Virtual reality task for telerehabilitation dynamic balance training in stroke subjects

Imre Cikajlo, Marko Rudolf, Nika Goljar, Zlatko Matjačić Institute for rehabilitation, Republic of Slovenia

Ljubljana, Slovenia e-mail: imre.cikajlo@ir-rs.si

Abstract—The paper presents a virtual reality based dynamic balance training. The telerehabilitation balance training in stroke subject took place in smarthome, simulating a home environment. Virtual environment was designed as a game in web-explorer enabling the medical professionals to remotely supervise and control the balance training. In preliminary testing a right-side hemiparetic subject participated and demonstrated high motivation, high level of learning and accomplished the therapy with promising clinical outcomes. The subject performed the therapy five times a week, each time for 17 to 20 minutes for four weeks. The results were evaluated by objective game parameter as track time, number of hits and clinical instruments Berg Balance Scale, Timed Up&Go and 10m Walk tests. The outcomes indicate similar progress to those obtained in the clinical environment.

Keyword: virtual reality, balance, telerehabilitation, stroke

I. INTRODUCTION One of the important issues in stroke population is balance

control. The loss of balance in most cases results in fall and consequently in injuries. Therefore the restoration of static and dynamic balance is important for restoration of functional capabilities of stroke subjects. The outcome of research performed in the USA and UK in a group of stroke subjects demonstrated that functional capabilities may improve with rehabilitation in acute and chronic phase, when intensive therapy with repeatable and targeted tasks are applied [1]. The therapy in these conditions in clinical environment can be assured almost only by applying an assistive device that from one point assure safety on the other provide an option for targeted tasks in repeatable conditions. Among these devices are rehabilitation robots and passive devices, e.g. dynamic balance trainer [2]. The dynamic balance trainer enable the stroke subjects to stand in the vertical position providing support at the pelvis level and help the physiotherapist at realization of static and dynamic balance training.

Balance training tasks in clinical environment traditionally comprise stepping on various surface, foam, and uneven

surface, head movement, catching or holding a ball. Similar, but more targeted and repeatable tasks for the rehabilitation of stroke subjects can be realized in virtual environments (VE) using virtual reality (VR) computer technology [3-5]. VR technology enables accurate task design, coordination, parameters change, task control and on the other hand relieve the physiotherapist of strenuous work keeping the task within repeatable conditions. Besides, the VE can be anytime adapted to the subject’s special needs and therapy requirements.

VE presents also a motivation factor, especially if designed as a game, where the subjects are required to beat their own score. This leaves the clinical staff some room to take care of more subjects at the same time. But as the technology is ready to set the VE based therapy out of the clinical environment when the subject is trained to the level that is safe to leave, the therapy can be monitored and controlled remotely – telerehabilitation [6]. Telerehabilitation comprises remote assistance, supervision, remote parameter setup, remote task control and daily instructions. Besides, as an option a videoconference call can be established.

In the world the most commonly used clinical tool to evaluate balance capabilities is Berg Balance Scale (BBS) [7], which requires a skilled medical professional. Burger et al [8] demonstrated that dynamic balance training with standing frame without additional assistance of the physiotherapist can be equally effective as classical clinical approach without any device. In the paper we present a VE based dynamic balance training in combination with potential telerehabilitation service providing objective score. The approach adding a VE motivation issue demonstrates one of the options how the standing frame based balance training may be transferred on subject’s home. According to the positive results with balance training in clinical environment we expected high level of motivation and similar improvement of the balance capabilities.

II. METHODOLOGY

A. Dynamic balance training The balance training was based on dynamic balance

training standing frame made of steel base construction placed on four wheels, which when unlocked enable the apparatus mobility. The later is important in clinical environment where the rehabilitation aids have no dedicated space. The standing

This work was supported in part by the Slovenian Research Agency and the Health Insurance Institute of Slovenia..

I. Cikajlo, PhD, M. Rudolf, PT, N.Goljar, MD, PhD and Z.Matjačić,. PhD are with the Institute for rehabilitation, Republic of Slovenia, Linhartova 51, SI-1000 Ljubljana, Slovenia, phone: +386 1 4758 150; fax: +386 1 4372 070; e-mail: imre.cikajlo@ ir-rs.si).

