2002 Conference Proceedings

Go to previous article 
Go to next article 
Return to 2002 Table of Contents


USING VIRTUAL REALITY TO IMPROVE WALKING FOLLOWING STROKE

David L. Jaffe, MS
VA Medical Center
3801 Miranda Ave., MS-153
Palo Alto, CA 94304
650/493-5000 ext 6-4480
jaffe@roses.stanford.edu

The ability to walk is an essential component for enabling people to live and work independently and to function safely in their homes and in the community. Individuals with locomotor disabilities have an increased fear and risk for falls and subsequent injuries due to their inability to respond quickly and appropriately to challenges within their environment such as stairs, inclines / slopes, and uneven surfaces. This is a particularly serious problem for persons with hemiplegia due to stroke whose locomotor disabilities include impaired balance, decreased stride length, compromised ability to step over objects, and decreased endurance [Said, 1999; Winter, 1990].

Stroke is the number one cause of adult disability in the US, costing more than $40 billion a year (NIH, July 99). It is estimated that 600,000 Americans suffer a first stroke each year and nearly 4 million people in the US who have survived a stroke are currently living with the consequences [American Heart Association, 2000].

Virtual reality systems have been used to enable people with disabilities to perceive and interact with objects in a non-physical environment that would be limited or impossible in the physical world. For example, with the virtual environment generated by the CAVE(tm) VR interface, a VR user interacts with both real and virtual objects to explore accessible transit design [Electronic Visualization Laboratory; Browning, 1995]. In another application, VR provided a way to manipulate and explore a virtual office without requiring physical access, dexterity, and strength [Nemire, 1994]. Telemanipulation and assessment applications of VR have been reported to be effective in analyzing hand and upper body function [Greenleaf, 1997]. A recent application of this technology employs a virtual kitchen to assess the cognitive functions of sequencing and meal preparation after brain injury [Christiansen, 1998; Christiansen, 1996; Zhang, 2001]. Other applied research utilizing VR includes operating a virtual wheelchair, controlling a mobile robot, and treating phobias [Smith, 1995; Inman, 1995; Simsarian, 1995; Weghorst, 1997].

The VA Merit Review project, Improving Stepping-Over Responses in the Elderly using Simulated Objects, compared two training interventions: stepping over real foam objects (Overground) and stepping over computer-generated objects (Treadmill). (In the following discussion, the acronym SOR [Stepping-Over Responses] will be used to refer to this project.) The SOR research protocol consisted of several elements: subject identification, cognitive screening, pre-training evaluation, training intervention, post-training evaluation, and follow-up evaluation.

Training Interventions - In the training interventions, subjects were instructed to step over objects of a selected height and length. A Physical Therapist provided feedback and encouragement during this training. Twelve trials of stepping over ten objects constituted one session. The intervention consisted of 6 sessions of approximately one-hour each, three times a week, over two weeks. Blood pressure was taken before and after each session. After each trial, the subject's heart rate, oxygen saturation, and perceived level of exertion was recorded. If a subject cleared 80% of the objects during a session, the next larger object in length was employed for the next session.

Overground Training Intervention - In the SOR overground training intervention, subjects stepped over foam objects in a hallway. The objects were placed on the floor, spaced at intervals ranging from 15 to 22 inches calculated from the subject's stride length.

Image of Overground Training Intervention

Treadmill Training Intervention - In the SOR treadmill training intervention, subjects walked on a treadmill at their normal walking speed and were held safely in place using an overhead harness. The harness did not suspend or support their body weight, but provided safely and prevented falls. Subjects held onto the treadmill's handrails for stability and orientation.

Image of harness used for Treadmill Training Intervention
Diagram of camera angle.
  1. foot with shoe
  2. computer-generated object
  3. rear sensitive area of computer-generated object
  4. front sensitive area of computer-generated object
  5. reflective tape on bootie
  6. footswitch between shoe and bootie

A color video camera was directed at the subject's legs from the right side. The subject wore a head-mounted display to view the camera's real time image. Despite the "immersive" nature of the images in the head-mounted display, no dizziness was reported, even with one subject who had reported episodes of mild claustrophobia at home.

The VR computer introduced a stationary image of a rectangular object of a selected height and length to the video of the subjects' feet. As the treadmill ran, the subject was instructed to step over the virtual object on each step. The computer detected any intersection of the user's feet with the objects. A collision by the toe on the front edge of the object indicated that the subject did not lift the foot high enough (Figure 4b), while a collision with the heel on the top of the object indicated that the subject did not step far enough (Figure 4c). Throughout the sessions, a therapist offered encouragement and advice to improve stepping performance.

Successful cleared obstacle. Toe collision with obstacle. Heel collision with obstacle.

Figure 4a - Successful

Figure 4b - Toe collision

Figure 4c - Heel collision

Foot switches worn between the shoes and "booties" tracked the stance and swing phases of each foot. When a collision with the virtual object occurred, a vibro-tactile feedback produced by pager motors was directed to the heel or toe of the colliding foot. An audio indication of the collision was also presented to the subject. The VR computer system monitored the subject's progress, keeping track of the steps and the number of collisions that occurred during the training sessions.

The visual, vibro-tactile, and auditory cues offered multiple ways of informing the subject of a collision, thereby provided the subject with meaningful multi-modal sensory feedback which, in turn, was utilized to train and correct their gait to avoid further collisions. Their ability to view the lateral side of their legs while walking gave the subjects a unique perspective of their efforts to negotiate the objects. The unique sensory feedback combined with the safety provided by the overhead harness afforded subjects a safe means to try different strategies recommended by the therapist to improve their performance.

