7

Click here to load reader

[IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

  • View
    214

  • Download
    0

Embed Size (px)

Citation preview

Page 1: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

Abstract—We present an at-home assessment system for upper extremity rehabilitation with simple motion capture that is functional and affordable. This system uses lighted targets to initiate a reach to grasp (or touch, if the patient is unable to grasp) to three touch- and force-sensitive cones. During the reach, end-point reach trajectory is captured using a low-cost, custom-built infrared motion capture system using two Wii Remotes. An embedded computer collects data for tracking patient’s progress over time. The system is a low cost way to track reaching trajectory, reaching time, reaction time and relative grasp forces. It requires minimal setup and instruction.

I. INTRODUCTION TROKE is the number one cause of disabilities in American adults. Because stroke mortality rates have decreased in recent decades due to improved medical

interventions, millions of Americans survive the accident to live with chronic upper limb disabilities that affect activities of daily living (ADL) [1]. These disabilities vary greatly among stroke survivors and affect a wide range of functional movements, posing a great challenge for developing comprehensive assessment devices for upper limb functions that address every individual’s motor impairments.

Assessing upper limb function is critical for evaluating rehabilitation outcomes and functional abilities during recovery. A physical therapist usually makes this evaluation in a rehabilitation clinic. While this provides essential information about the patient’s recovery, long-term rehabilitation can be prohibitively expensive since traditional insurance programs do not cover this type of care. Additionally, traveling to a rehabilitation clinic may be impossible for those with more severe disabilities and may be an additional burden to their caretaker or spouse. Another disadvantage to this method is that it requires one physical therapist per patient, which does not allow the therapist to treat a number of patients at once. One solution to the problems associated with the standard system of care is to

Manuscript received April, 16, 2008. Suneth Attygalle and Margaret Duff are full-time PhD students with the Harrington Department of Bioengineering and Arts, Media & Engineering Program at Arizona State University, Tempe, AZ 85287 USA. (phone: 480-965-9438, fax: 480-965- 0961, e-mail: [email protected], [email protected]).

Thanassis Rikakis is with the Arts, Media & Engineering Program at Arizona State University, Tempe, AZ 85287 USA. (e-mail: [email protected]).

Jiping He is with the Harrington Department of Bioengineering at Arizona State University, Tempe, AZ 85287 USA. (e-mail:[email protected]).

*These authors contributed equally to this work.

develop a simple and low-cost, assessment system that a patient could use at home. A physical therapist or doctor monitors the system in a process known as telerehabilitation [2].

Such a system provides long-term information on how a patient is recovering, or, if used after a rehabilitation program, how much of the therapy is retained. It also allows a patient to use the system at their leisure and in the comfort of their own home, providing data that is preferable to data from the rehabilitation clinic setting [3]. Other telerehabilitation systems implemented to train or assess upper limb function at home [4] that use computer based programs [5, 6, 7, 8], robotics [9], or other complicated equipment [10] face many technical and practical challenges [11]. One group has used Wii remotes primarily for its 3-axis accelerometer [12] but did not use the Wii remotes as three-dimensional (3D) motion capture cameras as proposed in this work. Another group has used the Xbox videogame console as a low-cost platform for stroke rehabilitation that records finger positions [13]. Our system includes low-cost motion capture of the end-point trajectory, thus enabling future use for real time kinematic feedback for upper limb recovery.

Since the target population of our home assessment system is older people who may not be comfortable with technology and computers [14], we present our system as a tabletop game. Our system uses a hidden computer without a monitor and the user has no direct interaction with it. Thus, our system can be used as an intermediary step towards using more complex systems to acclimatize older people to using interactive technologies in the home.

Our at-home system is designed to follow an interactive, multi-modal, reach-to-grasp therapy program developed at Arizona State University (ASU) [15]. Our home system is a game that can be used on top of any flat surface such as a dining room table (See Fig. 1). The tabletop has three target cone positions and a rest position that were set during the therapy program. A low-cost motion capture setup monitors end-point trajectory during the reach and low-cost sensors track reaction and reaching times, as well as relative grasping force. A microcontroller and small computer collect information from the sensors and motion capture cameras and provide simple feedback to the patient about their movement. We propose this system as an easy to use and minimally intrusive assessment tool that provides informative data during and after upper extremity

Low-cost, at-home assessment system with Wii Remote based motion capture

Suneth Attygalle*, Student Member, IEEE, Margaret Duff*, Student Member, IEEE, Thanassis Rikakis, Jiping He, Senior Member, IEEE

S

978-1-4244-2701-7/08/$25.00 ©2008 IEEE 168

Page 2: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

rehabilitation programs for approximately $1200.