121978-1-4244-4189-1/09/$25.00 ©2009 IEEE

frame is made of aluminum and fixed to the base with passive controllable spring defining the stiffness of the two degrees of freedom (2 DOF) standing frame. The stiffness of the frame is set up according to the individual’s requirements. On the top of the standing frame a wooden table with safety lock for holding the subject at the level of pelvis was mounted.

The subject was standing in vertical position in the balance trainer with his hands placed on the wooden table in front of him and secured with safety lock from behind at the level of pelvis, preventing to fall backward (Fig. 1). The standing

frame can tilt in sagital and frontal plane for ± 15o. The tilt of the balance standing frame (BalanceTrainer, Medica Medizintechnik, Germany) was measured by commercially available three-axis tilt sensor (Xsens Technologies, Enschede, The Netherlands).

B. Virtual reality task The task for dynamic balance training was based on

subject’s movement, i.e. weight transfer in sagital and frontal plane, resulting in BalanceTrainer tilt. The tilt was assessed by sensor and the action immediately resulted in the designed virtual environment. The virtual reality (modeled in VRML 2.0, running in MS Internet Explorer with blaxxun contact plug-in) based task required from the subject to “walk” by tilting the frame forward and “turn” by tilting the frame in frontal plane. The task (Fig. 2) was divided in three difficulty levels, each comprising additional obstacles on the way.

In the first and the easiest level (level 1) there were no obstacles on the tree lined path, where the subject passed by the woman on the left and at the dolphin statue turned left and continue to the buffet, where the obstacles were two tables and chairs. Afterwards the subject made a turn around the tables and returned back to the dolphin statue and continued his way passing by the security guard to the entrance of the building. When the subject entered through the door, the task restarted from the begging. At the second level (level 2) four benches were added as obstacles which the subject needed to avoid. And the third level (level 3) added three additional cans and two pools near the dolphin statue (Fig 3).

C. Telerehabilitation The VE was designed to run in web-explorer and therefore

enable the medical professionals to supervise, monitor and control the balance training process. Besides, the videoconferencing enabled them to provide important advices to the subject during the dynamic balance training. The videoconference was established by using a free-ware Skype (Skype Technologies S.A., Luxembourg, EU). Full-duplex voice and video enabled the physiotherapist to correct the subject’s posture during the therapy.

Figure 1. Telerehabilitation in practice. A physiotherapist (upper photo) is supervising and assisting the subject remotely during dynamic balance training (lower photo) in Smarthome Iris [9]. The BalanceTrainer apparatus was equipped with tilt sensor and detected the subject’s movement. Adequately the VE changed and the subject was required to avoid obstacles and to reach the goal – the door in the building.

Figure 2. Virtual reality technology was used to implement the dynamic balance training task, where the subject “walked” through the scenery by tilting the apparatus. The task runs in web-explorer, which made it easy to implement it in the telerehabilitation service.

122

Figure 3. The top view of the VE task objects (* cans, + benches, O pools, x hydrant,…) with training tracks appended. The VR task started at the lower end of the figure and finished at the upper right corner with entering the building. It is obvious that the subject has hit the tables at the buffet twice, crossed the pool and hit the 2nd and 3rd can.

D. Subject In the pilot study a 47 year old subject (male, 180 cm, 80kg,

intracerebral hemorrhage a month before the therapy) with right side affected and a slight motor dysphasia and dysarthria. The subject has been cardiovascular compensated to achieve 30 minutes or more of moderate intensity physical activity. Besides had no additional diseases and took any medications that may affect his balance. His cognitive functions were well-preserved, allowing him to follow the given instructions.

The methodology was approved by local ethics committee and the subjects gave informed consent.

E. Protocol The testing of the VE based telerehabilitation dynamic

balance training took place in the SmartHome Iris [9], a well equipped demonstrating apartment for people with special needs. The subject was standing in the BalanceTrainer (Fig. 1 below), secured with safety rod at the pelvis level, and placed his hands at the front table. In front of the subject was a LCD screen with a multimedia camera with the microphone on the top. An engineer or occupational therapist who was responsible for the subject in the SmartHome Iris established a videoconference call with the physiotherapist who was located in the remote room. The physiotherapist could see the subject and provided him instructions how to correct his posture while the subject was performing the VE based task.

The subject performed the therapy five times a week, each time for 17 to 20 minutes. On the first week the subject used level 1 of the VR task for balance training, on the second week, when the physiotherapist estimated the subject’s progress, the training started with level 2 and on the third and fourth week with level 3.

