An automated system of Kinect motion sensors that keep watch over children could spot telltale signs of autism.
Since the new set-up is automatic, it’s faster and cheaper than current detection methods for early diagnoses. New Scientist reports.
In young children, diagnosing an autism spectrum disorder (ASD) can be tricky. But it’s also better if the child begins speech therapy and gets help learning social and communication skills earlier on. Usually, detecting the subtle symptoms of an ASD requires video footage of a child playing and an experienced doctor to analyze it – but this is costly and time-consuming.
To find out if a computer can automate all or part of this process, researchers with the University of Minnesota's Institute of Child Development in Minneapolis have employed Microsoft’s gaming sensor, along with computer-vision algorithms trained to detect behavioral abnormalities.
- University of Minnesota's Guillermo Sapiro and colleagues fitted a nursery with 5 Kinect depth-sensing camera rigs to monitor groups of around 10 children ages 3 to 5 as they play.
- The cameras identify and track children based on their shape and the color of their clothes.
- The information is fed to 3 PCs that are running software that logs each child’s activity level (including how they move each of their limbs), and it plots this against the room’s average.
- The system flags the children who are hyperactive or unusually still, both possible markers for autism. (Medical staff can then decide whether the child requires a specialist for a one-on-one diagnosis.)
"The idea is not that we are going to replace the diagnosis, but we are going to bring diagnosis to everybody," Sapiro says. The same way a teacher might flag a child, “the system will do automatic flagging and say, 'Hey, this kid needs to see an expert'."
The team hopes to merge the Kinect work with another project they’re working on. By studying video footage of children interacting with a psychiatrist, computer-vision algorithms learn to identify behavioral markers as designated on the Autism Observation Scale for Infants [pdf]. The traits measured include: the ability to follow an object as it passes in front of the eyes and certain mannerisms or postures classified as being early signs of a possible ASD. Early tests have been in agreement with professional diagnosis, Sapiro says.
The system will be presented at the IEEE International Conference on Robotics and Automation in St Paul, Minnesota, this month.
[Via New Scientist]
Image by J. Fang
This post was originally published on Smartplanet.com