II. METHODS

A. Introduction of system The home system is an assessment tool to track functional

improvements over time and to track the retention of motor learning from an interactive multi-modal rehabilitation program for reaching and grasping. This system uses lighted targets to initiate a reach to grasp (or touch, if the patient is unable to grasp) to three touch- and force-sensitive cones. Reaching distance is maintained by monitoring a consistent rest position with a sensor. A low-cost, custom-built motion capture system using two Wii Remotes captures end-point trajectory. A small, hidden computer collects data for tracking patient progress over time. The system is a low-cost method to track reaching trajectory, reaching time, reaction time and relative grasp force. It requires minimal setup and instruction.

B. Biofeedback for Rehabilitation project at ASU The Arts, Media, and Engineering program at ASU has

developed an interactive multimedia rehabilitation program for reaching and grasping entitled Biofeedback for Rehabilitation [15]. This program uses audio and visual feedback to train stroke patients to modify their movement dynamics to use their distal joints more and with better synergy and to make their reaching movements closer to ideal trajectories. In the Biofeedback for Rehabilitation program, patients reach to different physical and virtual targets that are set according to their individual abilities and receive continuous feedback indicating measures of performance and results. Depending on the training task, the table height is set to one inch above their hanging olecranon, which allows their elbow to be supported by the table during rest periods, or one inch below so that the arm is unsupported. During this program, a physical therapist assesses the patients and notes the most significant movement impairments and the major strategies used to compensate for these impairments. The interactive feedback is then customized to focus on training the patient in the areas that need the most work. Patient progress is measured

using functional assessment scales as well as devices that measure grasp forces and reaction time. We use the information collected during these training sessions to customize our assessment system for the patient to bring home after they have completed training.

C. Physical Setup This system is portable and easily modified to meet any

patient’s reaching abilities. The system fits over any table so no special equipment is needed in the home. It is made from ¼” plywood, with a semi circle cut out for the torso. This mimics the table setup that the patients used during training. The targets are set based on the subject’s passive reaching abilities. Two targets (2 and 3) on the table are based on passive reaching to midline while a raised target (1) is based on an ipsilateral reach straight from the shoulder at a height of 6 inches. Target locations were chosen to elicit a range of reaching motions. The target locations are presented in Fig. 2.

The plywood has a grid of holes to secure the targets to the base (See Fig. 3.). The system is elevated 1” from the table to allow space for the embedded electronics. The system is secured to the table using custom made C-clamps. The protruding arm rest on the patient’s impaired side has an adjustable leg for support. The arm rest on the patient’s unaffected side has a hinge that opens for easy entry and exit. Once the target locations are determined, an overlay of softly textured beige plastic (Sintra) is secured over the base to cover the grid and provide a smooth reaching surface. The targets are based on plastic cones typically used in rehabilitation clinics [16] and were rapid prototyped using a fused deposition modeling (FDM) 3D printer.

D. Electronic Design A Wiring [17] microcontroller board controls the home

system’s inputs and outputs, monitors when the target has been touched, when the hand is in the correct rest position, and controls the lighted stimulus presentation. The Wiring board is an easy-to-use, low-cost, open source platform for

Fig. 1. Tabletop home system setup.

Fig. 2. Target location calculations for home system

169

Page 3: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

electronics prototyping based on the Atmel ATmega128 microprocessor. The microcontroller measures reaction time at one millisecond resolution and reaching- and return- times at tens of milliseconds resolution. It then transmits data to a Mac Mini [18] computer via a USB-serial connection. The Mac Mini was chosen for its small form factor, quiet operation and the ease of scripting using built in scripting languages. All electronics components are readily available off the shelf. High intensity red and green LEDs embedded in the top of each cone provide visual stimulus to elicit a reach.

Because stroke patients may successfully reach the target but be unable to grasp it, we want to ensure that all on-target reaches are detected. To detect and differentiate two types of

contact with the cone, we embedded a combination of capacitive touch sensors and force sensitive resistors in the cone. We chose the readily available, low-cost QTouch QT113 [19] touch sensor integrated circuit for its automatic calibration, drift compensation, and ease of use. To create the capacitive electrode we covered the cone targets with highly conductive silver fabric with a rubber coating so even a light graze is detected.

