Issue: Volume: 23 Issue: 10 (October 2000)

Balancing Act

By Karen Moltenbrey

Caren is a unique visualization system that uses high-speed motion capture, a motion platform, and 3D animation to immerse patients in a fully reactive virtual and physical environment. Medical experts use the system to determine, register, and evaluate human performance for therapeutic or diagnostic purposes.

The Caren system uses optical and magnetic sensors for capturing motion, determining on the fly which type of marker is producing the best results and processing that data only.

When we were toddlers, we had to learn to walk before we could run. And as videotapes and home movies show, learning to take those first steps was far from easy. We teetered in an unconfident and contorted stance that resulted in a shaky step forward or a sudden plop on the floor. But once we mastered the act, it soon became second nature, and we raced from here to there, leaving parents reminiscing about easier times. For many people, though, temporary injuries or crippling diseases can make the simple action of walking a difficult task. Now an innovative visualization environment from Motek Motion Technology (Amsterdam, The Netherlands; Manchester, NH) is enabling researchers and medical specialists to examine and analyze walking and other human movement in an entirely new way, and in some cases, help people with injuries, balance disorders, or other disabilities achieve better control of motor functions.
Caren is a unique visualization system that uses high-speed motion capture, a motion platform, and 3D animation to immerse patients in a fully reactive virtual and physical environment. Medical experts use the system to determine, register, and evaluate h

The new system, called Caren (Computer Assisted Rehabilitation Environment), immerses a patient in an interactive virtual and physical environment using a combination of motion-capture technology, 3D computer graphics, and a six-degree-of-freedom motion platform. A patient outfitted with optical or magnetic motion sensors stands on the motion platform in front of a large screen that projects a virtual 3D environment. As the person moves or walks, Motek's proprietary D-Flow program compares the patient's motion with an established database of normal body movement and adjusts the platform faster than the person can react to the change and the images they are seeing on the screen. This enables medical professionals to rehabilitate patients by reteaching them to walk, for instance, or to test and monitor movement disorders caused by coordination problems or joint deformities, for example, by accurately measuring subtle body motion.

"Caren started as an idea to use real-time motion capture for analysis, diagnosis, and the correction of balance disorders in people," says Oshri Even-Zohar, vice president/chief technology officer and a co-founder of Motek. "But to evaluate the subtleties of human balance behavior, you need a system that is several orders of magnitude faster than real time to discern the underlying deficiencies and deviations in their movement." The Caren system, he notes, processes the data at about 500hz, while a human's brain processes this motion data at about 5 to 7hz.
The Caren system uses optical and magnetic sensors for capturing motion, determining on the fly which type of marker is producing the best results and processing that data only. (Images courtesy of Motek Motion Technology.)

By placing a patient inside this faster than real-time feedback environment, physical therapists can introduce changes to the person's virtual reality (on the screen) and physical reality (on the motion platform) to alter the patient's motion memory or learned reaction to counteract an imbalance. By allowing the physical reality to occur slightly before the virtual reality, the patient is "tricked" into physically performing a movement earlier than anticipated.

For instance, a person with a broken ankle can be forced to inadvertently place a certain amount of pressure on the healing limb-rather than apply less pressure by limping-thereby speeding the rehabilitation process. "When you are engulfed in the Caren system, you don't have any motion memories that you can link to because they are being altered physically through the artificial environment-people cannot rely on their pre programmed motor reactions," ex plains Even-Zohar. As a result, a therapist can teach a patient new balance behaviors to compensate for a variety of impairments, including those related to a healing limb, the use of prosthetics, physical deformities, or debilitating tremors that may prevent the person from maintaining his or her balance.

For the most part, Caren is platform-independent, although Motek has chosen what it believes are the best components for the present Caren installations. These include an Ascension (Burlington, VT) mocap system, a Primas optical system from Delft Motion Analysis (Delft, The Netherlands), and animation software from Softimage (Montreal) for creating the virtual environment and the 3D body model on which the collected data is mapped. The motion base was designed specifically for this application by Rexroth Hydraudyne B.V. (Boxtel, The Netherlands) to accommodate the system's necessary response times.

The concept for Caren was born about 10 years ago in Amsterdam, The Netherlands. That's when Even-Zohar, a former animator who opened one of the first motion-capture studios for the film and game industry, discussed his idea with Edward Costello, who is credited with helping to create real-time motion capture while at motion-capture developer Polhemus (Colchester, VT).

Their vision seemed basic: Instead of creating virtual movements, the technology could be used for observing movements. "We were both busy with motion capture for the entertainment industry-capturing and linking a person's motion to a virtual character," says Costello, Motek's CEO and co-founder. "We realized we could use the same technology, along with simulation and training techniques, and apply them to the medical area."

The concept sat idle until about four years ago, when the European Commission, the governing body of the European Union, provided grant money for development of the Caren prototype. Even-Zohar put together a development team from Amsterdam that included medical researchers from the Amsterdam Academic Medical Center (AMC) and the University of Groningen's (RUG) physiology department, who provided invaluable medical diagnostic input and testing for the system. When the development project was completed, Even-Zohar's company, Lamalo BV, became the R&D, engineering, and manufacturing arm of the newly formed Motek company. "We had to wait for technology to catch up to our idea-computers, graphics processing speed, and motion-capture technology just weren't fast enough," Costello notes.
By projecting a person's real-time motion onto a 3D human, doctors can evaluate balance issues on the spot or save the data for later review.