The subject balance capabilities were also assessed with clinical instrument (BBS) [7] at the beginning of the therapy and after four weeks. Besides BBS, the standing alternatively on the healthy and the affected lower extremity, the »timed up & go« test [10] and the 10-m rapid walk test were also performed.

Between the VE task sessions the subject took part in other standard neurotherapeutic programs.

F. Data analysis The objective assessment of each task performed in the VE

was established on the basis of the time within a single task circle (from the start to entering through the door) was completed. Additionally the number of collisions with the objects was taken into account. The number of collisions was calculated offline after the subject finished the session. Each collision added a penalty time (+5s) to the achieved subject’s track time.

The mean and standard deviation were calculated for each VR task difficulty level (or week) to estimate the progress of the therapy. A paired T-test was performed on track time data to check the data relevancy.

III. RESULTS Fig. 4. shows the mean track times, achieved at the

beginning of the therapy and a week later (start end) for each difficulty level. In the first week the subject made an enormous progress, reducing the mean track time from 48s to 42s (p<0.05). On the second week the task was a step more difficult with obstacles on the way the score was still encouraging, from 56s to 52s (p = 0.031). The level 3 with even more obstacles,

123

Figure 4. The figure shows the mean track times achieved at VR task at different difficulty levels. Upper left demonstrates the successful task accomplishment with decreased mean track time for the level 1, upper right for the level 2 and lower graph for the task difficulty level 3.

which also lengthened the path a bit, also demonstrated progress from 53s to 51s (p = 0.0157).

The score of each training track was displayed in seconds and object hits and total score:

51s (3 hits +15s) = 66s

For the clinical purposes the therapy has been also evaluated with clinical instruments; BBS (Table I), Standing on a single extremity (Table II) and Timed Up&Go (TUG) test and 10m walk (Table III).

TABLE I. BERG BALANCE SCALE

BBS Assessment

Before therapy After therapy

value* 50 54

* maximal value of the BBS for neurologically intact subjects is 56.

TABLE II. STANDING ON SINGLE LIMB TEST

Standing on healthy (HE) and affected (AE) extremity Before therapy After therapy Paired T-value

HE 41.10 ± 9.24 46.82 ± 22.81 p= 0.71

AE 4.87 ± 2.89 13.63 ± 3.06 p= 0.023 *all values in second

TABLE III. TIMED UP AND GO TEST AND 10M WALK TEST

Timed up&go test and 10m walk test

Before therapy After therapy Paired T-value

TUG 14.93 ± 0.61 11.47 ± 0.64 p= 0.0024

10m Walk 12.57 ± 0.71 11.20 ± 0.10 p= 0.0298 *all values in second

IV. DISCUSSION The right-side affected subject had major problems with

balancing and weight transfer, especially when loading his affected extremity. Therefore the subject sometimes left the desired trajectory in VE and hit some obstacles. But as the subject was highly motivated, he improved his performance on a daily basis. The subject managed to accomplish the task faster, more accurate, even when the physiotherapist increased the difficulty level. The outcomes demonstrated that the subject achieved practically the same track time at the end of therapy with level 3 as at the beginning with level 1, where no obstacles were present. This suggests that the subject has completely mastered the VR task. Besides the objective engineering indicators also the clinical instruments revealed improvement of the subject’s clinical status; improved weight balance – standing on the affected extremity, faster in timed up&go and 10m walk test and increased BBS. Especially the 10m walk might be very relevant when considering that it is argued that timing of gait over 10 meters is a valid reliable measure that is currently underused [11].

The outcomes of the pilot study are comparable with the results of the study in stroke subject, where classical balance therapy was compared to balance training with the BalanceTrainer [8]. Besides we are aware that stroke subjects require full rehabilitation treatment, not only balance therapy and that in the proposed approach the only participating subject had functionally acceptable balance capabilities. But in general the subjective physical experience of moving and the limitation in field of view may also have had impact on performance [12] as well as the problems in cognitive processing due to conversion of tilting into “walking”. However, the rehabilitation in virtual environment or using virtual reality for

124

replacing real world objects may not be always transferrable real world applications or be effective immediately in the real world [13]. But exactly the virtual reality technology is the one that enables repeatability, target orientated rehabilitation, adding difficulty levels and remote activity, the advantages that are important for rehabilitation in acute phase, where the intensive therapy should be functionally targeted. There has been reported that task using virtual reality technology also improves gait in post-stroke subjects [14] and that VR in combination with robotic device demonstrated better results in chronic stroke subjects than robot therapy alone [15].

