By: Anita Neal Harrison
Photos by: Shane Epping
It was the second day of basketball practice for the Mansfield (Mo.) Lady Lions, and senior Kailey Morris — the team’s point guard — was pumped.
“We’d just had a 6-foot-1 girl move in, and we were supposed to be pretty good,” says Morris, who averaged 19 points her junior year and was described by her coach as “a tremendous passer.”
But during a simple keep-away drill, as Morris crossed over to make a pass, her right leg gave out. It was the same leg that three years earlier had suffered an anterior cruciate ligament tear. A more recent MRI had revealed that her ACL was at high risk for tearing again — which it did, just a few weeks later.
“Even after it tore, my doctor told me I could still keep playing, if I could manage the pain,” Morris says. “I was still getting offers to play college basketball, and that had been my dream ever since I was little. I wanted the chance to do that.”
Female athletes in sports that require jumping or “cutting” — sudden changes in direction — are four-to-six times more likely than males to tear their ACLs. Although there is little female athletes can do to change their anatomy and hormones (both of which contribute to the problem) they can adopt training strategies and styles of play that lower the odds of injury.
Identifying the athletes most at risk is the starting point, according to Aaron Gray, an MU School of Medicine assistant professor in the departments of family and community medicine and orthopedic surgery, and Marjorie Skubic, an MU professor of electrical and computer engineering. In the past, screening has been impractical for all but the most elite athletes. Gray and Skubic are changing that with a cheap, widely available gaming peripheral called the Kinect, a motion-sensing device that works with Microsoft’s Xbox.
“We want to prove that our system screens as well as the expensive, lab-based system to identify girls who are high risk,” Gray says. “Then, in the future, we could potentially start putting those girls through some game-based interventions to help correct these neuromuscular and biomechanical weaknesses. And, hopefully, we’d be able to show on our screening tool that the way that the girls move changes and that their ACL numbers drop.”
ACL injuries, especially full ACL tears, have serious consequences. Often they require reconstructive surgery and six-to-nine months of rehab. Costs of around $30,000 per injury are not unusual. Damaged knees mean a lot of missed school for athletes and a lot of missed work for their families.
Morris sustained her first ACL tear as a freshman after a hard drive to the bucket during a game. It required surgery and six months of rehabilitation, which she unwisely cut down to five-and-half to compete in district tournament play. When her ACL tore again, she postponed surgery until after the basketball season ended. Recovery that time required eight months of rehab. She also spent three weeks in a wheelchair, forcing her to go on homebound instruction for the final quarter of her senior year in high school. “Not being around my friends for the last time that we were going to be together was rough,” Morris says.
Repeat injuries such as Morris’ are common. And even when ACLs are fixed, often they’re not: About 50 percent of ACL tears lead to arthritis developing in the knee within 10 to 15 years. The lasting effects make preventing these injuries even more important. Gray first thought seriously about prevention back in 2005, when he was completing a sports medicine fellowship at UCLA. He attended a medical conference that featured research using Vicon, a 3D motion-capture system. Using the device, scientists were able to show how girls’ joint positions during a simple jump correlated positively to future ACL injury risk.
It just so happened that after that conference, Gray bought an Xbox 360 game system with a Kinect, the aforementioned add-on, which let him control the game with just his movements — no other controller or body markers required. As he watched his on-screen self accurately mimic his real-life moves, Gray wondered, “Could Kinect replace the Vicon system in ACL screenings?”
If it could, it would mean the screenings could move out of the lab and into school gyms. While the Vicon system costs $150,000, the Kinect costs about $200. And there were other potential advantages. The lab-based Vicon uses multiple cameras and is not really portable. Kinect can fit in a book bag. And while the Vicon system requires placing reflective “body markers” on subjects, a time-consuming and error-prone task, the Kinect is marker-less. “I thought, ‘I need to try to find some engineers who would want to work on the problem,’” says Gray.
Seven years passed before he found the right collaborator. In 2012, Gray was new to MU. At a University function, a colleague pointed out Skubic, telling Gray that she was using Kinect in an elder-care project, one that outfitted seniors’ apartments with technology to monitor the residents’ health.
A Kinect expert, Skubic knew the device’s secret is depth imagery. “In a color image, each pixel has a red, green and blue value to it that gives you the color,” she explains. “A depth image doesn’t have color information; it shows each pixel’s depth in relationship to [the camera].”
To create that image, the first generation Kinect emitted a grid pattern of small infrared dots that “paint” the objects in front of it. These dots are invisible to the human eye but were recognized by Kinect’s infrared-sensitive camera. The Kinect then read the distortion in the grid pattern to determine the depth value of each pixel, and used that data to create 3D images. These could be processed into a moving game character, or, more to Gray and Skubic’s purposes, used to provide information on a subject’s positions and movements. The more recent version of Kinect has updated and improved this technology.
