At the Palo Alto VA Rehabilitation Research and Development Center, we deal with problems of ambulation, manipulation, and also of applied technology. Our center has been known as a center of robotics for 20 years, from when Professors Leifer and Inder Perkash were working with robotic systems as adaptive devices. Our approach is to apply what we believe are sound engineering processes and tools in combination with medical science. Since we're in Silicon Valley, we take advantage of some technologies that allow us to conduct studies that previously were not feasible through the use of computer models and complex mathematical models.
With a valid model of movement, you can predict the motion from the knowledge of forces generated internally and the timing of the production of those forces. This allows you to determine whether the differences in observed data are truly responsible for the observed differences in behavior. It allows you to test hypotheses without doing invasive tests on subjects, often permitting you to predict what you will then see in the clinics and the laboratory.
The models we use are based on established anatomical, physiological, and engineering foundations. The process essentially is to take a user interface developed in Palo Alto, and link it to some particular process of motion we wish to investigate.
This approach has resulted not only in basic discoveries in biomechanics, but also in practical applications to address specific problems. We have developed, for instance, a package for simulating surgeries, such as tendon transfers. This software draws on anatomical, physiological, and engineering data to let you track outcomes of particular interventions: you can change your surgical technique and then look at the effects of your innovation. This has now been commercialized as SIM, produced by Muscular Graphics.
Apart from simulations, we do work with real people, primarily stroke patients and some incomplete spinal cord patients. We can measure their performance and create a model assuming activation levels and timing of the various muscles in the legs for pedaling. Pedaling is an easily modeled task, because of its constant relationship between the body and the environment that is essentially a four bar linkage that can consistently be modeled and the data applied to check the best fit. By doing this, we are able to predict how fast a person would pedal against a given work load, and, indeed, these models were accurate.
We then moved into the stroke population to measure muscle activation and timing. In stroke there are some muscles, the soleus for example, that behave about the same in control subjects as in stroke subjects. In other muscle, the activation is almost entirely out of phase. These observed patterns were then put into the model and successfully predicted the difficulty in phase transition observed in persons with stroke or hemiplegia on a monitored pedaling task.
Our initial apparatus has instrumented force transducers between the foot and the pedal and an ergometer to establish a workload. As it allows the person to pedal in either a supine or an upright position, we are able to test persons with a wide range of disabilities. This led to development of a clinical tool called the tilt cycle, which also features a backboard for body support that moves up and down, permitting force production with both limbs throughout the pedaling cycle.
With this apparatus, we can create novel pedaling tasks, like pedaling forward with one limb and then reverse with the other, to look at motor learning in patients with stroke. We are moving into the study of the influence of one limb on the performance of the opposite. With our newest device we can apply forces, let's say to the normal limb, so that it would sense if pedaling with hemiparatic limb on the other pedal. However, the two pedals are completely de-coupled: there is no direct mechanical coupling. Instead motors produce measurable position and force input to the two sides.
By making the hemiplegic limb feel as though it were pedaling normally, we can then look at the behavior of the opposite limb, or manipulate the forces that go into one limb to see what impact that has on performance of the opposite limb. All this comes from work we were doing on supported ambulation, and because of our proximity to NASA, we've taken advantage of a project intended to solve the problem of exercise in space.
In micro-gravity there is a problem holding a person against a treadmill. Rob Whalen at NASA came up with the idea of putting a partial vacuum in a sphere in which the person stands. The sphere has a waist seal and a pressure difference between outside and inside atmospheres exerts force. When the inside is negative, it pulls the person down against the treadmill. If the inside is positive, you simulate micro-gravity. A pressure difference of 1 pound per square inch within the bubble generates a 100-pound lifting force at the approximately 100 square inches at the waist.
As the waist is very close to the body's center of gravity, the lifting force is felt as if you are lighter. At 2 pounds per square inch within the bag, you can actually cause the person to be lifted off his or her feet. We applied no more than 1 pound and all subjects studies had significant reduction in unloading.
This led to a clinical version that is basically a vinyl bag with arm ports, so a therapist can reach inside. Subjects were able to walk much longer in this device then over ground and the therapist is able to reach through to guide the steps.
Our goal in upper limb rehab is to eventually help restore more function than currently possible after stroke. Those of you who work with stroke patients know that ambulation is more common than good upper limb function. We must first objectively quantify the upper limb movement during functional tasks by being able to make measures not subject to either observer or patient bias.
In this work we are using robotics to provide assistance or resistance to an upper limb during either programmed or subject-controlled trajectories. This allows us to make force measurements and look at training effects. Essentially we have a six degree of freedom digitizer that senses the position of one arm, while an industrial robot attached through a six axis force transducer to a trough on the other side produces motion in the hemiparetic limb.
Our current study is a randomized, controlled trial, comparing the typical tone reduction exercises, followed by rather simple motor skills, or motor tasks, progressing on to more functional tasks and to those assisted by a robot. We have a group of chronic stroke patients who are at least 6 months post acute and we compare their performances in a thrice weekly, 24-session protocol. The robot is able to provide the standard passive range of motion or an active range. In the former, the person is instructed to remain relaxed and to provide as little force between their arm and the robot as possible. For the active range, they help the robot to the point that the robot is applying no force to them and we can measure whether or not they're being successful with this.
Additionally, we have an active constraint mode, where the robot does not move unless the subject produces force in the correct trajectory, enough force to reach a target before they reverse direction. The subjects can do this for various programmed tasks. If they try to move out of the trajectory, the robot applies a spring-like restoring force. One last mode is the bilateral or mirror image mode, where whatever the person does with their unaffected limb is reflected about the midline, and assisted as the mirror image movement on the affected side. This allows the subject to control the therapy, the rate of movement, and the reach.
Basically, what we have shown is that there is a relationship between the upper limb component of the Fugelmeyer score (a standardized clinical measurement of functional motor strength and spasticity), and measures that we're getting from our torque and force transducers. We think that this is going to give us a more sensitive and more reliable measure of upper limb function.
Applications in MS Models of normal and impaired movement are useful in any clinical condition in which mobility is an issue. One of the problems that we have with MS is that it is a much more variable disease. Its course is not static or unnecessarily unidirectional. The objective assessments that we do in our work have a role in MS research because they allow you to look for perhaps more subtle changes, and they remove the subjective bias that can creep in from the therapists, other clinicians, and users. Some of the therapeutic methodologies, like the robot assistance and pressure-assisted ambulation may also have applications.
One possible advantage of the pressure system treadmill system to the MS population is that we can cool the air going into the chamber. This may ameliorate the fatigue factor which otherwise could actually compromise therapeutic progress in MS patients. Indeed, by improving energy efficiency or the mechanical efficiency of performing a task, and by unloading some of their body weight patients can better train the cardiovascular system and increase the ceiling of what maximal capacity could be. In that case, a given task represents a lower percentage of their maximal capacity; so by allowing them to train and increase their cardiovascular capacity, you are, in fact, addressing that fatigue issue.
Go to TOP.
|
|
|