Each cone target is covered in sixteen 0.5” circular force-sensing resistors (FSR) to detect successful grasps [20] (See Fig. 4.). FSRs are reliable low-cost sensors for qualitatively measuring force. Half inch FSRs were chosen since larger FSRs have a drifting offset force due to deformation on a curved surface. Each FSR is part of a voltage divider circuit with a 2.2 kΩ resistor, which provides the largest range of voltages for forces normally seen in grasping. The FSRs are attached to the cone using the factory provided adhesive. The output from the force sensors are multiplexed into one analog input on the microcontroller for each cone.

A bright red foam pad marks the rest position, where the subject's wrist will be between target reaches. The foam pad has two conductive cloth contacts on its surface, which comprise a switch. The subject wears a wristband with a piece of conductive cloth sewn onto it, which closes the rest position switch when in contact (See Fig. 5.). This type of sensor ensures that the subject’s wrist is in the correct

position before every reach. In addition, a tact switch is embedded in the part of the foam pad to detect that the forearm is resting on the pad instead to help ensure that the patient’s arm is relaxed before a reach.

A magnetic reed switch paired with a magnetized wristband was considered [21], however many patients we work with have pacemakers and it was decided that it would be too dangerous to use magnets in an uncontrolled setting, such as the home, around such a sensitive device. A sound plays if the hand is not resting in the correct position and the target stimulus will not light until the hand returns to the correct position.

The audio feedback is provided using the Mac Mini computer and its built-in speakers. A Max/MSP [22] patch communicates with the microcontroller using the serial protocol and receives flags from the microcontroller to trigger sounds for audio feedback. The patch also logs and timestamps the reaction and reaching time data. The Max/MSP patch then communicates with custom motion capture software to begin recording from the Wii Remote IR cameras. IR camera data is logged using a custom program described in the next section. The system architecture is presented in Fig. 6.

E. Motion Capture

Fig. 3. Physical set up of table with overlay removed with rest position sensor.

Fig. 4. Force sensitive cone with cover removed to reveal FSRs.

Fig. 5. Wristband with marker and conductive cloth used for rest position contact switch

170

Page 4: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

End point trajectory measurements are important for assessing stroke patients’ reaching patterns, because they

can be used to calculate the smoothness, shape, and velocity of the trajectory. Usually these measurements are made using costly and cumbersome motion capture systems that are inappropriate for home use [23]. We have developed an inexpensive camera setup use remote controllers for the Nintendo Wii gaming system [24]. The remote control for the Wii gaming system (the Wii Remote) contains an embedded infrared camera with 1024x768 resolution and an approximately 40° x 30° field of view for only $39.99. Since the controller uses Bluetooth to connect with the Wii game console, it is easily interfaced with other Bluetooth capable devices. Using open source software, it is possible to connect to and record data from the Wii Remote. The Wii infrared (IR) camera is a proprietary system-on-a-chip from PixArt [25] that is able track 4 infrared sources at 100 Hz and returns their planar coordinates and intensity. Due to the camera’s high resolution, ease of interface with a computer, and fast sample rate, we selected the Wii Remote as non-intrusive, low-cost motion capture camera. For endpoint trajectory tracking, we created a wristband with a single marker that is tracked by two Wii Remotes. Using only a single marker allows us to avoid marker labeling issues. We were able to sufficiently reconstruct three-dimensional end-point trajectory using two Wii Remotes (For validation, see the System Validation section). Using the open-source DarwiinRemote code [26], we developed software that can

record from multiple Wii Remotes at a time. To avoid relying on AA batteries powering the Wii Remotes, we connected the power terminals to a 3V regulator.

For ease of setup by a patient, we chose passive reflective markers instead of active infrared emitting markers. However, since passive reflective markers require a source of infrared light to reflect, an array of TSAL6400 [27] infrared light emitting diodes (LEDs) was designed and mounted on the Wii Remote. The array dimensions (8 LEDs by 6 LEDs) maximize the number of LEDs that can be powered on two arrays from a single 12V, 1500mA power supply. The LEDs have an optimal combination of intensity (40 mW/sr) and angle of half intensity (25 degrees). The remote and array is presented in Fig. 7.

To convert the standard camera coordinates returned by the two Wii Remotes into global coordinates, a camera calibration was performed using a trapezoidal frame with four reflective markers. The intensity values returned by the cameras were not used due to their low 4-bit resolution. By capturing the location of four markers using a calibration frame at various locations in space, we can create a transformation to the global coordinate system based on Zhang’s method [28].

