Matja

c, Z. (2010).

Continuation of balance training for stroke subjects

in home environment using virtual reality. In 8

Intl Conf. on Disability, Virtual Reality & Associated

Technologies (ICDVRAT), pages 235–240, Valparaso,

Chile.

Cipresso, P., Gaggioli, A., Serino, S., Pallavicini, F.,

Raspelli, S., Grassi, A., and Riva, G. (2012). Eeg

alpha asymmetry in virtual environments for the as-

sessment of stress-related disorders. Studies In Health

Technology And Informatics, 173:102–104.

Deutsch, J. E., Borbely, M., Filler, J., Huhn, K., and

Guarrera-Bowlby, P. (2008). Use of a low-cost, com-

mercially available gaming console (wii) for rehabil-

itation of an adolescent with cerebral palsy. Physical

therapy, 88(10):1196–1207.

Dominjon, L., L

ecuyer, A., Burkhardt, J., Andrade-

Barroso, G., and Richir, S. (2005). The ”bubble”

technique: Interacting with large virtual environments

using haptic devices with limited workspace. In

GRAPP 2016 - International Conference on Computer Graphics Theory and Applications

312

First Joint Eurohaptics Conference and Symposium on

Haptic Interfaces for Virtual Environment and Teleop-

erator Systems, WHC 2005, Pisa, Italy, March 18-20,

2005, pages 639–640.

Dorigo, M. and St

utzle, T. (2004). Ant Colony Optimiza-

tion. Computer Science and Intelligent Systems MIT

Press, Cambridge MA, USA.

Fitzgerald, D., Foody, J., Kelly, D., Ward, T., Markham,

C., McDonald, J., and B, B. C. (2007). Development

of a wearable motion capture suit and virtual reality

biofeedback system for the instruction and analysis of

sports rehabilitation exercises. In International Con-

ference of the III EMBS, pages 4870–4874.

Gaggioli, A., Keshner, E., Weiss, P., and Riva, G. (2009).

Advanced technologies in rehabilitation - empowering

cognitive, physical, social and communicative skills

through virtual reality, robots, wearable systems and

brain-computer interfaces. .

Goua

ıch, A., Hocine, N., Dokkum, L. V., and Mottet, D.

(2012). Digital-pheromone based difﬁculty adaptation

in post-stroke therapeutic games. Proceedings of the

2nd ACM SIGHIT symposium on International health

informatics - IHI ’12.

Halton, J. (2008). Virtual rehabilitation with video games:

a new frontier for occupational therapy. Occupational

Therapy Now., 9(6):12–14.

Hoare, B., Imms, C., Carey, L., and Wasiak, J. (2007).

Constraint-induced movement therapy in the treat-

ment of the upper limb in children with hemiplegic

cerebral palsy: a cochrane systematic review. Clinical

Rehabilitation, 27:675 685.

Hocine, N. and Goua

ıch, A. (2011). Motivation based difﬁ-

culty adaptation for therapeutic games. 16th Interna-

tional Conference on Computer Games (CGAMES),

16th Inter:257–261.

Huang, W., He, S., Chang, D., and Hao, Y. (2010). Dy-

namic difﬁculty adjustment realization based on adap-

tive neuro-controlled game opponent. Computational

Intelligence (IWACI), pages 66–71.

Kizony, R., Katz, N., and Weiss, P. (2003). Adapting

an immersive virtual reality system for rehabilitation.

J.Visual. Comp. Anim., 14:261–268.

Ortiz-Gutirrez, L.-O., de la Cuerda Pidrola, C., Snchez-

Camarero, A.-D., and Culebras, M. (2013). Kinect

xbox 360 as a therapeutic modality for children with

cerebral palsy in a school environment: a preliminary

study. NeuroRehabilitation, 33(4):513–521.

Pallavicini, F., Cipresso, P., Raspelli, S., Grassi, A., Serino,

S., Vigna, C., Triberti, S., Villamira, M., Gaggioli, A.,

and Riva, G. (2013). Is virtual reality always an effec-

tive stressors for exposure treatments? some insights

from a controlled trial. BMC Psychiatry, 13(52):265–

272.

Parnandi, A., Son, Y., and Gutierrez-Osuna, R. (2013). A

Control-Theoretic Approach to Adaptive Physiologi-

cal Games. 2013 Humaine Association Conference on

Affective Computing and Intelligent Interaction, pages

7–12.

Rand, D., Kizony, R., and Weiss, P. (2004). Virtual reality

rehabilitation for all: Vivid gx versus sony playstation

ii eyetoy. In 5

Intl Conf. on Disability, Virtual Reality

& Assoc. Tech., pages 87–94, Oxford, UK,.

Rand, D., Weiss, P. T., and Katz, N. (2009). Training multi-

tasking in a virtual supermarket: A novel intervention

after stroke. American Journal of Occupational Ther-

apy, 63:535–542.

Raspelli, S., Pallavicini, F., Carelli, L., Morganti, F., Pe-

droli, E., Cipresso, P., andB. Corra B, B. P., Sangalli,

D., Silani, V., and Riva, G. (2012). Validating the

neuro vr-based virtual version of the multiple errands

test: preliminary results. Presence-Teleoperators And

Virtual Environments, 21:31–42.

Riva, G. (2003). Applications of virtual environments in

medicine. Methods Inf Med., 45(5):524–534.

Schmidt, R. and Lee, T. (2005). Motor learning: A behav-

ioral emphasis. In 4

edn in Human Kinetics, Cham-

paign, IL, USA.

Sharan, D., P., S., A., Rameshkumar, R., Mathankumar, M.,

Paulina, R., and Manjula, M. (2012). Virtual reality

based therapy for post operative rehabilitation of chil-

dren with cerebral palsy. Work, 41(1):3612–5.

Tan, C. H., Tan, K. C., and Tay, A. (2011). Dynamic game

difﬁculty scaling using adaptive behavior-based AI.

IEEE Transactions on AI in Games, 3(4):289–301.

Virtual Reality System for Rehabilitation of Children with Cerebral Palsy: A Preliminary Study

313