c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265

j o ur na l ho me pag e: www.int l .e lsev ierhea l th .com/ journa ls /cmpb

Development of virtual reality proprioceptiverehabilitation system for stroke patients

Sangwoo Choa, Jeonghun Kub, Yun Kyung Choc, In Young Kima,Youn Joo Kangc,∗∗, Dong Pyo Janga, Sun I. Kima,d,∗

a Department of Biomedical Engineering, Hanyang University, Seoul, South Koreab Department of Biomedical Engineering, Keimyung University, Daegu, South Koreac Department of Rehabilitation Medicine, Eulji Hospital, Eulji University School of Medicine, Seoul, South Koread Osong Medical Innovation Foundation, Medical Device Development Center, Chungbuk, South Korea

a r t i c l e i n f o

Article history:

Received 14 May 2013

Received in revised form

8 September 2013

Accepted 10 September 2013

Keywords:

a b s t r a c t

In this study, the virtual reality (VR) proprioception rehabilitation system was developed

for stroke patients to use proprioception feedback in upper limb rehabilitation by block-

ing visual feedback. To evaluate its therapeutic effect, 10 stroke patients (onset > 3 month)

trained proprioception feedback rehabilitation for one week and visual feedback rehabil-

itation for another week in random order. Proprioception functions were checked before,

a week after, and at the end of training. The results show the click count, error distance

and total error distance among proprioception evaluation factors were significantly reduced

Proprioception feedback

Computer based rehabilitation

Virtual Reality

Stroke

upper-extremity limb

after proprioception feedback training compared to visual feedback training (respectively,

p = 0.005, p = 0.001, and p = 0.007). In addition, subjects were significantly improved in con-

ventional behavioral tests after training. In conclusion, we showed the effectiveness and

possible use of the VR to recover the proprioception of stroke patients.

© 2013 Elsevier Ireland Ltd. All rights reserved.

external stimuli with robotic or functional electrical stim-

1. Introduction

Rehabilitation training is essential for most stroke patientswho have symptoms such as declined or abnormal motorcontrol due to brain damage [1]. Generally, the elementsneeded for normal physical activities are motor control mod-ulating movement in real time as well as strength. Motorcontrol amends the motion by interaction between visualfeedback that recognizes the external space or movement

of oneself through vision and proprioception feedback thatrefers information about movement and position of body,which transverse from muscle spindles into central nervous

∗ Corresponding author at: 17, Hangdang-Dong, Sungdong-Gu, Seoul, S∗∗ Corresponding author at: 14 Hangul-Bisuk Gil, Nowon-Gu, Seoul, Sout

E-mail addresses: md52516@hanmail.net (Y.J. Kang), sunkim7114@g0169-2607/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reshttp://dx.doi.org/10.1016/j.cmpb.2013.09.006

system [2–4]. Stroke patients have difficulty in conductingexact motor control due to declined strength as well asability to utilize feedback [5]. In particular, stroke patientsshowed lower accuracy of motor control compared to healthyindividuals in situations without visual feedback of move-ment rather than with visual feedback [6,7]. In spite ofthese previous studies conventional rehabilitation therapyhave mainly focused on strength exercise with occupa-tional therapists support and motor control training using

outh Korea. Tel.: +82 2 2293 7280.h Korea. Tel.: +82 2 970 8315.mail.com (S.I. Kim).

ulation [8,9]. Moreover, it was reported that the trainingeffect of stroke patients could be reduced by reliance ofvisual feedback of their movement during training because

erved.

i n b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265 259

t[

smpttmiDrs[

ttusaIact

idppV

2

2

It5aToimapwmpvn[[suostBi

Dem

ogra

ph

ics

and

clin

ical

resu

lts

of

stro

ke

pat

ien

ts.

ex

Age

Are

a

ofLe

sion

Infa

rcti

on/

hem

orrh

age

Tim

e

Aft

erSt

roke

(mon

th)

MM

SEsc

ore

FMS

scor

ePr

opri

ocep

tion

fun

ctio

n

test

erro

r

angl

e(d

egre

e)

Box

and

Blo

ck

test

Jebs

en-T

aylo

r

han

dfu

nct

ion

test

-wri

tin

gsu

btes

t

(sec

) in

pre

-tes

t

Aff

ecte

dh

and

Un

affe

cted

han

dA

ffec

ted

han

dU

naf

fect

edh

and

50

Lt. B

G

(su

bcor

tica

l)

Hem

orrh

age

6.3

29

49

5.41

30

53

45.4

7

44.6

6

62

Rt.