Perhaps the pilot study with the subject with functional balance capabilities may not fully demonstrate the advantages of virtual reality based telerehabilitation balance training, but in combination with the short term full rehabilitation treatment in subject’s acute stage in clinical environment and later homecare can reduce cost and contribute to subject’s individual independence.

REFERENCES

[1] G. Kwakkel, R.C. Wagenaar, J. W. R. Twisk, et. al., “Intensity of leg and arm training after primary middle- cerebral- artery stroke: A randomised trial.” Lancet, vol. 354: pp. 191- 196 , 1999

[2] Z. Matjačić, I.L. Johannesen, T. Sinkjaer, “A multi - purpose rehabilitation frame: A novel apparatus for balance training during standing of neurologically impaired individuals.” J. Rehab. Res. Dev., vol. 37, pp. 68-91, 2000,

[3] M. K. Holden, “Virtual environments for motor rehabilitation: Review.” Cyber Psychol Behav., vol.8, pp.187-211, 2005,

[4] M. Holden, E. Todorov, J. Callahan, E. Bizzi, “Virtual environment training improves motor performance in two patients with stroke: Case report.” Neurology Report , vol. 23, pp. 57-67, 1999.

[5] D. Jack, B. Rares, A.S. Merians, M. Tremaine, G.C. Burdea, et. al., “Virtual reality- enhanced stroke rehabilitation.” IEEE Trans. Neural. System. Rehabil. Engineering, vol. 9, pp. 308-318, 2001,

[6] M. J. Rosen,. “Telerehabilitation.” NeuroRehabilitation, vol. 3, pp. 3-18, 1999,

[7] K. O. Berg, S.L. Wood-Dauphinee, J.I. Williams, B. Maki, “ Measuring balance in the eldery: validation of an instrument.” Can. J. Public. Health. vol 25, pp. 7-11, 1992,

[8] H. Burger, N. Goljar, M. Rudolf, I. Stanonik, Z. Matjačić, “Balance training in stroke patients.” V: KULLMANN, Lajos (ur.), BURGER, Helena (ur.). Proceedings of the 9th Congress of European Federation for Research in Rehabilitation, Budapest, Hungary, 26-29 August 2007, (International journal of rehabilitation research, Vol. 30, Suppl. 1). London: Lippincott Williams & Wilkins, 2007, pp. 17-18

[9] A. Zupan, R. Cugelj, F. Hočevar, “Dom IRIS smart home.” Quark (Engl. ed.). [English ed.], summer 2008, pp. 132-142. Available: http://www.ir-rs.si/en/smarthome,

[10] S.S. Ng, C.W. Hui-Chan, “The timed up & go test: its reliability and association with lower-limb impairments and locomotor capacities in people with chronic stroke”. Arch Phys Med Rehabil., vol. 86, no. 8, pp- 1641-1647, 2005

[11] D. T. Wade, V.A. Wood, A. Heller, J. Maggs, R. Langton Hewer. “Walking after stroke: measurement and recovery over the first 3 months.” Scand. J. Rehabil. Med. Vol 19, pp. 25-30, 1987,

[12] L. Nyberg, L. Lundin-Olsson, B. Sondell, A. Backman, K. Holmlund, S. Eriksson, M. Stenvall, E. Rosendahl, M. Maxhall, G. Bucht, “Using a virtual reality system to study balance and walking in a virtual outdoor environment: a pilot study”., Cyberpsychol Behav. Vol. 9, no. 4, pp-. 388-395, 2006,

[13] F. D. Rose, E.A. Attree, B.M. Brooks, D.M. Parslow, P.R. Penn, N. Ambihaipahan, “Training in virtual environments: Transfer to real world tasks and equivalence to real task training.” Ergonomics, vol. 43, pp. 494-511, 2000,

[14] J.E. Deutsch, C. Paserchia, C. Vecchione, et al. “Improved gait and elevation speed of individuals post-stroke after lower extremity training in virtual environments.” JNPT. vol. 28, pp. 185-186, 2004,

[15] A. Mirelman, P. Bonato, J.E. Deutsch, “Effects of training with a robot-virtual reality system compared with a robot alone on the gait of individuals after stroke.” Stroke. Vol 40, No. 1, 2009,

125