Microsoft Kinect motion sensor An automated system that keeps watch over children could spot the telltale signs of autism and lead to earlier diagnoses. The noise level is rising at the Shirley G. Moore Laboratory School, at the University of Minnesota.   Children are charging through the classroom shouting, playing, picking up toys and tossing them around. All the while, in the corners of the room, five Kinect motion sensors watch and record their every move.
The unusual set-up at the University of Minnesota\'s Institute of Child Development in Minneapolis is designed to look for signs of behavioural disorders. The plan is to find out if Microsoft gaming sensor, combined with computer-vision algorithms trained to detect behavioural abnormalities, can be used to automate the early diagnosis of  autism.
Diagnosing an autism spectrum disorder (ASD) in young children is difficult. However the earlier a child can begin speech therapy and get help learning social and communication skills, the better it is for the child. Many different symptoms may suggest a child has an ASD, but they are subtle. It usually takes an experienced doctor to spot the signs by analysing video footage of the child playing, which is a costly and time consuming process.
Guillermo Sapiro, Nikolaos Papanikolopoulos and colleagues at the University of Minnesota have fitted the nursery with five Kinect depth-sensing camera rigs to find out if a computer can automate all or part of this process. These Kinect cameras will monitor groups of around 10 children aged between 3 and 5 years old as they play.
The cameras identify and track children based on their shape and the colour of the clothes they are wearing. The information is fed to three PCs, which run software that logs each child\'s activity level - including how they move each of their limbs - and plots it against the room\'s average. The system can flag up children who are hyperactive or unusually still - both possible markers for autism.
Medical staff can then decide whether the child requires closer attention from a specialist for a one-on-one diagnosis.
Sapiro added: \"the idea is not that we are going to replace the diagnosis, but we are going to bring diagnosis to everybody. The same way a good teacher flags a problem child, the system will do automatic flagging and say, \'Hey, this kid needs to see an expert\'.\"
Ultimately, the team hopes to merge the Kinect work with another project it is working on. By studying video footage of children interacting with a psychiatrist, computer-vision algorithms learn to identify behavioural markers as designated on the Autism Observation Scale for Infants. The system measures traits like a child\'s ability to follow an object as it passes in front of the eyes, as well as noting certain mannerisms or postures that are classified as being early signs of a possible ASD. Early tests have been in agreement with professional diagnosis, says Sapiro.
\"Early diagnosis is critical in helping people with autism get the support they need,\" says Caroline Hattersley of The National Autistic Society in London, who stresses that specialists are still needed. \"While this technology could potentially identify some signs of autism, there are many factors, such as language delay and limited eye contact that it would miss.\"
\"We are trying to do very difficult and expert analysis that a psychiatrist would do, but automatically,\" says Sapiro. He envisages a specially developed video game for Kinect that would test a child as they played with a parent and flag up any concerns.