CR

(su

bcor

tica

l)

Infa

rcti

on

74.1

25

55

7.32

27

40

54.2

8

11.5

0

58

Lt. p

ons

Infa

rcti

on

5.2

28

47

8.91

42

50

85.1

6

94.1

8

63

Lt. p

ons

Infa

rcti

on

9.0

27

40

8.01

17

43

142.

10

76.5

0

40

Lt. C

R

(su

bcor

tica

l)

Infa

rcti

on

3.3

29

62

4.82

33

45

52.3

8

17.3

4

58

Lt. M

CA

(su

bcor

tica

l)

Infa

rcti

on

24.0

25

19

7.32

4

46

–

41.9

3

46

Lt. B

G

(su

bcor

tica

l)

Hem

orrh

age

65.1

25

18

17.2

3

6

45

–

50.1

9

64

Lt. C

R

(su

bcor

tica

l)

Infa

rcti

on

53.2

26

55

8.77

27

38

40.1

2

42.1

0

53

Lt. C

R

(su

bcor

tica

l)

Infa

rcti

on

146.

3

30

51

5.79

26

51

72

28

52

Lt. B

G

(su

bcor

tica

l)

Hem

orrh

age

8.1

29

66

11.6

2

30

51

29.8

5

22.9

7

Lt, l

eft;

MC

A, m

idd

le

cere

bral

arte

ry; B

G, B

asal

gan

glia

; CR

, cor

ona

rad

iate

; FM

S,

affe

cted

up

per

lim

b

scor

e

of

Fugl

-Mey

er

Ass

essm

ent

(ou

t

of

66);

MM

SE, M

ini-

Men

tal S

tate

Exam

inat

ion

(ou

t

c o m p u t e r m e t h o d s a n d p r o g r a m s

he vision of patient were intact rather than proprioception10,11].

Proprioception is evaluated by tests which measure aubject’s ability to detect an externally imposed passive move-ent, or the ability to reposition a joint to a predetermined

osition [5]. In order to improve proprioception, sensorimotorraining programs have been suggested to facilitate joint posi-ion sense and dynamic joint stability using rhythmic active

otion, angle repositioning and standing on an air cush-on with support to stimulate muscular co-activation [12,13].espite recently shedding a light on the proprioception in

ehabilitation, there are few studies related to rehabilitationystems focusing on the improvement of proprioception itself14,15].

The virtual reality (VR) technique can provide various vir-ual environments and has been used in rehabilitation therapyhat provides interaction between virtual objects and motionsing motion tracking [16–22]. This technique would be moreuitable for the proprioception rehabilitation because of itsbility to manipulate the visual feedback of virtual objects [23].n addition, the VR technique allows an objective assessments well as efficient rehabilitation training because a patientan confirm his own movement without the assistance of aherapist and also view the training results in near real time.

In this study, we developed a VR proprioceptive rehabil-tation system that could manipulate the visual feedbackuring upper-limb training and ask the subject to rely only onroprioception feedback. We also demonstrated the effect ofroprioception training on stroke patients using a developedR system that provides proprioception feedback.

. Methods and materials

.1. Subjects

n order to evaluate the developed proprioception feedback vir-ual environment system, we recruited 10 stroke patients (age:4.7 ± 7.83 years, onset: 3.29 ± 3.83 years) as shown in Table 1nd 10 healthy age-matched subjects (age: 56.4 ± 4.53 years).he stroke patients; (1) who had suffered a primary ischemicr hemorrhagic stroke as diagnosed by magnetic resonance

maging image scans or computed tomography; (2) presentedild to severe paresis of the upper extremity and lacked any

dditional neurological disease causing motor deficits; (3) canerform the active flexion of affected elbow more than 50◦; (4)ho was in more than 10 weeks from stroke onset becauseost motor recovery is almost completed within 10 weeks

oststroke [24,25]; (5) showed no deficits in visual field byisual field examination; (6) showed no severe defects in cog-itive function with a Mini-Mental Status Examination score