During the past year, AMC and RUG have been conducting clinical trials of the system for use in physiotherapy, orthopedics, neurology, and the early diagnosis of a wide range of balance and movement disorders. The hospitals were using the setup at Motek's Amsterdam office, and are now awaiting delivery of their own systems.

One of Caren's greatest potential uses thus far lies in the aforementioned area of physical therapy, where health-care providers can invoke a controlled physical situation for a patient. For instance, a person undergoing rehabilitation therapy for a broken leg often walks more gingerly than necessary at first, delaying the healing process. With Caren, the therapist can create a situation where the platform is raised slightly higher than the patient expects so he or she will step down sooner and apply more pressure on the leg.

"A big part of this rehabilitation process is overcoming the mental barrier-the remembrance of pain-that makes a person hesitant to place all his or her weight on the injured leg," says Dr. Imelda J.M. de Groot, head of the rehabilitation department at AMC and an early user of Caren. "This hesitation can cause a person to limp unnecessarily."

In one particular case, de Groot used Caren to help rehabilitate one of the engineers working on the project who had recently broken his ankle, shortening the process from three months to just three weeks. Using motion-capture data of the person before the accident (taken while de Groot was building a database of "normal" human motion), she could use the engineer's own methods of walking to retrain him to walk and stand. "Because it was the most natural way for him to move, his body adjusted more quickly," she explains. "During therapy, his brain was okay, and at that point, his leg was okay, but it was the interaction between the two that had to be retrained. So before he could limp, the motion platform would rise to meet his foot, forcing him to walk the way he did before he was hurt."
Medical experts can use various CG environments to invoke different physical responses in a patient. For instance, if a person is placed in a 3D boat environment, a doctor can evaluate dynamic balance issues.

Because each person uses a different type of motion strategy to move, it is difficult for therapists to optimize treatment. Instead of basing treatment on a doctor's observations, the system collects the patient's exact motion using 25 to 33 strategically placed optical or magnetic sensors. The movements are then compared against Motek's database of "normal" human skeletal and muscle motions. This enables a therapist to view the stresses placed on certain muscles and then devise the appropriate treatment.

"With Caren, I've been able to uncover which element of the movement cycle is the one invoking the problem, leading to new therapy strategies," de Groot says.

A unique aspect of Caren, in de Groot's opinion, is that the results can be tested and retested (after therapeutic interventions) and extensively analyzed. "Virtual reality rehabilitation is not a new idea, but what makes this system so different is that it operates in real time, and the results can be duplicated." In the past, she has performed 2D and 3D analysis using a two-camera setup, but the results always lagged about three to four weeks behind. "So the input for the patient was often too late," she notes.

While de Groot's focus has been on using the technology for corrective measures, Dr. Egbert Otten, associate professor specializing in motor control at RUG, is applying it to conduct research into balance disorders such as Parkinson's disease and multiple sclerosis. But before such anomalies can be detected, doctors need to determine what is "normal." This is done using an ever-expanding database of human motion collected by de Groot and Motek technicians from hundreds of volunteers without balance problems. "The Caren project lets us measure human motion in a way that wasn't possible before," notes Otten. "Prior to Caren, there was no way to establish a normal standard; every doctor had his or her own opinion as to what was normal."

As a result, it has been difficult to detect diseases such as Parkinson's in the early stages, when they can be treated more effectively, or to determine the rate of motion deterioration. Otten's study in this area centers on a person's attempt to maintain balance. "There is no such thing as standing perfectly still," explains Motek's Even-Zohar. "When we are standing still, we are constantly evaluating our posture and shifting our center of gravity, and Caren can quantify this movement, no matter how slight."

According to Otten, a study he conducted of Parkinson's patients show that as they stand up and begin to walk, there is a slight pause between the two movements, whereas in other people, these processes tend to overlap. "With the naked eye, the pause is not observable, but with Caren, you can see it immediately," he notes.
Caren can separate a person's visual input (the projected 3D image) from his or her mechanical input, forcing the person to resolve the situation through body motion.

Once the motion data is fed into Caren's human body simulation model, the joint forces and muscle activation can be calculated, and from the patterns of these responses, doctors can make inferences concerning the person's motor processing. "If a person moves his arm, the most obvious movement is of the arm, but the most complex movement is the stabilization of all his other joints, which Caren shows," says Otten. "There is still a lot to be discovered in the area of human body motion, and it must be done in an integrative fashion. You need to look at the entire body."

According to Otten, the project's real-time feedback in virtually normal and responsive surroundings makes it possible to analyze and train disabled persons for daily-life activities. "Caren offers not only a valuable research tool for motor control, but also a testing and learning environment for patients," he says. With the system offering so much potential for a range of motion anomalies, Otten and his colleagues are careful to take one step at a time, so to speak. "As soon as you start using the system, you have to re think your old ideas about human body motion," he says. "There's so much to explore."

In the future, Caren could prove valuable in screening people who have a family history of degenerative diseases. "The hope is that by precisely measuring the differences in a person's responses over time, doctors will be able to predict a tendency toward a disease such as Parkinson's, so they can intervene early enough to treat it more effectively," says Motek's Costello. The system also holds potential in the sports arena, for optimizing a runner's spring to convert more of that energy to forward motion rather than upward motion, for instance. Engineers could also use the system for determining why it takes a pilot a certain amount of time to react to an emergency trigger. The list seems endless.

"We've only invented the tool," says Costello. "It's up to the experts to find applications for it."

Karen Moltenbrey is an associate editor of Computer Graphics World.