This treadmill training technique produced an immediate and familiar visual stimulus for stepping and provided a unique perspective unavailable to a subject looking down at his/her feet or viewing their walking in a mirror. The treadmill and harness system offered a safe environment in which to respond to therapist's suggestions and try new strategies of moving and stepping. The therapist could view (via the computer's monitor) the same display as the subject to observe the subject's response to movement suggestions. The immediacy of the visual feedback when successfully negotiating with a virtual obstacle reinforced the subject's positive efforts. The computer added the ability to 1) fully document each session, 2) use objects of standardized size and placement, and 3) easily change the height and width of the object as the subject improved.

Results

The twenty-one SOR subjects in both training groups exhibited a wide range of pre-training capabilities and post-training performances in the major outcome measures.

Pre-training Post-training Significance
Walking speed (m/min) 32.7 (σ = 12.7) 41.8 (σ = 17.8) P = .001
Stride length (m) .873 (σ = .232) 1.065 (σ = .329) P < .001
6 minute walk (m) 726 (σ = 270) 780 (σ = 269) P < .05
Obstacle height (in) 9.3 (σ = 5.1) 11.4 (σ = 3.3) P < .05
Step length (m) 45.0 (σ = 11.8) 54.4 (σ = 14.2) P < .001
Chart graphs the percent improvement for the two training interventions for ten outcome measures.

The above chart graphs the percent improvement for the two training interventions for ten outcome measures. Improvement is defined as 100*(post-training-performance - pre-training-performance) / (pre-training-performance). Improvements in every test except one (overground step length of non-paretic leg during fast pace) occurred.

The group receiving the treadmill intervention training achieved significantly increased walking speed and stride length measures for both the self-selected and fast pace walking evaluation tests. The treadmill protocol results also show greater improvements in seven out of the ten performance measures. A minimal change in cadence (not graphed) was observed - subjects did not take more steps per trial, they took longer steps, which resulted in a greater walking speed.

Chart graphs the percent retention of the two training interventions for ten outcome measures.

The above chart graphs the percent retention of the two training interventions for ten outcome measures. It is defined as 100*(followup-training-performance) / (post-training-performance). A value greater than 100 indicates that the performance measured two weeks after the conclusion of training was greater than the performance measured just after the completion of training, i.e. additional improvements occurred after completion of the training. Averaged over all tests, both groups retained their post-training improvements two weeks after the end of the training.

Subjects in both training groups exhibited improvements in the outcome measures. Comparisons between the groups showed that the improvements of the group receiving treadmill training exceeded that of the overground group in 7 of 10 tests. However, the small number of subjects in the SOR study and their wide range of performance make it difficult to conclude that one training protocol is significantly better than the other.

Importantly, both training intervention protocols proved to be safe as no falls occurred.

Literature References

American Heart Association, Heart & Stroke Statistical Update, 2000.

Browning, Drew R, Virtual Reality and Accessible Transit Design: New Access Methods Project, Virtual Reality and Persons with Disabilities Conference, 1995. http://www.csun.edu/cod/95virt/0020.html

Christiansen, C, Abreu B, Ottenbacher K, Huffman K. Creating a virtual environment for brain injury rehabilitation and research: A preliminary report. Journal of Medicine and Virtual Reality, 6-9, 1996.

Christiansen, C, Abreu B, Ottenbacher K, Huffman K, Masel B, and Culpepper R. Task performance in virtual environments used for cognitive rehabilitation after traumatic brain injury. Archives of Physical Medicine and Rehabilitation. 79(8), 888-892, 1998.

Greenleaf, Walter J., Applications of Virtual Reality Technology to Therapy and Rehabilitation, New Technologies in Physical Therapy, 6:1, March, 1997.

Inman, Dean P. Teaching Motorized Wheelchair Operation in Virtual Reality, Virtual Reality and Persons with Disabilities Conference, 1995. http://www.csun.edu/cod/95virt/1001.html

Nemire K, Burke K, Jacoby R. Human factors engineering of a virtual laboratory for students with physical disabilities. Presence 3:216-226, 1994.

Reuben, D.B. and Siu, A.L., An objective measure of physical function of elderly outpatients, The physical performance test. J Am Geriatric Soc. 38: p. 1105-1112, 1990.

Said CM, Goldie PA, Patla AE, Sparrow WA, Obstacle crossing in subjects with stroke. Arch Phys Med Rehabil. Sep;80(9):1054-9, 1999.

Simsarian, Kristian T., Using Virtual and Augmented Reality to Control an Assistive Mobile Robot, Virtual Reality and Persons with Disabilities Conference, 1995. http://www.csun.edu/cod/95virt/0008.html

Smith, Stephen J., Virtual Presence and Autonomous Wheelchair Control: An Update, Virtual Reality and Persons with Disabilities Conference, 1995. http://www.csun.edu/cod/95virt/0017.html

Weghorst, Suzanne, Phobia Treatment by Virtual Reality Technology. ACM, 1997. http://www.kdinc.com/search7.htm

Winter DA, Patla AE, Frank JS, Walt SE. Biomechanical walking pattern changes in the fit and healthy elderly. Phys Ther. Jun;70(6):340-7, 1990.

Zhang L, Abreu BC, Masel B, Scheibel RS, Christiansen C, Huddleston N, and Ottenbacher KJ (in press). Virtual reality in the assessment of selected cognitive function after brain injury. American Journal of Physical Medicine and Rehabilitation. 80:597-604, 2001.


Go to previous article 
Go to next article 
Return to 2002 Table of Contents 
Return to Table of Proceedings


Reprinted with author(s) permission. Author(s) retain copyright.