26] >24; (7) showed no neglect by Albert test [27] or apraxia28]; (8) showed no serious depression by the Beck Depres-ion Inventory test [29]; (9) had no pain and dysfunction ofpper extremity by peripheral neuropathy, a rotator cuff tearf shoulder and complex regional pain syndrome; and (10) A

ummary of demographic variables and clinical measure forhe stroke group is included in Table 1. Most patients had lowDI scores (9–28). All subjects that consented to participate

n this study were informed about the experimental protocol,

Tabl

e

1

– S

P1

MP2

MP3

FP4

MP5

MP6

MP7

MP8

FP9

Fp

10

M

Rt,

righ

t;

of

30).

260 c o m p u t e r m e t h o d s a n d p r o g r a m s i n

Fig. 1 – VR-based proprioception training system. Theaffected hand performed a reaching movement towards thetarget position (virtual cylinder). When the subject believesthat the affected hand position corresponds to the target

position, unaffected hand would press the mouse button.

which was approved by the Department of Rehabilitation ofEulji Hospital. Healthy subjects had no history of neurologicaldiseases.

2.2. VR proprioception rehabilitation system

The virtual environment was designed to switch visual feed-back on/off during upper-limb training. The subjects’ armswill be positioned below an upper board to be out of view asshown in Fig. 1. The affected hand positions were tracked witha magnetic 3D position sensor (patriot 6-DOF tracker, POLHE-MUS, USA). On screen, a virtual cylinder moved according tothe position of the subject’s hand under the board. With theother hand, a button was installed for the subject to respondto designed tasks. VR tasks were programmed with 3D GameStudio (Conitec Datasystems Corp., Germany).

2.3. VR proprioception rehabilitation tasks

Two types of VR training tasks were designed to evaluatethe effect of the proprioception focused training comparedto visual feedback focused training. The VR environment wasdesigned like a living room environment which patients wereaccustomed. And the cylinder was used as a waste box inthe room in order to enhance the skill of daily life. In thevisual feedback virtual environment (VFVE) task, there is asemi-transparent cylinder showing the present position ofthe affected hand and opaque cylinder as the target posi-tion as shown in Fig. 2a. The subjects were asked to movetheir affected hand into the position of the opaque cylinder.

The subjects could see the semitransparent cylinder movingaccording to their arm’s movement as a visual feedback. Thesubjects pressed the mouse button when they believed theaffected hand position corresponded to the target position. If

b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265

the distance between the affected hand position and targetposition were within a predefined criterion, the opaque cylin-der would reappear. If they failed, repeated the process tillthey succeeded. On the contrary, in the proprioception feed-back virtual environment (PFVE) task, both semi-transparentand opaque cylinders are shown at staring point but the trans-parent cylinder disappears as soon as they start to movetheir hand into target position. Thus, they have to estimatethe current position and get to the target position by rely-ing on their own proprioceptive feedback information. Similarto VFVE, the subjects pressed the mouse button when theybelieved they reached the target position. If they failed, thesemi-transparent cylinder will be shown for 500 ms for thesubjects to check the current hand position and continue tillthey reached the target position. In both environments, twentytarget positions were provided. For performance analysis, thetotal number of trials was counted during twenty target tasks,which was the total number of mouse button clicks. In addi-tion, error distance measured the distance between targetposition and affected hand position when subjects pressedinitial-button. Total error distance was calculated by accu-mulating error distance every time the subjects pressed themouse button as confirmation. Lastly, movement distance wasmeasured from start position to affected hand position movedinitially. The total movement distance was measured by sum-ming the total movement distance from start position untilthe affected hand reached the target position successfully.

2.4. Experiments

2.4.1. Evaluation of characteristics of proprioceptionfeedback virtual environmentIn order to evaluate whether parameters measured in PFVEare related with proprioception function, all stroke patientssubjects performed the PFVE task as well as an elbow angleassessment task previously developed for the assessmentof proprioception function [23]. All parameters from PFVEwere compared with error angle from angle assessment task,which has a close relationship with proprioception func-tion [23,30]. In addition, occupational therapist measured theupper extremity FMS (Fugl-Meyer Assessment Scale) [31] andBox and Block test [32], Jebsen-Taylor Hand Function test [33]to measure the upper extremity function in stroke patients.Healthy subjects only performed the PFVE task in order tocompare with the behavioral characteristics of stroke patients.

