26770
Kinect Motion Capture of Toddlers with Autism during ADOS Assessments

Poster Presentation
Friday, May 11, 2018: 10:00 AM-1:30 PM
Hall Grote Zaal (de Doelen ICC Rotterdam)
I. Budman1, I. Menashe2, G. Meiri3 and I. Dinstein4,5,6, (1)Department of Biomedical Engineering, Ben Gurion University, Beer Sheva, Israel, (2)Public Health Department, Ben-Gurion University, Beer Sheva, Israel, (3)Soroka Medical Center, Beer Sheba, Israel, (4)Negev Autism Center, Ben Gurion University of the Negev, Beer Sheba, Israel, (5)Department of Cognitive and Brain Sciences, Ben Gurion University of the Negev, Beer Sheba, Israel, (6)Department of Psychology, Ben Gurion University of the Negev, Beer Sheba, Israel
Background:
Motion tracking is a very promising technology for quantifying repetitive movements and motor problems in autism. In addition, motion tracking of multiple interacting individuals (e.g., clinician and child) could yield informative measures regarding social capabilities and preferences. Such measures may aid not only in the initial clinical evaluation of toddlers with autism, but also as longitudinal measures that might change with age and in response to interventions. Most motion capture systems require placement of markers/sensors on the body of participants, which preclude their use with toddlers who have severe symptoms and sensory hyper-sensitivities. The Microsoft Kinect system, however, enables marker-free motion tracking that would be possible with all children. This system is being implemented as part of the data collection for the regional autism database initiative at the Negev Autism Center in Israel (www.negevautism.org).

Objectives:
To develop an automated motion tracking tool that will quantify repetitive movements, motor problems, and measures of social interaction during a 45 minute ADOS assessment involving a clinician and toddler. In addition, to identify motion tracking measures that are correlated with autism severity and can aid with the clinical diagnosis of autism and longitudinal follow-ups.

Methods:
We have developed an automated marker-less motion-capture system consisting of 4 Kinect sensors arranged in a rectangular arrangement, in the corners of the ADOS assessment room. Recordings of the clinician and the toddler are performed at 30fps for a length of at least 45 minutes. The sensors are calibrated and synched using ipiSoft, which also fits the depth-data from the sensors to a human skeletal model. Initial analyses were carried out with skeletal data of the torso and head only, separately for the clinician and the child. This enabled assessment of distance between clinician and child as well as the deviation angle between the direction that the clinician was facing and that of the child.

Results:
To date, we have successfully analyzed data from 8 children, which have revealed excellent quality and reliability throughout the 45 minute recordings. We extracted the distance and deviation angle for each frame in the recording and examined the relationship between the mean of each measure and symptom severity as estimated by the ADOS. Analysis of additional recordings that are currently carried out at a rate of 6 new children per week will enable us to perform statistical tests regarding these relationships.

Conclusions:
Marker-less motion tracking using the Kinect system are likely to yield important informative measures regarding the severity of autism in individual children. Further development of this tool will also enable quantification of repetitive behaviors and motor problems in the same children.

See more of: Technology Demonstration
See more of: Technology Demonstration