Trajectory Gaze Path Analysis and Isolating Areas of Interest in Eye-Tracking Data for Autism Spectrum Disorder Studies
Author:
Kartikeya Sharma ’22Co-Authors:
Dr. Vanessa Troiani, Dr. Antoinette DicriscioFaculty Mentor(s):
Dr. Brian R. King, Computer ScienceFunding Source:
PURAbstract
According to the CDC, about 1 in 59 children have autism spectrum disorder or ASD, representing a significant percentage of the population. Unfortunately, this condition often remains undiagnosed until later in childhood, which, in turn, delays many clinical treatments that could improve social functioning outcomes.
Researchers have identified abnormal visual attention as a hallmark symptom of ASD. With this finding, ASD researchers commonly deploy eye-tracking systems in their experiments. A typical experimental setup assesses how participants look at objects encapsulated within one static image or a stimulus. Eye-tracking systems collect real-time gaze data over a short, fixed time period. ASD experts have found that children without ASD generally focus more on objects associated with socializing, such as people or food items, than on inanimate objects. In contrast, children with ASD tend to focus on both categories of images with no preference. Heat maps, currently used in the clinical setting, forgo clinically crucial information about how children cognitively prioritize stimuli over time.
To better understand the cognitive process for prioritization of stimuli between children with and without ASD, clinical researchers need novel methods that yield visuals that show how participants prioritize stimuli over time. My work under the guidance of Dr. Brian King is developing multiple novel algorithms intersecting between the computer science fields of data mining and machine learning, including density-based clustering, object detection, and image classification. Further, we make these data visualization algorithms accessible to end-users through an interactive graphical user interface (GUI) encapsulated within a software toolkit.