2.4.2. Evaluation of the therapeutic effects ofproprioception feedback virtual environmentTen stroke patients participated in the experiment for eval-uation of the effects of proprioception feedback virtualenvironment training. They were randomly assigned into twogroups (five patients per group) in order to randomized theorder of training. One group performed VFVE five times dur-ing the first week. Then, the week after, the group did PFVE

training five times. The other group performed the tasks inreverse order. All patients performed the VFVE task and PFVEtask three times in order to check the improvement of propri-oception function. The Box and Block test and Jebsen-Taylor

c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265 261

Fig. 2 – Two virtual rehabilitation tasks. (a) Visual feedback virtual environment (VFVE): a visible semi-transparent cylindercorresponding to the current position of affected hand as a visual feedback. (b) Proprioception feedback virtual environment( comet ing o

ht

2

TFuicfbV

3

3e

CPad(t

PFVE): a semi-transparent cylinder is visible initially but bearget position. Subjects estimate the target position by rely

and function test was conducted with PFVE before and afterraining.

.5. Data analysis

he Pearson’s correlation has been used to compare betweenMS score or error angle and parameters measured in PFVEsing the SPSSWIN 18.0 software package. In addition, the

ndependent t-test has been used to compare the behavioralharacteristics between stroke patients and healthy controlrom PFVE. The repeated ANOVA and one sample T-test haveeen used in order to compare the training effects betweenFVE and PFVE.

. Results

.1. Characteristics of proprioception feedback virtualnvironment

omparison between an elbow angle assessment task andFVE, is shown in Table 2. The error angle from the elbow angle

ssessment task had a significant relationship with the erroristance (r = .640, n = 10, p = .046) and total movement distance

r = .751, n = 10, p = .012). Whereas in PFVE, it had no correla-ion with movement distance, click count or the total error

Table 2 – The correlation between two assessments (elbow angfeedback virtual environment.

Errordistance(r/p-value)

Movementdistance(r/p-value)

Cl(r/

Error angle 0.64/0.046* 0.36/0.304 0.5FMS score −0.66/0.037* −0.51/0.127 −0

∗ p < 0.05.

s invisible when the subject begins to move its hand inton their own proprioceptive feedback information.

distance. The two parameters (error distance and total move-ment distance) also significantly correlated with FMS scorereflecting the severity of stroke, respectively r = −.662, p = .037and r = −.726, p = .018. In addition, FMS score correlatedwith total error distance (r = −.714, p = .002) and click count(r = −.659, p = .038) but not with movement distance among theparameters measured in PFVE.

The comparison between stroke patients and healthycontrol subjects in PFVE shown in Table 3 below showsthat the total movement distance of stroke patientsincreased significantly compared to healthy control group(t = 1.87, p = 0.044). Additionally, movement distance instroke patient group increased compared to the healthycontrol group (t = 1.818, p = 0.043). However, the clickcount, error distance and total error distance of strokepatients did not significantly increase compared to thehealthy control, click count of stroke patients showedan increase pattern rather than healthy control (t = 1.488,p = 0.077).

3.2. Therapeutic effect of proprioception feedback

virtual environment rehabilitation

In order to evaluate the therapeutic effect of PFVE, we com-pared VFVE and PFVE after training stroke patients for two

le assessment, FMS) and parameters from proprioception

ick countp-value)

Total errordistance(r/p-value)

Total movementdistance(r/p-value)

5/0.095 0.62/0.057 0.75/0.012*

.66/0.038* −0.71/0.02* −0.73/0.018*

262 c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265

Table 3 – Comparison of behavioral characteristics between stroke patients and healthy control in proprioceptionfeedback virtual environment.