F. Device Costs The estimated cost for one system is $1146. A breakdown

of the costs is listed in Table I.

TABLE I APPROXIMATE COSTS FOR HOME SYSTEM

Component Cost

Mac Mini $599 3-D printing of cones $120 Two Wii Remotes $80 Wiring Microcontroller $65 Force Sensitive Resistors $65 Uninterruptible power supply $55 Misc. Electronics $40 Aluminum stock $30 Wood $30 Infrared LEDs $22 RGB LEDs $20 Sintra sheet $15 Conductive Cloth $5 Total $1146

G. Data collection The system will be installed in the homes of stroke

patients who have taken part in the Biofeedback for Rehabilitation therapy program at the Arts, Media, and Engineering program at ASU to determine the feasibility and usefulness of our home based system. The procedures for installing the devices in homes have been approved by the Institutional Review Board at ASU. The patients will play the assessment game three times a week for eight reaches to each cone at a specified time of day to monitor their reaching abilities for one month. To avoid fatigue, they can

Fig. 6. System architecture for the home system

Fig. 7. Wii Remote with attached IR array

171

Page 5: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

play for a maximum of 10 minutes per day. Chair position is maintained using adhesive markers on the floor for the chair feet.

The game begins when a subject presses a start button. A sound plays to indicate that their hand should return to the rest position. Once on the rest position, a red LED on a cone, chosen pseudo-randomly, lights up for a pseudo-randomized rest period between 3.5 to 5 seconds. This provides an adequate period of relaxation between reaches for the subject and allows the subject to get ready for a reach. The required duration on the rest position is variable and pseudorandom so that the subject performs discrete reaches and does not form a "rhythm" since there are different neural bases for rhythmic and discrete arm movements [28]. Varied rest times also reduce the subject’s tendency to anticipate the movement before the stimulus appears.

After this rest period, a green LED turns on to indicate that the subject may initiate a reach. Since sudden feedback at the initiation of a grasp could cause unnatural movement, no feedback is given for a successful grasp or touch other than the physical feedback from contacting the cone. If the subject is unable to reach the cone within 6 seconds, the light turns off to prevent frustration and proceeds to the next rest period and light after the subject returns to rest. If the subject anticipates the stimulus and comes off the rest position while the red light is on, the timer resets and the subject has to return to the rest position for another 3.5 to 5 seconds before the next light turns on. These timing values are based on the START assessment device [21] we are using as well as the Biofeedback for Rehabilitation training program [15].

End-point trajectory, relative grasp force, reaction time, and movement time are recorded and stored locally on the password-protected computer. A member of our research group visits the patient weekly to make sure the system is functioning and to collect data. If the patients have Internet connections in their homes, non-identifying system status reports are sent to a remote computer. However, without an encryption protocol or secure connection, data are not transmitted over the Internet. This is sufficient for assessment purposes but for training purposes an interactive remote component will become necessary.

III. SYSTEM VALIDATION

A. Patient acceptance The current system is based on two devices that have been

used and accepted by stroke patients and unimpaired subjects alike. The cones and reach time measuring components are based on devices that are used pre- and post-training in the Biofeedback for Rehabilitation system. Both have been used by 3 stroke patients and 4 unimpaired subjects. Since the current system is less complicated than the rehabilitation training that they have already participated in, and initial reactions from subjects to the idea of a home system has been positive, we expect acceptance of the home

system.

B. Motion capture The motion capture data from the Wii Remotes after 3-D

reconstruction was cross-validated with data recorded simultaneously from a 10 camera Eagle RealTime motion capture system and EvaRT software from Motion Analysis [30]. A reaching movement, similar to the system task, was performed using one reflective marker on the wrist. Both sets of data were first smoothed using a five-point moving average window in Matlab [31]. Displacement and velocity (difference in displacement divided by the sampling interval) were calculated for all data in the x, y and z directions for one representative reach (coordinate system is presented in Fig. 2). Absolute maximum errors were calculated (See Table II). Root mean squared error between the EVaRT data and Wii Remote data was calculated for each direction for displacement and velocity, then normalized by the maximum value of the EVaRT data (See Table III).