“Part of the reason I came to the University of Missouri is because I knew I wanted to do multidisciplinary research,” says Skubic, who has collaborated on projects ranging from equine surgery to robotics. “So when Aaron came to me and said, ‘Do you think we can do this?’ It wasn’t a big stretch to say, ‘Yeah, this looks like an important problem, and it looks like something we can do.’”
Although Skubic knew from her elder-care project that Kinect can accurately determine some biomechanical measures, such as walking speed and stride length, she wasn’t so sure that Kinect could accurately capture the types of hazardous motions that can lead to ACL problems.
Adds Gray: “One of the big things we see is that when female athletes get ready to jump, their knees kind of cave in and when they land, instead of landing softly, their legs are rigid, with their knees caved together, which, it’s thought, puts a lot more stress on the ACL.”
The drop vertical jump test — used in the Vicon research that led to Gray’s idea — requires subjects to stand atop a one-foot-tall box, drop off the box, and, upon landing, immediately jump as high as they can. There are two specific points during this jump when hip, knee and ankle positions need to be measured. First there is the point of initial contact, when subjects’ feet first touch the ground, and, next, the point of peak flexion — peak bending — which occurs after the initial contact and before subjects leave the ground for the vertical jump.
To find and record the joint positions at these points, Skubic’s team used the Microsoft Kinect SDK, a software-development kit that pulls out skeletal data and delivers X, Y, Z locations for the body’s major joints. The team also used the Vicon motion capture system to see how well the two systems agreed. Their findings, presented at the 2013 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, were “that the Kinect skeletal model likely offers acceptable accuracy for use as part of a screening tool for elevated ACL injury risk.”
“Since then, we have run another validation study with the latest version of the Kinect and with more subjects,” Skubic says. “The results are quite good for the knee-to-ankle separation ratio.”
Having validated the Kinect’s accuracy, the researchers say the next step is to test the biomechanical measures’ predictive power. This research, now underway, requires using Kinect to record athletes’ joint positions, then following up at six months and one year to count ACL injuries. So far the researchers have screened 89 female athletes (and 72 male athletes) between the ages of 14 and 18 at five high schools.
The setup at the screenings is incredibly simple. The only equipment needed is a laptop, a Kinect, a tripod to hold the Kinect and the box for athletes to drop from. As soon as an athlete jumps, data about their joint positions automatically pops up on the screen, thanks to algorithms Skubic’s team developed to process Kinect’s inputs.
So far, the knee-to-ankle separation ratio data from the Kinect screenings is showing that girls are much more likely than boys to have knees caved toward each other at the point of peak flexion, a finding consistent with previous studies. The researchers are working with MU’s Trent Guess, an associate professor in the departments of physical therapy and orthopaedic surgery and the director of the Mizzou Motion Analysis Center, to identify other biomechanical measures that might predict future ACL-injury risk.
“With the improvement of technology in the Kinect version 2, we hope to be able to measure other things, like how bent or straight the knees are when someone first lands,” Gray says. “We also plan to look to see if one foot touches the ground before the other, along with other things that would be difficult to see with the naked eye.”
As Gray and Skubic wait to see how the captured joint positions relate to ACL injury risk, they are already exploring new interventions, with a focus on video and phone games. In one prototype the MU team has developed, proper leg lifts result in a tiger character jumping on and over columns.
“If you look at the literature, there’s an issue with compliance,” Skubic says. This, she says, is why they are exploring new interventions. “I have a daughter who played sports in high school, and if I would’ve told her, ‘You need to do this really boring, repetitive exercise on a regular basis because you might have an ACL injury,’ I can tell you what her reaction would have been! And the research backs this up. … So people have been trying to figure out, ‘How can we encourage them to do these exercises?’ And Aaron came up with this idea of game-based exercises.”
The researchers have been surveying athletes participating in the screenings to find out which games and gaming platforms appeal to them. Results so far show girls prefer using their phones, while boys favor systems. “We want to identify how we can incorporate these exercise-based games into what they’re already doing and make them engaging so they’re fun to do,” Skubic says. “That’s the strategy behind it.”
Although a game-based intervention is still years down the road, Gray adds that he hopes the screenings themselves improve motivation.
“It’s the science of compliance,” he says. “There are programs out there to prevent ACL tears, but they target everybody on the team. What we hope is that we can tell girls: ‘Not only are all girls at higher risk, but your risk is higher still. You have a 20 percent chance this year to tear your ACL.’ We believe that will be powerful.”
It would have been for her, says Morris, who, despite her injuries, signed with William Woods University women’s basketball and is spending her freshman year redshirting; that is, preserving her four years of eligibility by attending classes and practicing with the team but not competing in games.
“My doctor told me I was about as ‘high risk’ as he’s seen because my legs are so loose-jointed,” she says. “I couldn’t have changed that, but I could’ve strengthened other things to compensate. That would’ve been nice to know.”