Stroke patientsaffected (N = 10)

Healthy controlaffected-cor (N = 10)

t p-Value(1-tailed)

Click count (number) 3.34 ± 1.31 2.58 ± 0.95 1.488 0.077Error distance (cm) 10.58 ± 3.48 9.02 ± 2.06 1.214 0.120Movement distance (cm) 42.20 ± 9.08 36.53 ± 3.82 1.818 0.043*

Total error distance (cm) 27.32 ± 15.29 18.75 ± 10.12 1.477 0.078Total movement distance (cm) 64.89 ± 32.41 44.43 ± 8.44 1.878 0.044*

ding u

Affected, affected upper extremity of stroke; Affected-cor, correspon∗ p < 0.05.

weeks with 2 × 2 ANOVA analysis with two factors (pre-postand feedback methods). Because total sample size is only ten,a normality of the data was assessed and confirmed for aparametric statistical analysis using the Kolmogorov-Smirnovtest. As shown in the Fig. 3(a), the error distance in PFVE sig-nificantly decreased in post-training rather than pre-training(F(1,9) = 25.194, �2 = .737, p = 0.001). In addition, there was asignificant interaction between two factors (F(1,9) = 9.924,�2 = .524, p = 0.012). Comparing the training effects betweena virtual environment with visual feedback and with pro-prioception, the click count, error distance and total errordistance were reduced more after training in PFVE than VFVE.(Click count: p = 0.005, error distance: p = 0.001, total errordistance: p = 0.007.)

3.3. Improvement of upper-limb function of stroke

patients

Therapeutic effect of the VR program was proven with con-ventional behavioral tests (Box and Block test, Jebsen-Taylor

Fig. 3 – Improvement effect of upper-limb of stroke patient in twError distance decreased more in PFVE training than VFVE trainibetween VFVE and PFVE (p = 0.005), (c) comparison of reduced ercomparison of reduced total movement distance between VFVE

pper extremity of age-matched control.

Hand Function test). As shown in Fig. 4a, upper-limb functionof stroke patients in the Box and Block test showed improve-ment pattern after training (n = 10, t = −2.05, p = 0.071). Becausetwo patients had difficulty grasping a block with an affectedhand (very low Box and Block test scores were shown inFig. 4a), re-analysis was performed with eight stroke patientsexcluding the two. With this adjustment, Box and Block scoreswere significantly improved after VR training (n = 8, t = −3.567,p = 0.009). On the other hand, the unaffected hand of strokepatients did not show improvement pattern after training(n = 10, t = 0.525, p = 0.613).

Seven stroke patients completed the Jebsen-Taylor handfunction test. Because two patients could not write with theaffected hand and one patient did not do the post-test, threepatients’ data were omitted. As shown in Fig. 4b, affected handfunction of stroke patients by writing subtest of Jebson-Taylor

test improved after training (n = 7, t = 2.508, p = 0.046). On theother hand, unaffected hand of stroke patients didn’t (n = 7,t = −1.442, p = 0.199). Other tests of Jebsen-Taylor also did notimprove.

o different virtual environments (*: p < 0.05, **: p < 0.01) (a)ng (p = 0.001), (b) comparison of reduced click countror distance between VFVE and PFVE (p = 0.001), (d)and PFVE (p = 0.007).

c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265 263

Fig. 4 – Evaluation of improvement of upper-limb functions (**: p < 0.01). (a) Box and Block scores improved significantly afterVR training (p = 0.009). Two stroke patients were omitted because their occupational therapists reported have troublegrasping the block with the affected hand. (b) Writing performance time of Jebsen-Taylor Hand Function test weresignificantly reduced after VR training (p = 0.046).

4

Ittio

ttivtoos

. Discussion

n this study, we developed a new type of rehabilitation systemhat focuses on the proprioception of stroke patients usinghe VR technology. The developed VR system was proven tomprove the motor control of stroke patients after VR propri-ception feedback training.

One of the important features of the developed system ishe use of proprioception feedback for upper limb rehabilita-ion by blocking visual feedback. VR proprioceptive paradigms novel because this type of exercise system cannot be pro-ided easily in a real environment. In order to perform a motorask, the central nervous system integrates multiple modes

f sensory information, especially vision and proprioceptionf a healthy individual [2,34,35]. However, the multiple sen-ory information integration could be changed in the patient

with the impairment of somatosensory information integra-tion after stroke, whereas stroke patients can use intact visualinformation rather than somatosensory. Stroke patients tendto excessively rely on visual input in motor task [11]. Bonanet al. suggested that visual influence is known to become pre-dominant when afferent input from other sources is reducedand the predominant influence of visual input constitutes anatural compensatory strategy for coping with initial strokedamage [10,11]. Traditional rehabilitation reinforces the exces-sive visual reliance by focusing on visual compensation ratherthan restoring sensory inputs that would have been nor-mally used. Therefore, the control of visual information suchas blocking visual feedback could play an important role inforcing patients to rely more on proprioception and learn

movement efficiently. In addition, previous work has shownthat subjects do not rely on visual feedback for learning novelmovement [33,36–39]. Even congenitally blind individuals are

s i n

r

264 c o m p u t e r m e t h o d s a n d p r o g r a m

able to adapt to the perturbing effects of a force field producedby a rotated room.