TABLE II

MAXIMUM ABSOLUTE ERRORS OF AN EXEMPLAR REACH

Axis Position Velocity

X 6.2 mm 28.5 mm/s Y 37.9 mm 65.3 mm/s Z 17.9 mm 48.2 mm/s

TABLE III

NORMALIZED ROOT MEAN SQUARE ERRORS OF AN EXEMPLAR REACH

Axis Position Velocity

X 1.9 % 2.9 % Y 9.8 % 9.7 % Z 11.6 % 9.7 %

Fig. 8-10 show the displacement and Fig. 11-13 show the

velocity from both systems in all three directions. While

there is an error in displacement in the y and z directions that increases as the marker moves further off the table and closer to the cameras, the velocities are almost identical in all directions. The displacement errors may be due to

Fig. 8. x-direction displacement curve for an exemplar reach (for Fig. 8-13, black dashed line is EVaRT reference and solid cyan line is from Wii remotes, for coordinate axes see Fig. 2.)

172

Page 6: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

inherent properties of the Wii Remote cameras or the calibration procedures. However, by staying in the accurate range of the system’s capture volume, we can avoid errors in displacement.

For our system, since the reach completion is detected by the touch sensor, the motion capture data can be used to assess hand transport velocity during the reach and return. Since the displacement curves from the Wiimote cameras are highly correlated to the EVaRT camera curves (R > .99 for all directions), we gain information about the end-point trajectory shape and smoothness.

C. Data storage To test the robustness of the data storage software and

avoid memory leaks, the system was left recording motion capture data 6 hours. Since the daily use is about 10 minutes, we felt the lack of errors during this period was adequate proof of software stability. Bluetooth connectivity performance was checked by leaving the Bluetooth connection to the Wii Remotes for 3 days without any problems. In the case of a loss of connectivity, an automated script is being developed to reconnect to the Wii Remotes.

IV. CONCLUSION The home system is a minimally intrusive, low cost, and

simple device for assessing upper limb rehabilitation in the home. Although there are tradeoffs in functionality and cost, it is clear that simple and low cost systems will be necessary to put results from expensive and complex stroke rehabilitation systems into practice. In this example, for approximately $1200, we are able to record parameters related to our lab setup including reaching trajectory, reach time, simple reaction time, and relative grasp force distributions. However, we are only able to record one marker and with some loss of spatial accuracy.

To the best of our knowledge, this is the first stroke rehabilitation system using the infrared cameras of Wii remotes to record end-point trajectory using 3-D reconstruction of reflective marker data. Cross-validation with reference system reveals that motion capture from Wii remotes is viable for recording end-point trajectory and for

generating feedback based on velocity. We expect ongoing validation with stroke survivors to

confirm the feasibility of such a system. Based on feedback from these feasibility tests and further development of the Biofeedback for Rehabilitation project, we will expand the system by using additional sensors and markers that can inform complex audio and visual knowledge-of-results and knowledge-of-performance feedback based on real time performance. This will allow the system to be used as a training system as well as an assessment device. This system can also be used to remotely monitor repetitive therapies assigned by a physical therapist.

Fig. 12. y-direction velocity for an exemplar reach

Fig. 13. z-direction velocity for an exemplar reach

Fig. 9. y-direction displacement for an exemplar reach

Fig. 10. z-direction displacement for an exemplar reach

Fig. 11. x-direction velocity for an exemplar reach

173

Page 7: [IEEE 2008 Virtual Rehabilitation - Vancouver, BC (2008.08.25-2008.08.27)] 2008 Virtual Rehabilitation - Low-cost, at-home assessment system with Wii Remote based motion capture

V. ACKNOWLEDGMENT The authors kindly thank Yu-fei Liu and Gang Qian for

help with camera calibration, David Lorig for support with the construction, Shawn Cook for help with software programming, Jeffrey Boyd and Michael Baran for help with motion capture testing, Leah Buechley for suggestions for conductive cloth and Johnny Cheung Lee for his demonstrations of alternate uses for Wii Remotes. We also thank two anonymous reviewers for their valuable comments.

REFERENCES [1] W. Rosamond, et al, "Heart disease and stroke statistics—2008

update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee," Circulation, vol. 117, pp. e25-146, Jan 29. 2008.

[2] J. Winters, "Motion analysis and telerehabilitation: healthcare delivery standards and strategies for the new millennium," Pediatric Gait, 2000.A New Millennium in Clinical Care and Motion Analysis Technology, pp. 16-22, 2000.

[3] M. J. Rosen, ‘‘Telerehabilitation,’’ NeuroRehabilitation, vol. 12, pp. 11---26, 1999

[4] G. C. Burdea, P. L. Tamar Weiss, D. Thalmann, "Guest Editorial Special Theme on Virtual Rehabilitation," Neural Systems and Rehabilitation Engineering, IEEE Transactions on, vol.15, no.1, pp.1-1, March 2007

[5] M. K. Holden, T. Dyar, E. Bizzi, L. Schwamm and L. Dayan-Cimadoro, "Telerehabilitation for motor retraining in patients with stroke,’’ Journal of Neurologic Physical Therapy: Volume, vol. 29, pp. 200, 2005.

