Computation in the Surgical Suite: Modeling Crouch Gait for Orthopedic Decision Making
Building surgeons' confidence
They say that what you don’t know can’t hurt you. But what your pediatric surgeon doesn’t know certainly can hurt your child. An orthopedist who recommends that a child with cerebral palsy undergo hamstring surgery to correct a gait problem, for example, often can’t be sure if the procedure will make things better, or worse.
Physicians and patients confront scenarios like this every day, with doctors making their best guesses about what might work, and those in their care trusting that everything will go according to plan. To boost their chances of being right, surgeons in various orthopedic specialties are turning to computational modeling and simulation to better predict patient outcomes and to improve accuracy and safety in the operating room. Some, for example, use modeling software to more accurately fit knee and hip implants to bone, while others go so far as to rely on computer-guided robots to do some of the cutting and sanding for them.
Here, we tell the tale of how one team of researchers and clinicians is using computers to help correct crouch gait, a problem that afflicts many patients with cerebral palsy.
Michael Schwartz, PhD, and Tom Novachek, MD, have been working hand in surgical glove for many years to integrate computational methods into the surgical decision-making process. Both men work at the James R. Gage Center for Gait and Motion Analysis at Gillette Children’s Specialty Healthcare in St. Paul, Minnesota—Novachek as the Center’s medical director, and Schwartz as its director of bioengineering research. (Both also hold appointments in the department of orthopedic surgery at the University of Minnesota.)
The Center uses the same motion-capture technology employed in Hollywood films to diagnose and plan treatments for people with walking and movement disorders. Approximately three quarters of the patients who enter its doors have cerebral palsy. Of those, some 90 percent are children, many of whom walk and run with their knees perpetually flexed, a painful and potentially debilitating condition known as crouch gait.
Because crouch gait is often attributed to excessively tight hamstring muscles, surgeons have typically tried to correct the disorder by lengthening them, a process that involves elongating the tendons that attach the hamstrings to the bones. But factors aside from hamstring length, such as neurological defects and bone deformities, may also contribute to crouch gait, and doctors have historically lacked good tools for predicting who will respond well to the procedure. As a result, while some young patients see improvement after surgery, others get worse; and even those who make headway may require further corrective surgery later on in life.
Since the mid-1990s, motion-capture hardware and 3-D simulation software have improved the situation somewhat by giving doctors a better understanding of how bone deformities and excessive tightness or looseness in other muscles can also lead to crouch gait. At Gillette, for example, every patient is outfitted with reflective markers and videotaped as he or she moves across the motion-capture lab. Commercial software uses the trajectories of the markers and the angles of the patient’s joints to generate an animated 3-D simulation that can be used to analyze the individual’s joint kinematics. Doctors can then use that information along with other patient-specific data to figure out what’s driving a particular problem. As sophisticated as it is, however, the system still has limitations; so Schwartz and Novachek have added another tool to their technological arsenal: OpenSim, the open-source software package for modeling and simulating movement developed through Simbios and the NIH Center for Simulation in Rehabilitation Research (NCSRR) at Stanford University.
In a series of articles, a group of researchers led by Scott Delp, PhD, professor of bioengineering, mechanical engineering, and orthopedic surgery at Stanford, used musculoskeletal models developed in OpenSim to demonstrate that the lengths and velocities of a patient’s hamstrings could be used to predict whether they would benefit from hamstring lengthening. (The velocity of a muscle describes the rate at which its length changes over time.) Schwartz, who contributed to some of those articles, explains that if the hamstrings are short enough and slow enough, then lengthening them can alleviate crouch gait. If not, lengthening may do nothing—or it may make matters worse.
Yet standard motion-capture gait analysis only offers an indirect measurement of muscle lengths and velocities; and basing surgical decisions on that kind of inference, says Novachek, involves “a little bit of a leap of faith.” Fortunately, OpenSim can use the data gathered during motion capture to compute muscle lengths and velocities directly. So now, marker trajectories and joint angles that are recorded for a child with crouch gait are fed into OpenSim, which spits out its best estimate of the lengths and velocities of their hamstrings. The software is even capable of minimizing the errors that can arise from differences between the observed marker positions (which can be thrown off by inaccurate marker placement or soft tissue motion) and what it predicts those marker positions should be based on the recorded joint angles.
Powerful as it is, OpenSim can’t yet tell Novachek and his fellow surgeons exactly how much a patient’s hamstrings ought to be lengthened. Nonetheless, simply identifying those patients who are genuinely good candidates for the procedure has already bumped up surgical success rates, and Novacheck has also seen reductions in under- and over-corrections, the latter of which is especially difficult to treat. More generally, Novacheck says that the confidence afforded by gait analysis has made the Gillette surgeons more aggressive about the number of gait-related corrections they’re willing to tackle during a single session—performing anywhere from five to fifteen at once, for example, rather than just one or two—thereby saving their patients from having to undergo multiple follow-up surgeries.
In the near term, Novacheck looks forward to seeing similar data on the lengths and velocities of other muscles implicated in crouch gait. In the long term, he hopes to one day bring motion-capture equipment right into the operating room in order to precisely document just how much hamstring lengthening is actually required to normalize a person’s gait—information that should enable OpenSim to predict how many millimeters or centimeters need to be trimmed from a given patient’s hamstrings before they go under the knife.