In our results, the significant decrease of error distanceand the total error distance shown after just proprioceptionfeedback rehabilitation training, suggests that the propriocep-tion feedback rather than visual feedback could be an effectivemeans to enhance motor control during rehabilitation training[40]. It could be explained by the fact that cortical excitabil-ity increase in the condition without visual feedback ratherthan with visual feedback in our previous study [41]. To ourknowledge, there are few rehabilitation systems that focus onenhancing proprioception itself in stroke patients by blockingview of arm movements. Jun et al. demonstrated the effect ofproprioception feedback training without visual feedback instroke patients [42].

One of the other advantages in the developed VR rehabilita-tion training is the ability to evaluate proprioception functionof stroke patients as well as training itself. In order to provethis, it was compared to a previously developed upper extrem-ity proprioceptive assessment system, which include anglesassessment task which is necessary to match imposed jointpositions without visual feedback to check for angle error[23]. Among all other parameters objective parameters forevaluating the proprioceptive assessment in PFVE, total move-ment distance had a significant relationship with error angleobtained from previous elbow angle assessment task as shownin Table 3. Most of all it significantly correlated with the FMSscore (Table 2) and had higher value compared to healthy sub-jects (Table 3). Moreover, previous study verified the significantdifferences of click count and error distance between affectedupper extremity of stroke group and control group in largegroup (n = 30) [30]. Stroke patients tend to take couple triesto reach to the target due to the damage of proprioception(the number of trials, p = 0.07) and this lead to the total move-ment distance and error distance which were accumulated.This explains why it could be sensitive to the functionality ofthe upper limb and could be used as an index parameter toevaluate proprioception of stroke patients.

Despite the results showing the improvement of propri-oception function after PFVE training, careful considerationshould be taken in interpretation because the PFVE was usedas an evaluation tool after training session. There is a possi-bility that subjects became accustomed to the PFVE and knewhow to adjust their arm to target without visual feedbackto get a high score. A therapeutic effect was also demon-strated in the conventional behavioral tests (Box and Blocktest, Jebsen-Taylor test) before and after training as shown inFig. 4., although two patients were excluded in the statisticalanalysis because they were not able to perform the con-ventional behavioral tests (Box and Block test, Jebsen-Taylortest). This corresponded to the previous study that showedimprovements after rehabilitation training with sensory feed-back using a VR system, even in stages [43–45].

An insufficient number of subjects and the heterogeneityof the lesions, onset time, manual dexterity, functional levelmay represent the limitations of our study. We did try to min-

imize the large variability effect due to individual differenceincluding motor recovery influence in the experimental con-dition and analysis method by performing the experimentin short period, recruiting the patients who are in almost

b i o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265

completed motor recovery. In addition, significant differenceswere not consistently found between stroke and age–matchedcontrol group for the major proprioceptive indices due toinsufficient number of subjects. Furthermore, it is hard to gen-eralize the efficacy of PFVE training in stroke patients becausethere is no comparison study with conventional rehabilita-tion therapeutic methods. In future investigations, clinicalstudies examining the effectiveness of the proprioceptive VRrehabilitation program according to various stroke subgroups,systemic experimental design, and training duration shouldbe considered, and a method that enhances the effect via inte-gration with other treatment methods, such as robot therapyand the various feedback paradigms, should be developed.

Conflicts of interest

None.

Acknowledgements

This work was supported by the National Research Foundationof Korea Grant funded by the Korean Government (NRF-220-2008-D00112).

e f e r e n c e s

[1] A. Dogan, et al., The rehabilitation results of hemiplegicpatients, Turkish J Med Sci 34 (6) (2004) 385–389.