[6] M. Rydmark, J. Boeren and R. Pasher, "Stroke rehabilitation at home using virtual reality, haptics and telemedicine". Studies in Health Technology and Informatics 85: 434-7, 2002

[7] W. Durfee, S. Weinstein, J. Carey, E. Bhatt and A. Nagpal, "Home Stroke Telerehabilitation System to Train Recovery of Hand Function," Rehabilitation Robotics, 2005.ICORR 2005.9th International Conference on, pp. 353-356, 1928.

[8] R. D. Willmann, G. Lanfermann, P. Saini, A. Timmermans, J. te Vrugt and S. Winter, "Home stroke rehabilitation for the upper limbs," Conf. Proc. IEEE Eng. Med. Biol. Soc., vol. 2007, pp. 4015-4018, 2007.

[9] R. Sanchez, D. Reinkensmeyer, P. Shah, J. Liu, S. Rao, R. Smith, S.

Cramer, T. Rahman and J. Bobrow, "Monitoring functional arm movement for home-based therapy after stroke," Conf. Proc. IEEE Eng. Med. Biol. Soc., vol. 7, pp. 4787-4790, 2004.

[10] L. Johnson and J. Winters, "Enhanced TheraJoy technology for use in upper-extremity stroke rehabilitation," Conf. Proc. IEEE Eng. Med. Biol. Soc., vol. 7, pp. 4932-4935, 2004.

[11] A.A Rizzo, D. Strickland, and S. Bouchard (2004). "The challenge of using virtual reality in telerehabilitation," Telemedicine Journal and e-Health., vol. 10, no.2, pp. 184-95.

[12] K. Morrow, C. Docan, G. Burdea and A. Merians, ‘‘Low-cost Virtual Rehabilitation of the Hand for Patients Post-Stroke’’, Virtual Rehabilitation, 2006 International Workshop on, vol., no., pp.6-10, 2006.

[13] P.H. Wilson et al., ‘‘A virtual tabletop workspace for the assessment of upper limb function in Traumatic Brain Injury (TBI),’’ Virtual Rehabilitation, 2007, 2007, pp. 14-19.

[14] D. P. Lorence and H. Park, "New technology and old habits: The role of age as a technology chasm," Technol. Health Care, vol. 14, pp. 91-96, 2006.

[15] H. Huang, Y. Chen, W. Xu, H. Sundaram, L. Olson, T. Ingalls, T. Rikakis and J. He, "Novel design of interactive multimodal biofeedback system for neurorehabilitation," Conf. Proc. IEEE Eng. Med. Biol. Soc., vol. 1, pp. 4925-4928, 2006.

[16] Sammons Preston. http://www.sammonspreston.com [17] Wiring. http://www.wiring.org.co [18] Apple Inc. http://www.apple.com [19] Quantum Research Group Ltd. http://www.qprox.com [20] Interlink Electronics, Inc. http://www.interlinkelectronics.com [21] M. Duff, S. Attygalle, T. Rikakis, J. He “A Portable, Low-cost

Assessment Device for Reaching Times,” IEEE EMBC 2008. August 20-24, 2008.

[22] Cycling ’74. http://www.cycling74.com [23] H. Zheng, N. D. Black and N. D. Harris, "Position-sensing

technologies for movement analysis in stroke rehabilitation," Med. Biol. Eng. Comput., vol. 43, pp. 413-420, Jul. 2005.

[24] Nintendo http://www.nintendo.com [25] PixArt Imaging Inc. http://www.pixart.com.tw [26] DarwiinRemote. http://sourceforge.net/projects/darwiin-remote [27] Vishay Intertechnology Inc. http://www.vishay.com [28] Z. Zhang. “A flexible new technique for camera calibration,” IEEE

transactions on Pattern Analysis and Machine Intelligence, 22(11):1330-1334, 2000.

[29] S. Schaal, D. Sternad, R. Osu and M. Kawato, "Rhythmic arm movement is not discrete," Nat. Neurosci., vol. 7, pp. 1136-1143, Oct. 2004.

[30] Motion Analysis Corp. http://www.motionanalysis.com [31] The Mathworks Inc. http://www.mathworks.com

174