[2] S.J. Sober, P.N. Sabes, Multisensory integration during motorplanning, J Neurosci 23 (18) (2003) 6982–6992.

[3] R.W. Fremerey, et al., Proprioception after rehabilitation andreconstruction in knees with deficiency of the anteriorcruciate ligament: a prospective, longitudinal study, J BoneJoint Surg Br 82 (6) (2000) 801–806.

[4] M. Bayramoglu, R. Toprak, S. Sozay, Effects of osteoarthritisand fatigue on proprioception of the knee joint, Arch PhysMed Rehabil 88 (3) (2007) 346–350.

[5] L.M. Carey, L.E. Oke, T.A. Matyas, Impaired limb positionsense after stroke: a quantitative test for clinical use, ArchPhys Med Rehabil 77 (12) (1996) 1271–1278.

[6] R.L. Sainburg, et al., Control of limb dynamics in normalsubjects and patients without proprioception, JNeurophysiol 73 (2) (1995) 820–835.

[7] N. Leibowitz, et al., Automated measurement ofproprioception following stroke, Disabil Rehabil 30 (24)(2008) 1829–1836.

[8] J. Kim, C.M. Sapienza, Implications of expiratory musclestrength training for rehabilitation of the elderly: tutorial, JRehabil Res Dev 42 (2) (2005) 211–224.

[9] J.L. Patton, F.A. Mussa-Ivaldi, Robot-assisted adaptivetraining: custom force fields for teaching movementpatterns, IEEE Trans Biomed Eng 51 (4) (2004) 636–646.

[10] I.V. Bonan, et al., Reliance on visual information after stroke.Part I: balance on dynamic posturography, Arch. Phys. Med.Rehabil. 85 (2) (2004) 268–273.

[11] I.V. Bonan, et al., Reliance on visual information after stroke.Part II: effectiveness of a balance rehabilitation program withvisual cue deprivation after stroke: a randomized controlled

trial, Arch. Phys. Med. Rehabil. 85 (2) (2004) 274–278.

[12] T.E. Hewett, M.V. Paterno, G.D. Myer, Strategies forenhancing proprioception and neuromuscular control of theknee, Clin. Orthop. Relat. Res. 402 (2002) 76–94.

i n b i

c o m p u t e r m e t h o d s a n d p r o g r a m s

[13] R.A. Schmidt, T.D. Lee, Motor control and learning, HumanKinetics, Champaign, IL, 2005.

[14] J.D. Wong, et al., Can proprioceptive training improve motorlearning? J. Neurophysiol. 108 (12) (2012) 3313–3321.

[15] M.D. Lewek, et al., Use of visual and proprioceptive feedbackto improve gait speed and spatiotemporal symmetryfollowing chronic stroke: a case series, Phys. Ther. 92 (5)(2012) 748–756.

[16] M.C. Moreira, et al., Use of virtual reality in gait recoveryamong post stroke patients – a systematic literature review,Disabil. Rehabil. Assist. Technol. (2013).

[17] K.E. Laver, et al., Virtual reality for stroke rehabilitation,Cochrane Database Syst. Rev. 9 (2011), p. CD 0083 49.

[18] S.E. Kober, et al., Virtual reality in neurologic rehabilitationof spatial disorientation, J. Neuroeng. Rehabil. 10 (2013) p17.

[19] E. Wade, C.J. Winstein, Virtual reality and robotics for strokerehabilitation: where do we go from here? Top. StrokeRehabil. 18 (6) (2011) 685–700.

[20] G. Saposnik, M. Levin, Virtual reality in stroke rehabilitation:a meta-analysis and implications for clinicians, Stroke 42 (5)(2011) 1380–1386.

[21] L.F. Lucca, Virtual reality and motor rehabilitation of theupper limb after stroke: a generation of progress? J. Rehabil.Med. 41 (12) (2009) 1003–1100.

[22] S. Subramanian, et al., Virtual reality environments forpost-stroke arm rehabilitation, J. Neuroeng. Rehabil. 4 (2007)p20.

[23] K.H. Lee, et al., Upper extremity proprioceptive assessmenttest using virtual environment technique in patients withstroke, J. Korean Acad. Rehabil. Med. 34 (2) (2010)141–149.

[24] G. Kwakkel, B.J. Kollen, Predicting activities after stroke:what is clinically relevant? Int. J. Stroke 8 (1) (2013) 25–32.

[25] P. Langhorne, J. Bernhardt, G. Kwakkel, Stroke rehabilitation,Lancet 377 (9778) (2011) 1693–1702.

[26] M.F. Folstein, S.E. Folstein, P.R. McHugh, Mini-mental state. Apractical method for grading the cognitive state of patientsfor the clinician, J. Psychiatr. Res. 12 (3) (1975) 189–198.

[27] M.L. Albert, A simple test of visual neglect, Neurology 23 (6)(1973) 658–664.

[28] Q.J. Almeida, S.E. Black, E.A. Roy, Screening for apraxia: ashort assessment for stroke patients, Brain Cogn. 48 (2–3)(2002) 253–258.

[29] H.M. Hahn, et al., A standardization study of Beckdepression inventory in korea, J. Korean Neuropsychiatr.Assoc. 25 (3) (1986) 487–500.

[30] D.J. Gladstone, C.J. Danells, S.E. Black, The fugl-meyerassessment of motor recovery after stroke: a critical reviewof its measurement properties, Neurorehabil. Neural Repair16 (3) (2002) 232–240.

o m e d i c i n e 1 1 3 ( 2 0 1 4 ) 258–265 265

[31] A.R. Fugl-Meyer, et al., The post-stroke hemiplegic patient,1. a method for evaluation of physical performance, Scand. J.Rehabil. Med. 7 (1) (1975) 13–31.

[32] V. Mathiowetz, et al., Adult norms for the Box and Block testof manual dexterity, Am. J. Occup. Ther. 39 (6) (1985)386–391.

[33] R.H. Jebsen, et al., An objective and standardized test ofhand function, Arch. Phys. Med. Rehabil. 50 (6) (1969)311–319.

[34] M.S. Graziano, Where is my arm? The relative role of visionand proprioception in the neuronal representation of limbposition, Proc. Natl. Acad. Sci. U.S.A. 96 (18) (1999)10418–10421.

[35] S.J. Sober, P.N. Sabes, Flexible strategies for sensoryintegration during motor planning, Nat. Neurosci. 8 (4)(2005) 490–497.

[36] S. Bamford, et al., The COSMIC (catalogue of somaticmutations in cancer) database and website, Br. J. Cancer 91(2) (2004) 355–358.

[37] R. Osu, et al., Random presentation enables subjects toadapt to two opposing forces on the hand, Nat. Neurosci. 7(2) (2004) 111–112.

[38] R.A. Scheidt, et al., Interaction of visual and proprioceptivefeedback during adaptation of human reaching movements,J. Neurophysiol. 93 (6) (2005) 3200–3213.

[39] C. Tong, D.M. Wolpert, J.R. Flanagan, Kinematics anddynamics are not represented independently in motorworking memory: evidence from an interference study, J.Neurosci. 22 (3) (2002) 1108–1113.

[40] L.B. Bagesteiro, F.R. Sarlegna, R.L. Sainburg, Differentialinfluence of vision and proprioception on control ofmovement distance, Exp. Brain Res. 171 (3) (2006) 358–370.

[41] Y.J. Kang, et al., Upper extremity rehabilitation of stroke:facilitation of corticospinal excitability using virtual mirrorparadigm, J. Neuroeng. Rehabil. 9 (2012) 71.

[42] A.Y. Jun, K.I. Jung, E.H. Choi, The effect of visual feedbacktraining on proprioception of the wrist joint in strokepatients, Korean J. Stroke 5 (1) (2003) 78–82.

[43] M.S. Cameirao, et al., The combined impact of virtual realityneurorehabilitation and its interfaces on upper extremityfunctional recovery in patients with chronic stroke, Stroke43 (10) (2012) 2720–2728.

[44] M. da Silva Cameirao, et al., Virtual reality basedrehabilitation speeds up functional recovery of the upperextremities after stroke: a randomized controlled pilot studyin the acute phase of stroke using the rehabilitation gaming

system, Restor. Neurol. Neurosci. 29 (5) (2011) 287–298.

[45] K. Eng, et al., Interactive visuo-motor therapy system forstroke rehabilitation, Med. Biol. Eng. Comput. 45 (9) (2007)901–907.